{"site":{"name":"Koji","description":"AI-native customer research platform that helps teams conduct, analyze, and synthesize customer interviews at scale.","url":"https://www.koji.so","contentTypes":["blog","documentation"],"lastUpdated":"2026-10-31T05:59:05.919Z"},"content":[{"type":"documentation","id":"d37a2942-fda7-4d9e-a238-fc8c3f3151c1","slug":"first-click-testing-guide","title":"First-Click Testing: The Complete Guide to Validating Navigation and Findability (2026)","url":"https://www.koji.so/docs/first-click-testing-guide","summary":"A complete practitioner guide to first-click testing — the UX research method that uses a single click to predict task success. Covers methodology, the Bailey & Wolfson research foundation, sample size guidance (15-100 participants), step-by-step setup, and how AI-moderated platforms like Koji extend the method with conditional follow-up questions.","content":"## First-click testing in 30 seconds\n\nFirst-click testing is a usability research method that asks participants to complete a task on a static screen or prototype by clicking the *single first place* they would go to accomplish it. The click location, time-to-click, and a short follow-up question are recorded. Why it matters: research by Bob Bailey and Cari Wolfson found that participants who clicked the correct first link succeeded on the overall task **87% of the time**, versus only **46%** when the first click was wrong — a near-2x success gap that holds across thousands of follow-up studies. ([MeasuringU](https://measuringu.com/first-choice/))\n\nUse first-click testing when you need to validate **information architecture, navigation labels, button placement, or findability** before investing in full usability sessions. Modern AI-native platforms like Koji extend the method by pairing click data with a brief AI-moderated interview — capturing not just *where* users clicked but *why* they expected the answer to live there.\n\n---\n\n## What is first-click testing?\n\nFirst-click testing — sometimes called \"click testing\" or \"findability testing\" — is an evaluative usability method that isolates a single moment in the user journey: the very first decision a participant makes when starting a task. Participants see a screenshot, wireframe, or live page, read a task scenario (\"You want to cancel your subscription. Where do you click first?\"), and click once. The test records:\n\n- **Click coordinates** — visualized as a heatmap\n- **Time to first click** — hesitation indicates uncertainty\n- **Correctness** — did they hit the intended target area?\n- **Optional follow-up** — why did they choose that location?\n\nUnlike full usability tests, first-click tests do not measure task completion across multiple steps. They measure the leading indicator that *predicts* completion.\n\n## Why the first click matters: the science\n\nBob Bailey and Cari Wolfson published \"FirstClick Usability Testing: A New Methodology for Predicting Users' Success on Tasks\" in 2009, analyzing 12 scenario-based user tests. Their conclusion reshaped how UX teams validate navigation:\n\n> \"If users get the first click right, they have an 87 percent chance of completing the task correctly. If they get it wrong, that drops to 46 percent.\"\n> — Bob Bailey, *FirstClick Usability Testing* (Bailey & Wolfson, 2009)\n\nA larger validation study analyzing eight new studies with **1,000+ users across 137 tasks** found the success gap was even more dramatic: 80% success when the first path was correct, only 14% when it was wrong — meaning users were **more than 6x as likely to succeed** when their first click landed in the right area. ([MeasuringU](https://measuringu.com/first-choice/))\n\nWhy does the first click predict so much? Cognitive psychology offers an answer. Once a user commits to a path — even a wrong one — they tend to keep going down it before backtracking. This is the well-documented \"tunnel vision\" or path-dependence effect. As Optimal Workshop summarized in their analysis of TreeJack data: **correct first clicks lead to 3x higher task success rates** in tree-testing studies. ([Optimal Workshop](https://www.optimalworkshop.com/blog/correct-first-click-lead-to-3x-higher-task-success))\n\n## When to use first-click testing\n\nFirst-click testing shines in specific scenarios. Use it when:\n\n- **You're evaluating information architecture or navigation labels.** \"Will users find Pricing under 'Plans' or under 'For Business'?\" — first-click testing answers this in hours, not weeks.\n- **You have low-fidelity wireframes or mockups.** No working prototype required. A flat screenshot is enough.\n- **You need quick directional data before a full usability test.** First-click tests cost a fraction of moderated sessions.\n- **You're comparing design alternatives (A/B).** Run the same task on two variants and compare correct-click rates.\n- **You're auditing a live site for findability.** Identify the highest-friction tasks before redesign.\n\nFirst-click testing is **not** the right method for evaluating multi-step workflows, emotional reactions, complex form interactions, or content comprehension. For those, pair the click test with an AI-moderated interview or a [usability testing study](/docs/usability-testing-guide).\n\n## How to run a first-click test: step by step\n\n### 1. Define the task scenarios\n\nWrite tasks the way real users describe them, not the way internal teams label them. Avoid using the exact word that appears on the button you want them to find — that turns the test into word-matching, not findability.\n\n- **Bad:** \"Click on 'Pricing'.\" (just word matching)\n- **Good:** \"You want to know how much the Pro plan costs. Where would you click first?\"\n\nAim for 5–8 scenarios per session. Beyond that, attention drops and click data becomes noisy.\n\n### 2. Choose your stimulus\n\nA static screenshot of the page or wireframe is the most common stimulus. For more advanced tests, you can use clickable prototypes from Figma, Sketch, or your live product. Ensure the image renders cleanly across screen sizes — distorted layouts skew results.\n\n### 3. Set the correct-click region\n\nDraw a hit area around each target element (the link or button you intended users to click). Most platforms let you define this as a polygon. Be generous — include adjacent label text and surrounding padding.\n\n### 4. Recruit and screen participants\n\nFollow a [purposive sampling approach](/docs/purposive-sampling-guide) — recruit participants who match your target user profile. A first-click test on enterprise B2B navigation needs participants familiar with B2B SaaS, not the general public. In Koji, open Panel recruit, describe that profile, and Koji finds them across 130+ countries at a price per completed interview in credits you approve first (launching is on paid plans); your screener questions run before the test starts. If the users are already yours, import them so each gets a personalized link and you can see who took part. If you prefer to run recruitment yourself, marketplaces like User Interviews or Respondent can source a specialist audience: share your Koji link with the people they find and the results land in the same study.\n\n### 5. Decide on sample size\n\nSample size depends on your confidence requirements:\n\n| Confidence Level | Recommended Sample | Source |\n|---|---|---|\n| Quick directional insight | 15–30 per task | [Lyssna](https://www.lyssna.com/guides/first-click-testing/) |\n| Standard testing | 20–30 per task | UserTesting |\n| High statistical confidence | 50–100 per task | [Optimal Workshop](https://support.optimalworkshop.com/en/articles/9679633-how-many-participants-you-need-for-reliable-results) |\n\nFor most product teams iterating on IA, **30 participants per task** is the sweet spot — enough to spot clear patterns without burning budget.\n\n### 6. Add a follow-up question\n\nA click alone tells you *where* users went. To learn *why*, add an open-ended question after each task: \"What made you click there?\" This is where first-click testing benefits enormously from AI moderation — instead of a one-shot text box, an AI interviewer can probe: \"You said the icon looked like settings — what specifically made it look that way?\"\n\n### 7. Analyze results\n\nLook at four signals:\n\n1. **Heatmap distribution** — concentrated clicks (good) vs. scattered clicks (bad)\n2. **Correct click rate** — percentage of participants who clicked inside the target area\n3. **Time to first click** — long times signal hesitation and unclear labels\n4. **Open-text reasoning** — themes in *why* people clicked where they did\n\nA correct-click rate above 80% indicates strong findability. Below 60% means the navigation pattern is broken and warrants redesign.\n\n## How Koji modernizes first-click testing\n\nWhile dedicated click-testing tools like Optimal Workshop, Lyssna, and Maze handle the click-recording mechanic, they all share the same limitation: the open-text follow-up is a dead-end. Participants type a sentence, and you're stuck with shallow rationale.\n\nKoji takes a different approach. Instead of a static \"why?\" textbox, Koji's [AI moderator](/docs/understanding-the-ai-consultant) probes intelligently after every click:\n\n- **Conditional follow-ups** — when a participant clicks the wrong area, the AI asks \"What did you expect to happen when you clicked there?\" If they click the correct area, the AI explores confidence: \"On a scale of 1–5, how sure were you?\"\n- **Voice or text** — participants can speak their reasoning while their click is fresh, dramatically improving recall accuracy ([AI voice interviews](/docs/ai-voice-interviews-definitive-guide))\n- **Structured + qualitative blend** — pair scale questions (\"How easy was it to know where to click?\" 1–5) with open-ended questions (\"What confused you about the layout?\") in a single study using [structured questions](/docs/structured-questions-guide)\n- **Automatic thematic analysis** — Koji groups all \"wrong click\" rationales into themes automatically, surfacing patterns like \"users mistook the help icon for settings\"\n\nA recent industry survey by UserZoom found that teams using AI-assisted research tools report **60% faster time-to-insight** compared to traditional unmoderated platforms. Koji's combination of click-data-style structured questions + automatic moderation makes first-click testing a 1-hour task instead of a 1-week project.\n\n## Common mistakes to avoid\n\n1. **Word-matching tasks.** If your task says \"Find the Pricing page,\" any user who recognizes \"Pricing\" in the nav clicks it — you're not testing findability, you're testing reading.\n2. **Too many tasks per session.** Beyond 8 tasks, attention fatigue degrades data quality. Split into multiple shorter studies.\n3. **Testing only success cases.** Include \"trick\" tasks where the desired action *isn't* obviously available — these reveal whether users invent paths or give up.\n4. **Ignoring time-to-click.** A correct click that took 45 seconds is functionally a failure. Cap acceptable response times at 15–20 seconds.\n5. **Skipping the why.** Click coordinates without reasoning are diagnostic but not prescriptive. You know *what* failed but not *what to fix*.\n6. **Recruiting the wrong audience.** A first-click test on senior-care navigation should not be run on a 25-year-old developer panel. Match participants to your real user base via [screener questions](/docs/screener-questions-guide); when Koji recruits, those screening questions run before anyone reaches the first task.\n\n## First-click testing vs. tree testing vs. usability testing\n\n| Method | What it measures | Stimulus | Best for |\n|---|---|---|---|\n| First-click testing | First navigation choice on a visual | Screenshot/prototype | Validating button placement, label clarity, IA |\n| [Tree testing](/docs/tree-testing-guide) | Path through a text-only IA | Bare hierarchy | Validating IA structure independent of design |\n| [Usability testing](/docs/usability-testing-guide) | End-to-end task completion | Live product | Evaluating full flows and emotional response |\n\nThe three methods complement each other. Run tree tests early to validate IA, first-click tests when applying that IA to visual design, and full usability tests on the integrated product.\n\n## Real-world example: SaaS pricing page\n\nA B2B SaaS team noticed their \"Compare plans\" CTA had a 12% click-through rate — far below industry average. They suspected the button was visually weak. A 30-participant first-click test using the prompt *\"You're evaluating which subscription tier fits your team. Where would you click first?\"* revealed:\n\n- 41% clicked on the navigation \"Pricing\" link (correct alternative path)\n- 34% clicked on individual plan tiles (acceptable, but not the comparison view)\n- Only 18% clicked the \"Compare plans\" CTA\n- 7% clicked elsewhere\n\nFollow-up AI interviews surfaced the why: participants didn't recognize \"Compare plans\" as a comparison tool; they expected a tooltip or hover state, not a button. The team renamed the CTA to \"See full feature comparison\" and increased its visual weight. Re-test: 47% correct first clicks.\n\nTotal time from research design to insight: 5 days. Cost: under $400 in incentives.\n\n## Related Resources\n\n- [Structured Questions Guide](/docs/structured-questions-guide) — Combine click data with scale, single-choice, and open-ended questions\n- [Usability Testing Guide](/docs/usability-testing-guide) — When you need full task-flow evaluation\n- [Tree Testing Guide](/docs/tree-testing-guide) — Validate information architecture independent of visual design\n- [Card Sorting Guide](/docs/card-sorting-guide) — Build the IA before you test it\n- [Purposive Sampling Guide](/docs/purposive-sampling-guide) — Recruit the right participants for findability studies\n- [How to Conduct Remote User Interviews](/blog/how-to-conduct-remote-user-interviews-2026) — Add depth to click data with conversation\n\n\n## Further reading on the blog\n\n- [Concept Testing: The Complete Guide for Product Teams (2026)](/blog/concept-testing-guide-2026) — Concept testing validates whether your idea is worth building before you build it. This guide covers methods, question templates, analysis a\n- [Koji vs Maze: Which Research Tool Is Right for Your Team? (2026)](/blog/koji-vs-maze-2026) — Koji and Maze both claim to power product research — but they do very different things. Here’s an honest 2026 comparison to help you choose \n- [Koji vs Userlytics: AI-Moderated Customer Interviews vs Usability Testing (2026)](/blog/koji-vs-userlytics-2026) — Koji and Userlytics both help you understand users — but they answer completely different research questions. Here's how to choose the right\n\n<!-- further-reading:blog -->\n","category":"Research Methods","lastModified":"2026-10-31T05:50:55.657107+00:00","metaTitle":"First-Click Testing: Complete Guide to Findability & Navigation (2026) | Koji","metaDescription":"Master first-click testing with proven methodology, sample size guidance, and the Bailey & Wolfson research that showed correct first clicks predict 87% task success. Modern guide for UX teams.","keywords":["first-click testing","findability testing","navigation testing","click test","first click test","usability testing","information architecture testing"],"aiSummary":"A complete practitioner guide to first-click testing — the UX research method that uses a single click to predict task success. Covers methodology, the Bailey & Wolfson research foundation, sample size guidance (15-100 participants), step-by-step setup, and how AI-moderated platforms like Koji extend the method with conditional follow-up questions.","aiPrerequisites":["Basic familiarity with usability testing","A wireframe or live page to test","Defined target user audience"],"aiLearningOutcomes":["Understand when to use first-click testing vs tree testing vs full usability testing","Design unbiased task scenarios that test findability not word-matching","Choose appropriate sample size based on confidence requirements","Analyze click distribution heatmaps and correct-click rates","Pair click data with AI-moderated follow-up questions for deeper insight"],"aiDifficulty":"intermediate","aiEstimatedTime":"12 min"},{"type":"documentation","id":"b21bb9cb-f187-45f3-90d2-139baf1b1167","slug":"qualitative-research-sampling-methods","title":"Sampling Methods in Qualitative Research: A Complete Guide for Choosing the Right Approach (2026)","url":"https://www.koji.so/docs/qualitative-research-sampling-methods","summary":"A complete reference guide to the eight sampling methods used in qualitative research: purposive, theoretical, snowball, convenience, quota, criterion, maximum variation, and homogeneous. Includes the empirical research on saturation (Guest et al. 2006, Hennink et al. 2017), a decision framework for choosing methods, sample size guidance, and how AI-moderated platforms like Koji enable larger and more sophisticated sampling strategies.","content":"## Qualitative sampling in 30 seconds\n\nQualitative research uses **non-probability sampling** — instead of randomly drawing from a population to make statistical generalizations, you deliberately select participants who can produce the richest insight on your research question. The eight most common qualitative sampling methods are **purposive, theoretical, snowball, convenience, quota, criterion, maximum-variation, and homogeneous** sampling. The right choice depends on what you're trying to learn — exploratory studies favor maximum-variation and snowball; theory-building studies favor theoretical sampling; tightly-scoped product research favors purposive and criterion.\n\nA landmark 2006 study by Guest, Bunce, and Johnson found that **basic themes emerge in the first six interviews and saturation typically occurs by the twelfth** — but only when sampling is *strategic*. Bad sampling can leave you stuck at twelve interviews with no usable theme structure. ([Sage Journals](https://journals.sagepub.com/doi/10.1177/1525822X05279903))\n\nModern AI-native platforms like Koji change the economics of qualitative sampling. Where a traditional researcher might cap a study at 12–15 interviews because moderating and analyzing more is too expensive, AI moderation makes 50 or 100 interviews feasible — letting you run methods like maximum variation that previously required massive budgets.\n\n---\n\n## Why qualitative sampling is different\n\nQuantitative sampling aims for **statistical representativeness** — random samples large enough to generalize a finding to a population. Qualitative sampling aims for **conceptual representativeness** — purposeful selection of participants whose experience can illuminate the phenomenon you're studying.\n\nAs Patton (2002) put it: *\"The logic and power of purposeful sampling lies in selecting information-rich cases for study in depth. Information-rich cases are those from which one can learn a great deal about issues of central importance.\"*\n\nThis means three things in practice:\n\n1. **Sample size is determined by saturation, not by power calculation.** You stop adding participants when new interviews stop producing new themes.\n2. **Strategy matters more than randomness.** A poorly chosen 30-person sample is worse than a well-chosen 8-person sample.\n3. **Combining methods is normal.** Real studies often start with convenience sampling, transition to purposive, then end with theoretical sampling as a theory takes shape.\n\n## The eight qualitative sampling methods\n\n### 1. Purposive sampling\n\n**Definition:** The researcher deliberately selects participants who possess specific characteristics relevant to the research question.\n\n**When to use:** Most product, UX, and customer research. When you know the profile of who has insight on the question — power users, churned customers, recent buyers, decision-makers — and you can recruit against that profile.\n\n**Strengths:** High signal density. Every interview is on-target.\n\n**Weaknesses:** Requires you to know the right profile in advance. Misses unexpected user types that randomization would catch.\n\n**Example:** Recruiting 12 users who churned within 30 days of signup to understand cancellation drivers.\n\nSee the dedicated [Purposive Sampling Guide](/docs/purposive-sampling-guide) for criteria design and recruitment workflows.\n\n### 2. Theoretical sampling\n\n**Definition:** Sampling driven by an emerging theory. As patterns appear in early interviews, the researcher targets new participants whose experiences will test, refine, or extend those patterns.\n\n**When to use:** Grounded theory studies, deep customer-discovery work, and any research where you start with a hypothesis-free stance and build the model as you go.\n\n**Strengths:** Maximizes theoretical coverage. Avoids confirmation bias by deliberately seeking disconfirming cases.\n\n**Weaknesses:** Requires concurrent analysis — you can't batch all interviews up front. Hard to plan timelines or budgets in advance.\n\n**Example:** After 5 interviews suggest a theme of \"switching costs,\" recruit 3 more participants who recently *did* switch successfully to test what made it possible.\n\n### 3. Snowball sampling\n\n**Definition:** The researcher recruits a few initial participants, then asks each of them to refer others who fit the criteria. The sample grows like a snowball.\n\n**When to use:** Hard-to-reach populations — niche professional roles, sensitive topics, communities the researcher isn't embedded in. Also useful when participants share insider knowledge of who else has the relevant experience. ([NCBI](https://pmc.ncbi.nlm.nih.gov/articles/PMC5774281/))\n\n**Strengths:** Reaches populations standard recruitment platforms can't. Builds trust through participant referrals.\n\n**Weaknesses:** Sample is socially clustered — referrals tend to share characteristics, biasing the data. Not suitable for studies that need demographic spread.\n\n**Example:** Studying the experience of CTOs at pre-Series-A startups by starting with 3 known contacts and asking each for two referrals.\n\n### 4. Convenience sampling\n\n**Definition:** Recruiting whoever is easy to reach — your customer list, your social network, intercept on a website.\n\n**When to use:** Pilot studies, quick directional reads, exploratory research where speed matters more than rigor. Often used as the *starting* sample in a study that later transitions to purposive.\n\n**Strengths:** Fast and cheap.\n\n**Weaknesses:** Low representativeness. Findings cannot be generalized beyond the sampled group.\n\n**Example:** Posting a recruitment link in your customer Slack community to gauge initial reactions to a feature concept.\n\nThis is the most-criticized method and the most-used. The pragmatic stance: convenience sampling is acceptable when the research question is exploratory and the team treats results as hypotheses to be tested, not conclusions.\n\n### 5. Quota sampling\n\n**Definition:** A non-probability variant of stratified sampling. The researcher defines target quotas (e.g., 10 enterprise users, 10 mid-market, 10 SMB) and recruits until each cell is filled.\n\n**When to use:** Studies where you need representation across known segments. Common in B2B research, multi-region studies, and anywhere you suspect different segments have meaningfully different experiences.\n\n**Strengths:** Forces segment diversity. Allows segment-level comparison in analysis.\n\n**Weaknesses:** Within each cell, recruitment is convenience-based — so the segment itself may be skewed.\n\n**Example:** A pricing-research study with 8 participants from each of three plan tiers (Free, Pro, Enterprise) to compare willingness-to-pay drivers.\n\n### 6. Criterion sampling\n\n**Definition:** All participants must meet a predefined set of criteria. A specific subtype of purposive sampling.\n\n**When to use:** Quality assurance studies, churn analysis, edge-case investigation. Anywhere you need participants who *all* share a defining experience.\n\n**Strengths:** Highly targeted. Strong internal validity within the chosen criterion.\n\n**Weaknesses:** Findings apply only to the defined criterion group.\n\n**Example:** Recruiting 15 customers who completed onboarding *but* did not return within 7 days, to understand the drop-off mechanism.\n\n### 7. Maximum variation sampling\n\n**Definition:** Deliberately selecting participants who span the widest possible range on key dimensions — geography, role, tenure, use case, demographic — to capture both common patterns and edge variation.\n\n**When to use:** When you suspect the phenomenon varies significantly across user contexts and you want both the *core* themes that survive variation and the *edge* themes specific to subgroups.\n\n**Strengths:** Highest descriptive completeness. Themes that emerge across maximum variation are robust.\n\n**Weaknesses:** Larger sample needed. Harder to recruit deliberately.\n\n**Example:** A user-research study on remote work that recruits 4 participants each from solo founders, agency consultants, enterprise engineers, and creative freelancers — explicitly capturing the spread.\n\n### 8. Homogeneous sampling\n\n**Definition:** The opposite of maximum variation — selecting participants who are tightly similar on key dimensions to study a specific shared experience in depth.\n\n**When to use:** Focus-group-style studies, deep dives on a specific persona, studies where context is so determinative that mixing populations would muddy the analysis.\n\n**Strengths:** Rich depth on a tightly-scoped phenomenon.\n\n**Weaknesses:** Limited generalization. Easily mistaken for representativeness.\n\n**Example:** Eight female founders of bootstrapped SaaS businesses with $1M-$5M ARR, interviewed about hiring decisions.\n\n## How to choose the right sampling method\n\nUse this decision framework:\n\n| If your research question is... | Use this method |\n|---|---|\n| \"What do customers experience when X?\" (exploratory) | Maximum variation or purposive |\n| \"Why did churned users leave?\" (criterion-defined) | Criterion sampling |\n| \"How do different segments differ?\" (comparative) | Quota sampling |\n| \"Build a theory of how X works\" (grounded theory) | Theoretical sampling |\n| \"Reach a niche population\" (hard-to-find) | Snowball sampling |\n| \"Quick directional read\" (exploratory pilot) | Convenience sampling |\n| \"Deep dive on one persona\" (focused) | Homogeneous sampling |\n\nMost real studies combine two or more methods. A common pattern: **convenience-then-purposive** (start with easy-to-reach pilots, then targeted recruitment) or **purposive-with-quota** (define a profile, then quota across a key segment dimension).\n\n## Sample size in qualitative research\n\nThe field has converged around several rules of thumb based on empirical research:\n\n- **Guest, Bunce, & Johnson (2006)** — analyzing 60 in-depth interviews — found basic themes emerged within the first 6 interviews and saturation occurred within 12. ([Sage Journals](https://journals.sagepub.com/doi/10.1177/1525822X05279903))\n- **Hennink et al. (2017)** — comparing code saturation vs. meaning saturation — found code saturation at 9 interviews but meaning saturation at 16-24. ([NCBI](https://pmc.ncbi.nlm.nih.gov/articles/PMC9359070/))\n- **Nielsen Norman Group** — for usability research — recommends 5 users per persona, with multiple personas as appropriate.\n\nA pragmatic playbook for product research:\n\n- **Pilot / discovery study:** 5–8 interviews\n- **Focused thematic study:** 10–15 interviews per segment\n- **Comparative study (multiple segments):** 8–10 interviews per segment cell\n- **Generalizable thematic study:** 20–30 interviews\n- **Grounded theory:** 25–40 interviews with iterative theoretical sampling\n\nThe deeper truth: sample size is determined by saturation, not by a target. Stop when new interviews stop yielding new codes. See [Data Saturation in Qualitative Research](/docs/data-saturation-qualitative-research) for how to operationalize saturation tracking.\n\n## How Koji changes the economics of sampling\n\nClassical qualitative research methods were built around a constraint: every interview required a human moderator for an hour, and analysis took days per study. That cost structure forced researchers to cap samples at 12–15, choose the cheapest sampling method (often convenience), and rarely use methods like maximum variation or theoretical sampling that need larger samples to work properly.\n\nAI-native platforms like Koji break that constraint. With [AI-moderated interviews](/docs/ai-moderated-interviews), the marginal cost of interview number 50 is the same as interview number 5. This unlocks sampling strategies that were previously the domain of well-funded research orgs:\n\n- **Maximum variation at scale.** Recruit 40 participants spanning your full user spread. Koji moderates all 40 in parallel; [thematic analysis](/docs/thematic-analysis-guide) groups themes automatically across the variation.\n- **Quota sampling without bottlenecks.** Run 10 participants in each of five segments simultaneously without adding moderator hours.\n- **Theoretical sampling in real time.** As Koji surfaces emerging themes in [the report](/docs/generating-research-reports), researchers can launch follow-up cohorts that target specific theoretical gaps within hours, not weeks.\n- **Always-on continuous research.** Set up a [continuous discovery program](/blog/continuous-discovery-handbook-weekly-customer-interviews) that maintains a rolling sample of 15–20 conversations per week — a sampling cadence impossible with manual research.\n\n[Structured questions](/docs/structured-questions-guide) within Koji also make sampling-aware analysis trivial. Filter scale-question results by segment, compare ranking-question outcomes across maximum-variation cohorts, or quota a yes/no question across plan tiers — all without re-coding transcripts.\n\nIndustry data: teams using AI-assisted research report **60% faster time-to-insight** and run an average of **3x more interviews per study** compared to traditional methods, dramatically expanding the sampling strategies available to them.\n\n## Common qualitative sampling mistakes\n\n1. **Confusing convenience sampling with rigor.** \"We interviewed 20 customers\" sounds robust until you notice all 20 came from the same Slack channel. Document recruitment source explicitly.\n2. **Stopping at a number, not at saturation.** Eight interviews where the eighth surfaced three new themes is *not* a saturated study. Keep going.\n3. **Ignoring within-cell homogeneity.** Quota sampling fixes between-segment representation but doesn't fix within-segment selection bias.\n4. **Using snowball sampling without disclosure.** Snowball samples are socially clustered. Always note this limitation in research reports.\n5. **Treating purposive sampling as bias.** It's not bias — it's deliberate strategy. The bias is failing to *justify* the purposive criteria.\n6. **Skipping screener questions.** Even purposive sampling fails if you don't verify each participant matches the criteria. See [Screener Questions Guide](/docs/screener-questions-guide).\n7. **Mixing recruitment methods without documenting.** A study that started purposive and quietly switched to snowball mid-recruitment will produce contaminated themes. Document every method change.\n\n## Quick reference: sampling method comparison\n\n| Method | Sample size guidance | Effort to recruit | Best signal density |\n|---|---|---|---|\n| Purposive | 8–15 per segment | Medium | Very high |\n| Theoretical | 15–40 iterative | High | Highest |\n| Snowball | 8–20 | Low (after seeds) | Medium |\n| Convenience | 5–15 | Low | Low |\n| Quota | 8–10 per cell | Medium-high | High |\n| Criterion | 10–20 | Medium | High |\n| Maximum variation | 20–40 | High | High (across spread) |\n| Homogeneous | 6–12 | Medium | Very high (narrow) |\n\n## Related Resources\n\n- [Structured Questions Guide](/docs/structured-questions-guide) — Build segment-aware analysis directly into your interview questions\n- [Purposive Sampling Guide](/docs/purposive-sampling-guide) — Deep dive on the most common qualitative sampling method\n- [How Many Interviews Are Enough](/docs/how-many-interviews-enough) — The empirical evidence on saturation\n- [Data Saturation in Qualitative Research](/docs/data-saturation-qualitative-research) — Track when sampling can stop\n- [Screener Questions Guide](/docs/screener-questions-guide) — Verify every participant matches sampling criteria\n- [Finding Research Participants](/docs/finding-research-participants) — Practical recruitment workflows\n\n\n## Further reading on the blog\n\n- [B2B Customer Research: The Complete Guide for Product Teams (2026)](/blog/b2b-customer-research-guide-2026) — B2B customer research is harder than B2C — you are navigating buying groups of 10+ stakeholders, gatekeepers, and enterprise procurement cyc\n- [B2C User Research: How to Understand Consumer Behavior at Scale (2026)](/blog/b2c-user-research-guide-2026) — B2C user research is systematically underinvested at most consumer companies. While B2B teams run structured customer discovery as a matter \n- [Best Qualitative Research Tools in 2026: The Complete Buyer's Guide](/blog/best-qualitative-research-tools-2026) — Comparing the top qualitative research platforms in 2026 — from AI-native interview tools to research repositories. Find the right tool for \n\n<!-- further-reading:blog -->\n","category":"Research Methods","lastModified":"2026-10-31T05:50:55.657107+00:00","metaTitle":"Qualitative Research Sampling Methods: 8 Approaches Explained (2026) | Koji","metaDescription":"Compare the eight qualitative sampling methods — purposive, theoretical, snowball, convenience, quota, criterion, maximum variation, and homogeneous. With sample size guidance and decision framework.","keywords":["qualitative research sampling","qualitative sampling methods","sampling techniques qualitative research","purposive sampling","theoretical sampling","snowball sampling","maximum variation sampling","qualitative sample size"],"aiSummary":"A complete reference guide to the eight sampling methods used in qualitative research: purposive, theoretical, snowball, convenience, quota, criterion, maximum variation, and homogeneous. Includes the empirical research on saturation (Guest et al. 2006, Hennink et al. 2017), a decision framework for choosing methods, sample size guidance, and how AI-moderated platforms like Koji enable larger and more sophisticated sampling strategies.","aiPrerequisites":["Familiarity with qualitative research basics","A defined research question","Understanding of target user populations"],"aiLearningOutcomes":["Distinguish the 8 main qualitative sampling methods and when each applies","Apply a decision framework to pick the right sampling strategy","Determine appropriate sample sizes based on saturation research","Recognize and avoid common sampling pitfalls","Combine sampling methods in real research workflows"],"aiDifficulty":"intermediate","aiEstimatedTime":"14 min"},{"type":"documentation","id":"69bc06e1-0f69-42c7-b4e7-3c675ebaf18f","slug":"jobs-to-be-done-interviews","title":"Jobs-to-Be-Done Interviews: The Timeline Interview and Job Statements","url":"https://www.koji.so/docs/jobs-to-be-done-interviews","summary":"A JTBD interview reconstructs one real decision on a timeline, from first thought to first use, and the patterns across 8 to 12 interviews become job statements in the form When [situation], I want to [motivation], so I can [outcome]. This page covers running the timeline interview, exploring the four forces, analysing transcripts and writing job statements. The framework itself is covered in the JTBD framework guide, and the switch interview question bank in the switch interviews guide.","content":"# Jobs-to-Be-Done Interviews: Running the Timeline Interview and Writing Job Statements\n\nA Jobs-to-Be-Done (JTBD) interview reconstructs one real decision, from the first moment someone thought \"this isn't working any more\" to the day they started using something new. You then turn the patterns across interviews into job statements: \"When [situation], I want to [motivation], so I can [outcome].\" This page covers exactly that: how to run the timeline interview and how to get from transcripts to job statements.\n\nFor the framework itself (its history, the two schools and well-known examples), read the [Jobs to Be Done framework guide](/docs/jobs-to-be-done-framework). For a full switch interview question bank and a ready template, see [switch interviews](/docs/switch-interviews-jtbd-method).\n\nIf you're a founder with a handful of customers and no researcher, start this week: pick 8 to 12 people who signed up or left recently, run the timeline below with each one, and write job statements from what repeats. Koji's Jobs to Be Done Interviews template gives you the questions to start from.\n\n## What Is Jobs-to-Be-Done?\n\nThe JTBD framework originated from the work of Clayton Christensen at Harvard Business School, building on Theodore Levitt's famous insight that customers don't want a quarter-inch drill — they want a quarter-inch hole. Tony Ulwick further developed the framework into Outcome-Driven Innovation, providing a structured methodology for identifying underserved customer needs.\n\nThe central premise: **people don't buy products — they hire them to get a job done.** A \"job\" is the progress someone is trying to make in a particular circumstance. Understanding the job gives you a much more durable foundation for product strategy than demographic profiles or feature wishlists.\n\nAccording to Christensen, over 95% of new products fail because companies focus on improving products along dimensions customers don't care about, rather than understanding the job the customer is trying to do.\n\n## JTBD vs. Traditional User Interviews\n\n| Dimension | Traditional User Interview | JTBD Interview |\n|---|---|---|\n| **Focus** | Current experience with a product or process | The decision to adopt or switch |\n| **Timeframe** | Present and recent past | The full decision timeline (first thought to adoption) |\n| **Key question** | \"What's your experience with X?\" | \"Walk me through how you came to use X.\" |\n| **Output** | Pain points, needs, workflows | Causal forces driving and resisting change |\n| **Best for** | Improving existing experiences | Product strategy, positioning, new product development |\n\nBoth approaches are valuable, and they complement each other. Traditional interviews (covered in our [user interview guide](/docs/user-interview-guide)) are better for understanding ongoing experience. JTBD interviews excel at understanding the decision to adopt.\n\n## The Four Forces of Progress\n\nJTBD theory identifies four forces that act on every switching decision:\n\n1. **Push of the current situation** — What's going wrong? What frustration is building? This is the dissatisfaction that creates openness to change.\n\n2. **Pull of the new solution** — What's attracting them toward the alternative? What promise does it hold?\n\n3. **Anxiety of the new solution** — What fears or uncertainties hold them back? \"What if it doesn't work? What if I lose my data? What if it's too hard to learn?\"\n\n4. **Habit of the present** — What's comfortable about the status quo? \"It's not great, but I know how to use it. All my stuff is there.\"\n\nFor someone to switch, the combined strength of Push + Pull must outweigh Anxiety + Habit. Your JTBD interviews should uncover all four forces.\n\n## The Timeline Interview Technique\n\nThe timeline interview is the signature JTBD method. It reconstructs the entire decision journey — from the first moment something triggered the thought of change, through evaluation and deliberation, to the actual purchase or adoption.\n\n### Structure of a Timeline Interview\n\n**1. Start at the purchase moment and work backward.**\n\n\"You mentioned you started using [product] about [timeframe] ago. Can you take me back to that moment — what was happening when you signed up?\"\n\nThen keep asking: \"And before that?\" \"What led to that?\" \"When did you first start thinking about this?\"\n\n**2. Map the key moments:**\n\n- **First thought:** When did the idea of changing first enter their mind? What triggered it?\n- **Passive looking:** When did they start noticing alternatives without actively searching?\n- **Active looking:** When did they start deliberately evaluating options?\n- **Decision:** What tipped them over the edge to actually commit?\n- **First use:** What was the onboarding experience like? Did it deliver on the promise?\n\n**3. For each moment, explore:**\n\n- What were you doing? Where were you?\n- Who else was involved in the decision?\n- What were you feeling?\n- What alternatives did you consider?\n- What almost stopped you?\n\n### Sample Timeline Interview Questions\n\n- \"Take me back to when you first started thinking about finding a new way to handle [job]. What was going on?\"\n- \"How long did you know about [product] before you actually tried it?\"\n- \"Was there a specific moment when you went from thinking about it to actually doing something about it?\"\n- \"What other options did you look at? Why did those not work out?\"\n- \"Was there a point where you almost didn't switch? What almost held you back?\"\n- \"When you first started using [product], did it do what you expected?\"\n\n## The Switch Interview Technique\n\nThe switch interview is a more focused variant that zeroes in on customers who recently switched from a competitor or alternative solution. Research from the Rewired Group suggests that interviewing customers within 90 days of their switching decision produces the most reliable recall of the forces at play.\n\n### Who to Interview\n\n- Recent new customers who switched from a known competitor\n- Recent new customers who switched from a DIY or manual process\n- Recent churned customers who switched away from your product\n\nAll three groups are in your CRM: import them into Koji so each person gets a personalised link, the interviewer greets them by name, and the report is already split by the alternative they came from. Churned customers often speak more freely on the generic anonymous link. For switchers in a market you have no list for, such as buyers who moved between two competitors, open Panel recruit and Koji finds them across 130+ countries at a price you approve first.\n\n### Sample Switch Interview Questions\n\n**Exploring the push:**\n- \"What was the thing that finally made you say 'enough' with your old approach?\"\n- \"How long had that been bothering you before you decided to do something about it?\"\n\n**Exploring the pull:**\n- \"When you first heard about [new solution], what specifically caught your attention?\"\n- \"What did you hope would change by switching?\"\n\n**Exploring anxiety:**\n- \"What concerns did you have about switching? What was the biggest risk?\"\n- \"Was there anything that almost stopped you from making the change?\"\n\n**Exploring habit:**\n- \"What was good enough about your old approach that you stuck with it as long as you did?\"\n- \"What's the thing you miss most, if anything, about how you used to do it?\"\n\n## Analyzing JTBD Interviews\n\nAfter conducting 8–12 JTBD interviews, you'll analyze the data to identify:\n\n1. **The core job statement.** \"When I [situation], I want to [motivation], so I can [expected outcome].\"\n2. **The trigger events** that push people out of the status quo.\n3. **The hiring criteria** — what people actually evaluate when choosing a solution.\n4. **The anxiety barriers** that prevent or delay adoption.\n5. **Competing solutions** — including \"do nothing,\" which is often the biggest competitor.\n\nYou can apply [thematic analysis](/docs/thematic-analysis-guide) to your JTBD interview data, coding for the four forces across all interviews and looking for patterns in triggers, criteria, and barriers.\n\nKoji's report groups open answers into themes with the quotes behind each one, so you start coding from clusters of similar triggers and criteria rather than from raw transcripts. Checking each theme against the four forces is still your job.\n\n## Writing Job Statements\n\nA well-formed job statement follows this format:\n\n**\"When [situation/trigger], I want to [progress/motivation], so I can [desired outcome].\"**\n\nExamples:\n- \"When I'm onboarding a new team member, I want to give them context on past decisions, so they can contribute meaningfully within their first week.\"\n- \"When my quarterly report is due, I want to pull together data from multiple tools, so I can show stakeholders the impact of our work.\"\n\nGood job statements are solution-agnostic — they describe the progress people want to make without specifying how.\n\n## Common JTBD Interview Mistakes\n\n- **Starting with the product instead of the situation.** Don't ask \"Why did you choose us?\" Start with the circumstance that triggered the search.\n- **Accepting vague answers.** \"It was better\" is not enough. Probe for specifics: \"Better in what way? Can you give me an example?\"\n- **Ignoring the emotional dimension.** JTBD decisions are driven by both functional and emotional progress. \"How did that make you feel?\" is a critical probe.\n- **Only interviewing happy customers.** Churned customers and people who almost-but-didn't switch offer equally valuable data about the forces at play.\n\n## Related Resources\n\n- [Jobs to Be Done framework](/docs/jobs-to-be-done-framework): the full framework, its two schools and examples\n- [Switch interviews](/docs/switch-interviews-jtbd-method): the switch interview question bank and template\n- [Thematic analysis guide](/docs/thematic-analysis-guide): code transcripts for the four forces\n- [Structured questions guide](/docs/structured-questions-guide): add scale or choice questions next to the timeline\n- [User interview guide](/docs/user-interview-guide): foundational interview skills\n- [Writing interview questions](/docs/writing-interview-questions): craft better questions for any interview type\n\n## Further reading on the blog\n\n- [The Complete Guide to Jobs-to-Be-Done Interviews (JTBD Framework 2026)](/blog/jobs-to-be-done-interview-guide-2026) — Jobs-to-be-done interviews reveal why customers really switch to — or away from — your product. This complete guide covers the JTBD framewor\n- [The Product Manager's Guide to Customer Discovery with AI (2026)](/blog/product-manager-guide-customer-discovery-ai) — Most PMs know they should talk to customers — but scheduling, note-taking, and analysis eat the clock. Here’s how to run faster, deeper cust\n\n<!-- further-reading:blog -->\n","category":"Research Methods","lastModified":"2026-10-31T05:50:43.353793+00:00","metaTitle":"JTBD Timeline Interviews and Job Statements | Koji Docs","metaDescription":"Run a Jobs-to-Be-Done timeline interview and turn what you hear into job statements. Sample questions, analysis steps and common mistakes.","keywords":["jobs to be done","JTBD","switch interview","timeline interview","product strategy","Clayton Christensen","Tony Ulwick","customer switching"],"aiSummary":"A JTBD interview reconstructs one real decision on a timeline, from first thought to first use, and the patterns across 8 to 12 interviews become job statements in the form When [situation], I want to [motivation], so I can [outcome]. This page covers running the timeline interview, exploring the four forces, analysing transcripts and writing job statements. The framework itself is covered in the JTBD framework guide, and the switch interview question bank in the switch interviews guide.","aiPrerequisites":["user-interview-guide"],"aiLearningOutcomes":["Understand the four forces that drive switching decisions","Conduct a timeline interview to map the customer decision journey","Run a switch interview focused on recent adopters","Write well-formed job statements","Analyze JTBD interview data to inform product strategy"],"aiDifficulty":"intermediate","aiEstimatedTime":"10 min read"},{"type":"documentation","id":"94038d7c-0895-4369-b254-9d094bdddc6b","slug":"logo-testing","title":"Logo Testing: How to Test a Logo With Real People Before You Commit","url":"https://www.koji.so/docs/logo-testing","summary":"Logo testing shows a logo to people from the target audience and measures recall, what it communicates about the company, fit with brand words, distinctiveness and unwanted meanings, with liking asked last. Lindgaard et al. (2006) found visual appeal ratings after a 50 millisecond view were consistent with ratings after 500 milliseconds, so short timed views capture real first impressions. Tropicana's 2009 redesign was followed by a 20% fall in unit sales between 1 January and 22 February and a reversal on 23 February (Ad Age, IRI data). Gap replaced its logo on 6 October 2010 and went back on 12 October; its North America president said the company did not go about this in the right way. Familiarity raises liking (Zajonc, 1968), so new logos should be compared on communication and fit, not popularity. The guide gives an eight-step method, a ten-question bank, sample sizes of 15 to 30 per logo for themes and 30 or more for shares, common mistakes, and how Koji's question images, timed views and follow-up questions run a logo test.","content":"Logo testing means showing a logo to people from your target audience and measuring what they remember, what it tells them about the company and whether it fits the brand you are trying to build. The most useful tests do not ask \"Do you like it?\". They show the logo briefly, ask what people recall and what kind of company they picture, then compare those answers across two to four candidates. Liking is the least useful number in a logo test, because a new mark almost always loses to a familiar one at first.\n\nThis guide covers what a logo test can and cannot tell you, the measures that matter, a step-by-step method, a reusable question bank and the mistakes that sink most tests.\n\n## Why test a logo at all\n\nA logo is cheap to change on a design file and expensive to change once it is on a product, a website, packaging and every sales deck. Two well-known redesigns show what happens when the reaction comes after the launch:\n\n- **Tropicana, 2009.** PepsiCo replaced Tropicana's packaging, including the logo, in January 2009. According to Information Resources Inc. data reported by Ad Age, unit sales of Tropicana Pure Premium fell 20% between 1 January and 22 February, and dollar sales fell 19%, about $33 million. The company announced on 23 February that it would bring back the old design, less than two months after the change.\n- **Gap, 2010.** Gap introduced a new logo on 6 October 2010 and went back to the old blue box on 12 October, after a wave of criticism online. Marka Hansen, then president of Gap North America, said the company \"did not go about this in the right way\" and \"missed the opportunity to engage with the online community.\"\n\nNeither case proves a test would have saved the launch. Both show that customers have strong, fast reactions to brand marks, and that those reactions are much cheaper to hear before launch than after.\n\n## What a logo test can and cannot tell you\n\nPeople judge visual design very fast. Lindgaard and colleagues (2006) showed web pages to participants for just 50 milliseconds and found their ratings of visual appeal were consistent with ratings made after a 500 millisecond view. First impressions of a design form almost instantly, and they are fairly stable.\n\nThat is good news for logo testing: a short exposure gets a real reaction. But a logo test has limits you should plan around.\n\n**It can tell you:**\n\n- whether people can recall the name or shape after a brief look\n- what the logo suggests about the company: industry, price level, personality\n- whether it fits the words you want associated with the brand\n- whether it is confused with a competitor\n- whether it carries a meaning you did not intend, in some markets or groups\n\n**It cannot tell you:**\n\n- how people will feel about it after a year of seeing it on a product they like. Familiarity raises liking, an effect psychologists call the mere exposure effect, first described by Robert Zajonc in 1968. A new logo competing against your current one starts at a disadvantage on liking alone.\n- whether it will \"work\" commercially. A logo takes on the reputation of the company behind it over time, good or bad. No test can show you that in advance.\n\nSo test for **communication and fit**, not for a popularity contest.\n\n## The five measures that matter\n\n![Checklist of what to measure in a logo test, from recall after a short view to unwanted associations and legibility at small sizes](/visuals/logo-test-what-to-measure-checklist/logo-test-what-to-measure-checklist.png \"Measure what the logo communicates before you ask whether people like it\")\n\n1. **Recall.** After a short view, can people describe the logo or name? A mark people cannot remember after five seconds has a harder job ahead.\n2. **Communication.** What kind of company do they think this is? What does it sell, who for, at what price? Compare the answers with your positioning.\n3. **Fit with brand attributes.** Give people your three to five target words (for example \"trustworthy\", \"modern\", \"for professionals\") alongside a few you want to avoid, and ask which describe the logo.\n4. **Distinctiveness.** Does it remind them of another company? Show it next to two or three competitor logos and ask which is which.\n5. **Problems.** Unwanted associations, meanings in other languages or cultures, and legibility at small sizes such as an app icon or a favicon.\n\nLiking can be a sixth measure. Ask it last and weight it least.\n\n## How to run a logo test, step by step\n\n1. **Pick two to four finalists.** More than that and people blur them together. Include your current logo if you have one, so you have a baseline.\n2. **Write down what the logo should communicate.** Three to five brand words and one sentence on who the company is for. This is what you score against.\n3. **Recruit people from your target audience.** Existing customers react to change; prospects react to the mark itself. If you are rebranding, test both groups and report them separately.\n4. **Start with a timed first look.** Show one logo for about five seconds, hide it, then ask what people remember and what kind of company they think it is. This is the logic of a [five-second test](/docs/5-second-test-guide).\n5. **Then show it again, for as long as they like,** and ask the fit and association questions.\n6. **Rotate the order.** If each person sees more than one logo, change which one comes first, or use a [monadic design](/docs/monadic-testing-guide) where each person sees only one. Otherwise the first logo shown colours every answer after it.\n7. **Show it in context last.** On a product, an app icon, a storefront or a letterhead. Many logos that look fine on a white page fail at 32 pixels.\n8. **Analyse the open answers by theme** and compare candidates on recall, communication and fit, not on a single \"favourite\" vote.\n\n## A logo test question bank\n\n![Question bank for a logo test, with questions for a first impression after a timed view, brand fit, distinctiveness and context](/visuals/logo-test-question-bank/logo-test-question-bank.png \"Reusable questions for a logo test, from a timed first look to the logo in context\")\n\nUse these as a starting point and adjust the wording to your brand:\n\n**First impression (after a five-second view)**\n\n- What do you remember about the logo you just saw?\n- What kind of company do you think this is?\n- What do you think this company sells, and who for?\n\n**Brand fit**\n\n- Which of these words describe this logo? (show your target words mixed with words you want to avoid)\n- If this company were a person, how would you describe them?\n- Does this logo feel more premium or more budget to you? Why?\n\n**Distinctiveness**\n\n- Does this logo remind you of any other company or product?\n- What, if anything, would make it hard to tell apart from others in this space?\n\n**In context**\n\n- Here it is on the product. Does it look like it belongs there?\n- Is there anything about it that bothers you or feels off?\n\nFollow every answer with \"What makes you say that?\". The reasons are where you find out that \"modern\" meant \"cold\", or that a shape reads as a medical cross.\n\n## How many people to test with\n\nFor a qualitative logo test focused on reasons and associations, 15 to 30 people per logo usually surfaces the main themes. If you want to compare candidates on shares, such as the percentage who correctly recall the name, aim for at least 30 people per logo, and more if the differences are small. The [survey sample size guide](/docs/survey-sample-size-guide) covers the maths.\n\n## Common mistakes\n\n- **Asking \"Which do you like best?\" first.** It turns the test into a beauty contest and primes every later answer. Ask about recall and meaning first.\n- **Testing the new logo against the old one on liking.** The old one has years of familiarity behind it. Compare them on communication and fit instead.\n- **Showing logos side by side for the first look.** People compare instead of reacting. Show one at a time.\n- **Testing only on a white background.** Check the logo at small sizes and in the places it will actually live.\n- **Testing with designers and colleagues.** They look at craft. Customers look at what it says about the company.\n- **Letting one loud comment decide.** One person seeing a rude shape is worth checking; it is not a verdict. Look at how often a theme comes up.\n\n## Running the test with real people\n\nA logo test lives or dies on the people in it. Recruit from the audience you want to win, not from whoever is nearby. Give each person a quiet first look without anyone explaining the design, because a designer's explanation becomes the answer. And let people talk: the most useful findings in logo tests are usually in a throwaway remark (\"it looks like a bank\") rather than in a rating.\n\n![A small group looking at printed design sketches on a table and talking about their first reactions to each one](https://images.unsplash.com/photo-1683199804561-ce75c2c7bf61?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHw4fHxwZW9wbGUlMjByZXZpZXdpbmclMjBkZXNpZ24lMjBza2V0Y2hlcyUyMHRhYmxlfGVufDF8MHx8fDE3OTMyNTY2MTF8MA&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Sweet Life](https://unsplash.com/@sweetlifediabetes?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\n## How Koji helps\n\nKoji runs AI-moderated interviews by voice or text, so every logo reaction comes with the reasons behind it.\n\n- **Attach the logo to any question.** Upload an image of a logo, a mockup or the logo on a product, and participants see it with the question. Koji knows what the image shows, so its follow-up questions are about the logo, not generic.\n- **Timed first-impression views.** Set an image to be shown once, full screen, for 3, 5, 10 or 15 seconds, then hidden. Koji then asks what people remember and what they think, which is exactly the first step of a logo test.\n- **Follow-ups that ask why.** For each question you choose how many follow-ups Koji asks, from none to three, so \"it looks cheap\" turns into what specifically made it look cheap.\n- **Fair comparisons.** Choice questions can shuffle their answer options for each participant, with the last few options kept in place for answers like \"None of these\", so brand words and candidate names are not always read in the same order.\n- **Themes, not just votes.** The report groups open answers into themes with the quotes behind each one, so you can compare logos on what they communicate.\n\n### Set up a logo test in Koji\n\n1. Add one question per logo, for example \"What do you remember about the logo you just saw?\".\n2. In the question editor, add an image to that question: a PNG, JPEG or WebP of the logo, or of the logo on a product.\n3. Check the description Koji writes of what the image shows, and edit it if it misses something. Screen readers and voice interviews rely on it.\n4. For the first look, set the view to \"Visible for 5 seconds\" (or 3, 10 or 15). Participants tap \"Show image\", see the logo once, full screen, and then Koji asks the question. Leave it on \"No time limit\" for the fit and context questions, so people can look as long as they like.\n5. Add the brand-word question as a choice question with shuffled options, and set how many follow-ups each open question gets.\n\nImages work in text and voice interviews, not on phone calls. Uploaded images are resized, stripped of metadata and stored privately. You can pair a logo test with a [purchase intent question](/docs/purchase-intent-scale) or a price test when you are testing a whole new brand or package.\n\n![Koji research report with participant, duration, themes and quality stats above a summary of findings](/email/shots/shot-report.jpg \"The report groups what people said about each logo into themes, with the quotes behind them\")\n\n## Related Resources\n\n- [Five-second test guide](/docs/5-second-test-guide)\n- [Preference testing guide](/docs/preference-testing-guide)\n- [Monadic testing guide](/docs/monadic-testing-guide)\n- [Brand perception survey guide](/docs/brand-perception-survey-guide)\n- [Name testing research](/docs/name-testing-research)\n- [Structured questions guide](/docs/structured-questions-guide)\n\n## Sources\n\n- Lindgaard, G., Fernandes, G., Dudek, C., & Brown, J. (2006). Attention web designers: You have 50 milliseconds to make a good first impression! _Behaviour & Information Technology_, 25(2), 115–126.\n- _Advertising Age_ (2009). Tropicana line's sales plunge 20% post-rebranding. April 2009, reporting Information Resources Inc. data.\n- Christian Science Monitor (2010). New Gap logo withdrawn: The \"blue box\" lives on. 12 October 2010.\n- Zajonc, R. B. (1968). Attitudinal effects of mere exposure. _Journal of Personality and Social Psychology_, 9(2), 1–27.\n","category":"Research Methods","lastModified":"2026-10-31T05:49:14.617361+00:00","metaTitle":"Logo Testing: Method, Questions and Mistakes (2026 Guide)","metaDescription":"How to test a logo with real people: timed first impressions, recall, brand fit and distinctiveness, a 10-question bank, sample sizes and common mistakes.","keywords":["logo testing","how to test a logo","logo test questions","logo testing survey","logo preference test","test logo with customers","logo research"],"aiSummary":"Logo testing shows a logo to people from the target audience and measures recall, what it communicates about the company, fit with brand words, distinctiveness and unwanted meanings, with liking asked last. Lindgaard et al. (2006) found visual appeal ratings after a 50 millisecond view were consistent with ratings after 500 milliseconds, so short timed views capture real first impressions. Tropicana's 2009 redesign was followed by a 20% fall in unit sales between 1 January and 22 February and a reversal on 23 February (Ad Age, IRI data). Gap replaced its logo on 6 October 2010 and went back on 12 October; its North America president said the company did not go about this in the right way. Familiarity raises liking (Zajonc, 1968), so new logos should be compared on communication and fit, not popularity. The guide gives an eight-step method, a ten-question bank, sample sizes of 15 to 30 per logo for themes and 30 or more for shares, common mistakes, and how Koji's question images, timed views and follow-up questions run a logo test.","aiPrerequisites":["Two to four logo candidates as image files","A short list of the words you want the brand to stand for"],"aiLearningOutcomes":["Explain what a logo test can and cannot predict","Measure recall, communication, brand fit, distinctiveness and unwanted meanings","Run a logo test with a timed first look and rotated order","Use a ten-question bank for logo tests","Avoid the mistakes that turn logo tests into beauty contests"],"aiDifficulty":"beginner","aiEstimatedTime":"9 min read"},{"type":"documentation","id":"6a9b1b64-b712-4571-bb7a-83a939d3be0d","slug":"5-second-test-guide","title":"The 5-Second Test: How to Measure First Impressions and Visual Hierarchy (2026 Guide)","url":"https://www.koji.so/docs/5-second-test-guide","summary":"A complete practitioner guide to the 5-second test — a usability method that captures first impressions of a design by showing it for exactly 5 seconds, then asking recall and sentiment questions. Covers methodology, the Lindgaard 50ms research foundation, question design, sample sizes, and how AI-moderated platforms like Koji enhance the method.","content":"## The 5-second test in 30 seconds\n\nThe 5-second test is a usability research method where you show participants a design — a landing page, hero section, ad creative, or onboarding screen — for **exactly five seconds**, then hide it and ask them what they remember and understood. The technique measures what users absorb during the brief window when humans actually form first impressions: research by Lindgaard et al. found people form aesthetic judgments about web pages in as little as **50 milliseconds**, and most users decide whether to stay on a page within 10–20 seconds. ([Nielsen Norman Group](https://www.nngroup.com/videos/5-second-usability-test/))\n\nFive seconds is short enough that participants can't read body copy or memorize details — but long enough to register the headline, dominant visual, and primary call-to-action. That makes it the cleanest way to evaluate **clarity of message, visual hierarchy, brand perception, and CTA prominence**. Modern AI-native platforms like Koji extend the method beyond the standard timed-image format, pairing the screen reveal with conversational follow-up that probes the *why* behind every recall.\n\n---\n\n## What is a 5-second test?\n\nA 5-second test is a quantitative-leaning usability method designed to capture the gut reaction users have to a design *before* they have time to overthink it. The flow is simple:\n\n1. The participant is told they will see a screen for a short time.\n2. The image is shown for exactly five seconds.\n3. The image disappears.\n4. The participant answers a small set of recall and reaction questions.\n\nNotice the test does *not* ask \"did you like it?\" while the image is visible — that defeats the entire purpose. The whole methodology hinges on hiding the image before participants can hunt for the answer.\n\n## Why five seconds? The science of first impressions\n\nThe original 5-second test was popularized in the late 2000s, but the research foundation is older. Three studies are critical to understanding why this duration works:\n\n**Lindgaard et al. (2006)** — *Attention web designers: You have 50 milliseconds to make a good first impression* — found that visual-appeal judgments were stable from 50 ms exposures all the way through 500 ms exposures. Once a user has glanced at a page, their aesthetic verdict is essentially locked in. ([ResearchGate](https://www.researchgate.net/publication/220208334_Attention_web_designers_You_have_50_milliseconds_to_make_a_good_first_impression_Behaviour_and_Information_Technology_252_115-126))\n\n**Jakob Nielsen** observed that \"users often leave web pages in 10–20 seconds, but pages with a clear value proposition can hold people's attention for much longer.\" Five seconds sits inside that initial-decision window. ([Nielsen Norman Group](https://www.nngroup.com/articles/how-long-do-users-stay-on-web-pages/))\n\n**Cognitive load research** demonstrates that five seconds is too short for systematic reading but long enough for **automatic processing** — pattern recognition, color absorption, layout parsing. NN/g researcher Therese Fessenden notes that \"first impressions are often automatic, not deliberate. Designers can't override automaticity, but they can design for it.\" ([Nielsen Norman Group](https://www.nngroup.com/articles/first-impressions-human-automaticity/))\n\nIn short: five seconds isolates the *automatic* layer of cognition. Anything users tell you they remembered after that window represents the strongest signal in the design — the parts that survived without conscious attention.\n\n## When to use a 5-second test\n\nUse it when you need to evaluate:\n\n- **Message clarity** — does the headline communicate what the product does?\n- **Visual hierarchy** — does the eye land on the right element first?\n- **Brand perception** — does the design feel premium / friendly / serious / playful?\n- **Primary CTA recall** — do users remember the action the page wants them to take?\n- **Logo and identity recognition** — is the brand identity strong enough to register?\n- **Ad creatives** — does a banner or social post communicate the value prop in a glance?\n\nDo *not* use 5-second tests for evaluating long-form content, complex flows, deep features, or anything below the fold. The method is a flashbulb, not a microscope.\n\n## How to design 5-second test questions\n\nQuestion design is where most 5-second tests fail. Bad questions either prime the participant or measure something other than first impressions.\n\n### The five canonical question types\n\n**1. Open-ended recall**\n- \"What do you remember about the page?\"\n- \"What is this product or service?\"\n- \"Who is this designed for?\"\n\n**2. Comprehension**\n- \"What did the page say it does?\"\n- \"What action did the page want you to take?\"\n\n**3. Sentiment**\n- \"What was your first impression?\"\n- \"On a scale of 1 to 5, how trustworthy did the page feel?\"\n\n**4. Brand attribute recall**\n- \"What three words describe the design's personality?\"\n- \"Did the design feel modern, traditional, or somewhere in between?\"\n\n**5. Memory check**\n- \"Was there a price visible?\" (binary, fact-checking attention)\n- \"What color was the main button?\" (testing whether the CTA broke through)\n\nA well-constructed test combines 4–6 questions. More than that and you're no longer testing first impressions — you're running a memory study.\n\n### Question types you should always avoid\n\n- \"Did you like the design?\" — too leading and too vague\n- \"Was the navigation easy to use?\" — they couldn't use it; this measures nothing\n- \"Did the design feel cluttered?\" — primes participants toward a specific judgment\n- Anything that requires reading body copy — five seconds isn't enough\n\n## Sample size for 5-second tests\n\nThe Nielsen Norman Group rule of thumb for unmoderated visual tests is **20 participants per variant** as a strong baseline. For directional insight on a single design, you can run useful tests with as few as **6–10 participants**, though signal will be noisy. For statistical confidence on attribute or sentiment ratings, aim for **30–50 participants**. ([Nielsen Norman Group](https://www.nngroup.com/articles/testing-visual-design/))\n\nFor A/B comparisons of two designs, double the sample so each variant gets its own group. Mixed-method 5-second tests that include open-ended recall benefit from larger samples because you need enough text responses to spot themes — Koji's [thematic analysis](/docs/thematic-analysis-guide) typically requires 20+ open responses to surface stable themes.\n\n## Step-by-step: running a 5-second test\n\n### 1. Pick a single design to test\n\nThe golden rule: **one stimulus per study**. Mixing four landing pages into one session contaminates the data because participants compare instead of reacting. Run separate tests, or use a forced-randomization design.\n\n### 2. Prepare the image at correct dimensions\n\nUse the resolution your real users see. Testing a desktop landing page on a mobile-cropped image will distort hierarchy. If the design has both desktop and mobile variants, run two separate tests.\n\n### 3. Set up the timing mechanic\n\nDedicated platforms like Lyssna, UserTesting, and Maze auto-hide the image after five seconds. If you're running it on a generic survey tool, you cannot reliably enforce timing — participants will linger. Don't try to retrofit a 5-second test into SurveyMonkey.\n\n### 4. Write the prep text\n\nTell participants: *\"You will see a screen for five seconds. Look at it carefully. After it disappears, you'll be asked questions about what you saw.\"* Don't hint at what they should look for. Don't mention the brand. Don't describe the test as \"first impressions\" until afterward.\n\n### 5. Recruit your audience\n\nMatch participants to the page's real visitors. Use [screener questions](/docs/screener-questions-guide) to filter for relevant demographics, prior brand awareness, or task context. A 5-second test of an enterprise security landing page should not be run on consumers who've never bought B2B software.\n\nPick the route that fits the test. If the page is for people you already have, customers, users in your product, a CRM segment, import them so each person gets a personalized link: you see who responded and the report is already segmented by whatever you imported. If you need people who have never heard of the brand, open Panel recruit, describe the audience, and Koji finds them across 130+ countries at a price per completed interview in credits you approve before launch (launching is on paid plans); your screener questions run before the interview. If you prefer to run recruitment yourself, marketplaces like User Interviews, Respondent or Prolific can source a specialist audience: share your Koji link with the people they find and the interviews land in the same study.\n\n### 6. Run the test and collect responses\n\nAim for completion within 24–48 hours of launch. Quick turnaround prevents stale recruitment and gives you a clean cohort. Most platforms deliver 30 participants in under a day.\n\n### 7. Analyze the data\n\nLook for four signals:\n\n- **Recall accuracy** — did participants identify what the product does? Strong designs get 80%+ correct.\n- **CTA recognition** — did participants remember the primary action? Below 50% means your CTA isn't breaking through.\n- **Sentiment distribution** — are reactions clustered or split? Bimodal sentiment usually signals brand confusion.\n- **Theme density** — what words appeared most in open recall? These are the elements that *survived* the five seconds.\n\n## How to run a 5-second test in Koji\n\nKoji runs the timed view itself, inside an AI-moderated interview, so the recall questions that follow can ask why.\n\n1. **Attach the design to a question.** In the question editor, add an image (PNG, JPEG or WebP) to the question you want people to answer after the view, such as \"What do you remember about the page you just saw?\". Koji writes a short description of what the image shows, which you can edit. Screen readers and voice interviews use it, and the AI moderator uses it to ask follow-ups about the design rather than generic ones.\n2. **Set the timed view.** Choose \"Visible for 5 seconds\" (3, 10 and 15 seconds are there too, or no time limit). Participants tap \"Show image\", see it once, full screen, with a countdown bar along the top, and then it is hidden. It is shown once per participant, so nobody can go back and hunt for the answer.\n3. **Ask first, then probe.** Koji asks the question only after the image has been seen, so people answer from the picture and not from the wording. Then it follows up on what they said:\n   - A participant says \"I think it's a project management tool\" and Koji asks what about it made them think that.\n   - A participant rates trustworthiness 2 out of 5 and Koji asks what would move it up.\n   - A participant can't recall the call to action and Koji asks what their eye landed on first.\n\n   You set how many follow-ups each question gets, from none to three, so a pure recall check can stay a single question.\n4. **Mix in structured checks.** [Structured questions](/docs/structured-questions-guide) (open-ended, scale, single choice, multiple choice, ranking and yes/no) sit in the same interview, so a yes/no \"Was there a price on the page?\" can follow the open recall question. Shuffle choice options for each participant when you list brand words or design names.\n5. **Read the themes.** The [report](/docs/generating-research-reports) groups open answers into themes with the quotes behind each one, next to the charts for your structured questions.\n\nTimed images work in text and voice interviews, not on phone calls. Uploaded images are resized, stripped of metadata and stored privately. For a full first-impression study of a brand mark, the [logo testing guide](/docs/logo-testing) uses the same setup.\n\n![Koji research report with participant, duration, themes and quality stats above a summary of findings](/email/shots/shot-report.jpg \"After the timed view, the report groups what people remembered into themes with the quotes behind them\")\n\n\n## Common 5-second test mistakes\n\n1. **Testing the wrong fidelity.** A wireframe with placeholder copy will fail every recall question — participants can't remember \"Lorem ipsum\" because it isn't a message.\n2. **Asking too many questions.** Six questions is the upper limit. Beyond that, recall decays and the data is noise.\n3. **Showing the same design to the same participant twice.** Repeat exposure breaks the methodology entirely.\n4. **Including movement or video.** Animated heroes change what participants see at second one vs second five. Use static frames or run a separate motion test.\n5. **Skipping comprehension questions.** \"Did you like it?\" without \"What does it do?\" gives you sentiment without grounding. Always pair feeling with understanding.\n6. **Comparing competitor designs against your own.** Halo effects from brand awareness contaminate the test. Run them as separate, blinded studies.\n\n## 5-second test vs. preference test vs. first-click test\n\n| Method | What it measures | Time per task | Best for |\n|---|---|---|---|\n| 5-second test | First impression, message clarity, visual hierarchy | ~30 sec | Hero sections, ads, brand identity, CTA prominence |\n| Preference test | Which of N options users prefer and why | ~1 min | A/B variant selection, brand direction |\n| [First-click test](/docs/first-click-testing-guide) | Findability of a specific element | ~30 sec | Navigation labels, IA validation |\n\nThe three methods are complementary. Run a 5-second test to validate hero clarity, a first-click test to validate navigation findability, and a preference test to choose between two finalist directions.\n\n## Worked example: a fintech homepage\n\nAn illustrative example (the numbers show how to read results; they are not from a real study). A fintech team tests two homepage hero variants, one built around \"Save smarter\" and one around \"Spend less, invest more\", with 30 participants on each:\n\n- **Variant A (\"Save smarter\")**: 73% identify the product as a savings app, 41% recall the \"Open account\" button and trustworthiness averages 3.4 out of 5.\n- **Variant B (\"Spend less, invest more\")**: 38% think it is a budgeting app, 22% think it is an investment platform, only 19% recall the button and trustworthiness averages 2.9 out of 5.\n\nThe follow-up answers on Variant B explain the split: in five seconds, people were torn between two ideas, tracking spending and investing. Variant A communicates one idea, so it is the one to take forward, and a live A/B test then tells you whether clearer recall turns into sign-ups.\n\n\n## Related Resources\n\n- [Logo Testing](/docs/logo-testing): how to test a logo with a timed first look, a question bank and the measures that matter\n- [Structured Questions Guide](/docs/structured-questions-guide) — Combine recall, scale, and sentiment questions in one study\n- [First-Click Testing Guide](/docs/first-click-testing-guide) — Validate findability and navigation\n- [Usability Testing Guide](/docs/usability-testing-guide) — Evaluate full task flows beyond first impressions\n- [Concept Testing Methodology](/docs/concept-testing-methodology) — Test propositions before designing the page\n- [Brand Perception Survey Guide](/docs/brand-perception-survey-guide) — Run brand recall studies at scale\n- [Thematic Analysis Guide](/docs/thematic-analysis-guide) — Code open-ended recall responses\n\n\n## Further reading on the blog\n\n- [Best AI Market Research Tools in 2026: The Complete Buyer's Guide](/blog/ai-market-research-tools-2026) — AI has fundamentally changed market research. This guide compares the leading AI market research platforms—from AI-native interview tools li\n- [B2B Customer Research: The Complete Guide for Product Teams (2026)](/blog/b2b-customer-research-guide-2026) — B2B customer research is harder than B2C — you are navigating buying groups of 10+ stakeholders, gatekeepers, and enterprise procurement cyc\n- [B2C User Research: How to Understand Consumer Behavior at Scale (2026)](/blog/b2c-user-research-guide-2026) — B2C user research is systematically underinvested at most consumer companies. While B2B teams run structured customer discovery as a matter \n\n<!-- further-reading:blog -->\n","category":"Research Methods","lastModified":"2026-10-31T05:48:47.429251+00:00","metaTitle":"5-Second Test: First Impressions Methodology Guide (2026) | Koji","metaDescription":"Run effective 5-second tests to validate visual hierarchy, message clarity, and brand perception. Learn the methodology, recommended sample sizes, question design, and how AI moderation extends the method.","keywords":["5 second test","five second test","first impression test","visual hierarchy testing","landing page testing","5-second test ux","first impression usability"],"aiSummary":"A complete practitioner guide to the 5-second test — a usability method that captures first impressions of a design by showing it for exactly 5 seconds, then asking recall and sentiment questions. Covers methodology, the Lindgaard 50ms research foundation, question design, sample sizes, and how AI-moderated platforms like Koji enhance the method.","aiPrerequisites":["Basic familiarity with usability testing","A static design (landing page, hero, ad)","Defined target audience"],"aiLearningOutcomes":["Understand when 5-second tests are appropriate vs other UX methods","Design recall, comprehension, and sentiment questions that avoid priming","Choose appropriate sample size (10-50 participants) based on confidence needs","Analyze recall accuracy, CTA recognition, and sentiment distribution","Combine timed reveal with AI follow-up probing for deeper insight"],"aiDifficulty":"beginner","aiEstimatedTime":"11 min"},{"type":"documentation","id":"cb3352f7-69db-4095-a130-00f59ef8b7f4","slug":"gabor-granger-pricing-method","title":"Gabor-Granger Pricing Method: Find Your Revenue-Maximizing Price","url":"https://www.koji.so/docs/gabor-granger-pricing-method","summary":"The Gabor-Granger method finds the revenue-maximizing price by asking respondents whether they would buy a product at a series of prices, then aggregating responses into a demand curve and a revenue curve whose peak is the optimal price. Created by economists André Gabor and Clive Granger, it is fast and decision-ready but works best when respondents are real buyers and the reasons behind each rejection are captured.","content":"# Gabor-Granger Pricing Method: How to Find Your Revenue-Maximizing Price\n\n**Bottom line:** The Gabor-Granger method is a direct survey technique that finds your revenue-maximizing price by showing each respondent a single product at a series of prices and asking, at each one, whether they would buy. Aggregating the \"yes\" responses across everyone produces a demand curve (purchase likelihood at each price) and, when multiplied by price, a revenue curve whose peak is your optimal price point. It's the fastest quantitative way to answer \"what should we charge?\" — and it's far more reliable when the buyers you survey are the right ones and you also capture *why* they say no.\n\nPricing is the highest-leverage number in your business. A [McKinsey analysis](https://www.mckinsey.com/capabilities/growth-marketing-and-sales/our-insights/the-power-of-pricing) of a typical S&P 1500 company found that a 1% improvement in price, with volume held constant, lifts operating profit by an average of more than 8% — a larger effect than equivalent improvements in variable cost, sales volume, or fixed cost. Yet most teams set price by copying competitors or marking up cost. Gabor-Granger replaces that guesswork with willingness-to-pay data straight from buyers.\n\n## Who created it?\n\nThe method is named after economists André Gabor and Sir Clive Granger — the latter a 2003 Nobel laureate in Economic Sciences — who developed it in the 1960s to study how price affects purchase intent. Decades later it remains a staple of pricing research because it is simple to run, fast to analyze, and produces a directly actionable output: a price.\n\n## How the Gabor-Granger method works\n\n1. **Pick a price range.** Choose a realistic span of prices to test, usually five to seven points bracketing your candidate price.\n2. **Show one price at a time.** Each respondent sees the product at a randomly selected starting price and answers a purchase-intent question (\"How likely are you to buy at $X?\" or a simple yes/no).\n3. **Adapt the next price.** If they would buy, show a higher price; if they would not, show a lower one. This continues until you have found each respondent's highest acceptable price.\n4. **Aggregate into a demand curve.** Across all respondents, calculate the share who would buy at each price. As price rises, that share falls — that is your demand curve.\n5. **Derive the revenue curve.** Multiply purchase likelihood at each price by the price itself. The result typically rises, peaks, and falls. The peak is your revenue-maximizing price.\n\nBecause higher prices reduce the number of buyers but raise revenue per sale, the two effects trade off — and the revenue curve's peak is the price where that trade-off is optimized.\n\n## A worked example\n\nSuppose you test a SaaS plan at $20, $30, $40, $50, and $60:\n\n| Price | % who would buy | Revenue index (price × %) |\n|---|---|---|\n| $20 | 85% | 17.0 |\n| $30 | 72% | 21.6 |\n| $40 | 58% | 23.2 |\n| $50 | 38% | 19.0 |\n| $60 | 21% | 12.6 |\n\nDemand falls steadily as price climbs, but revenue peaks at **$40**, where the higher price still retains enough buyers. Drop to $30 and you leave money on the table; push to $50 and you lose more buyers than the price increase recovers. Gabor-Granger surfaces that peak directly.\n\n## What Gabor-Granger is good at — and where it falls short\n\n**Strengths:**\n- Fast and inexpensive to field.\n- Produces a clear, decision-ready output: demand, revenue, and an optimal price.\n- Lets you estimate price elasticity for a single product.\n\n**Limitations:**\n- It tests one product in isolation — it ignores competitor prices and substitution (use conjoint analysis or a competitive study for that).\n- Stated willingness to pay tends to overstate real willingness to pay; people say they will buy at prices they would not actually pay.\n- It surfaces a revenue-maximizing price, not necessarily a profit-, growth-, or positioning-optimal one.\n- It tells you *what* price people accept, not *why* — the reasoning that tells you whether you can move the curve with better packaging or messaging.\n\nThat last gap is the important one. A number without a narrative is dangerous in pricing, because the \"why\" is what tells you whether a low willingness-to-pay reflects the price, the perceived value, or simply the wrong audience.\n\n## How to run Gabor-Granger in Koji\n\nKoji has a dedicated **Price Test (Gabor-Granger)** question type, so the ladder runs inside an AI interview instead of a survey with hand-built branching logic. You can set one up yourself from a free account, next to every other kind of question in the same study.\n\n**Setting it up**\n\n- **Describe what is priced.** The question text is the offer itself (\"Team plan: unlimited projects, 5 seats\"). Koji introduces it once and the screen asks the prices. If the question has an image, participants see it with the price.\n- **Enter 3 to 10 prices.** The editor recommends 5 to 7, evenly spaced. Pick a currency per question and an optional unit such as \"per month\".\n- **Choose who counts as a buyer.** Every price is rated on a 5-point purchase-intent scale, from \"Definitely would not buy\" to \"Definitely would buy\". By default \"Probably would buy\" or higher counts as a buy (the usual top-2-box rule). You can make it stricter (\"Definitely\" only) or more generous (\"Might or might not\" and up).\n- **Choose where people start.** By default each participant starts at a random price in the middle of the ladder, so nobody anchors on the cheapest one. You can let the start fall anywhere on the ladder instead.\n\nPrices lock once interviews start, so every participant sees the same ladder.\n\n**What each participant does**\n\nKoji shows one price at a time. If the answer counts as a buy, the next price is higher; if not, it is lower. The walk stops when it finds the highest price that person would still buy at, usually in about 3 steps rather than the full ladder. Then the AI interviewer asks why, in the participant's own words: what they compared it to, what would make the price feel fair, what the higher price would need to include. It works in text and voice interviews, not on phone calls.\n\n**What the report shows**\n\n- The share of participants who would buy at each price (the demand curve)\n- The revenue each price implies and the price that earns the most (the revenue curve)\n- When a cheaper price earns nearly as much, that trade-off spelled out, so you can weigh reach against revenue\n- The quotes behind each point, so a drop in demand comes with its reasons\n\nThe report labels the result as directional, for two reasons this guide already covers: people overstate what they would buy, and the best price is always one of the prices you tested. It reads best from about 30 interviews. That is enough to see the shape of the curve and hear the reasons; for a precise price point per segment, the classic advice of 100 or more buyers per segment still applies.\n\n**Where the AI interview adds something a survey can't**\n\n- **The \"why\" is built in.** A flat yes or no becomes a reason you can act on. You see not only that demand drops above a price, but whether it drops because buyers compare you to a cheaper alternative (a positioning problem) or because the offer is missing something (a packaging problem).\n- **Automatic thematic analysis** groups those reasons across everyone, so you aren't reading hundreds of answers by hand.\n- **Screening questions** keep people who aren't real buyers out of the study, which matters more for pricing accuracy than raw volume.\n- **Concept and price together.** Show the product, landing page or packaging as an image on an earlier question, collect first reactions, then run the price test in the same interview.\n- **Set it up by asking.** Tell the research assistant \"help me test pricing for our team plan\" and it drafts the study, price ladder included. AI assistants connected to Koji can create and read price tests too.\n\nSee the [structured questions guide](/docs/structured-questions-guide) for how the price test sits alongside scale, choice, ranking and yes/no questions, and the [purchase intent scale guide](/docs/purchase-intent-scale) for how to read the 5-point scale.\n## Gabor-Granger vs Van Westendorp vs conjoint\n\n- **Gabor-Granger** finds the revenue-maximizing price for one product via direct purchase-intent at set prices. Best when you have a specific product and a price range in mind.\n- **Van Westendorp Price Sensitivity Meter** maps a *range* of acceptable prices using four open-ended price-perception questions. Best for early-stage products with no anchor price yet.\n- **Conjoint analysis** models how price trades off against features and brand in a realistic, competitive choice. Best when price is one of several variables and you need to understand substitution.\n\nMany teams sequence them: Van Westendorp to find the plausible range, Gabor-Granger to pinpoint the revenue peak, and qualitative interviews to understand the \"why\" behind both.\n\n## Best practices\n\n- **Screen hard for real buyers.** A clean curve from the wrong audience is precise and useless.\n- **Bracket the range generously.** If demand stays high at your top price, your range was too low and you have capped your own answer.\n- **Pair numbers with reasons.** Always capture *why* a price was rejected — it is the difference between lowering price and raising perceived value.\n- **Validate against behavior.** Treat stated willingness to pay as a directional input, then confirm with a real pricing test or pre-orders where you can.\n- **Use enough respondents.** Aim for at least ~100 qualified buyers per segment so each point on the demand curve is stable.\n\n## Related Resources\n\n- [Choosing the Prices to Test](/docs/how-many-prices-to-test-price-ladder): how many price points to use, and how far apart\n- [Structured Questions Guide](/docs/structured-questions-guide): the price test next to scale, choice, ranking and yes/no questions\n- [Purchase Intent Scale](/docs/purchase-intent-scale): how to ask \"would you buy this?\" and read the answer\n- [Van Westendorp Price Sensitivity Meter](/docs/van-westendorp-price-sensitivity-meter) — find your acceptable price range\n- [Pricing Research Interviews](/docs/pricing-research-interviews) — the qualitative \"why\" behind willingness to pay\n- [Pricing Research Survey Guide](/docs/pricing-research-survey-guide) — design rigorous pricing surveys\n- [Conjoint Analysis Guide](/docs/conjoint-analysis-guide) — model price against features and competition\n- [Willingness to Pay Interview Template](/docs/willingness-to-pay-interview-template) — structure WTP conversations","category":"Research Methods","lastModified":"2026-10-31T05:15:12.235238+00:00","metaTitle":"Gabor-Granger Pricing Method: Find Your Optimal Price","metaDescription":"Learn the Gabor-Granger pricing method: how the demand and revenue curves work, a worked example, limitations, Gabor-Granger vs Van Westendorp vs conjoint, and how to run it with AI.","keywords":["Gabor-Granger pricing method","Gabor Granger method","optimal price research","revenue-maximizing price","price sensitivity research","willingness to pay survey","price elasticity research","pricing research method","demand curve survey","Gabor-Granger vs Van Westendorp"],"aiSummary":"The Gabor-Granger method finds the revenue-maximizing price by asking respondents whether they would buy a product at a series of prices, then aggregating responses into a demand curve and a revenue curve whose peak is the optimal price. Created by economists André Gabor and Clive Granger, it is fast and decision-ready but works best when respondents are real buyers and the reasons behind each rejection are captured.","aiPrerequisites":["Basic familiarity with pricing or market research concepts"],"aiLearningOutcomes":["Explain how the Gabor-Granger method estimates demand and revenue curves","Identify the revenue-maximizing price from purchase-intent data","Recognize the methods limitations, including stated vs real willingness to pay","Choose between Gabor-Granger, Van Westendorp, and conjoint analysis","Run a Gabor-Granger study that captures both the price and the reasoning"],"aiDifficulty":"intermediate","aiEstimatedTime":"10 minutes"},{"type":"documentation","id":"b729f6ff-3755-461b-8610-923ad5995453","slug":"how-many-prices-to-test-price-ladder","title":"Choosing the Prices to Test: How Many Price Points, and How Far Apart","url":"https://www.koji.so/docs/how-many-prices-to-test-price-ladder","summary":"For a Gabor-Granger price test, use 5 to 7 prices, evenly spaced (or spaced by percentage for wide ranges), from a floor just above the cheapest credible alternative to a ceiling most people would refuse. Start each participant at a random price in the middle to spread anchoring, count Probably would buy or higher as a buy, and treat results as directional because stated willingness to pay runs about 21% above real payment on average.","content":"# Choosing the Prices to Test: How Many Price Points, and How Far Apart\n\n**Short answer:** for a Gabor-Granger price test, use 5 to 7 prices, evenly spaced, running from a floor just above your cheapest credible alternative to a ceiling you suspect is too high. Fewer than 5 leaves big gaps where the best price could hide. More than 7 rarely helps, because the best price you can find is always one of the prices you tested, and the extra points mostly split the same answers more thinly. Set the ceiling high enough that some people turn it down: a ladder where most people still buy at the top price can't tell you where demand ends.\n\nIf you're a founder setting a first price, this page is the part most guides skip. The method is simple (show one price at a time, ask \"would you buy at this price?\", move up or down), but the answer can only be as good as the ladder you give it. Pick the wrong prices and the test will confidently tell you the best of a bad set.\n\n**Try it this week:** write down three numbers before you read on: what your cheapest real alternative costs, what you'd love to charge, and a price that makes you nervous. Those three numbers are the skeleton of your ladder.\n\n## Why the ladder matters so much\n\nPrice is the lever with the most effect on profit. A [McKinsey analysis](https://www.mckinsey.com/capabilities/growth-marketing-and-sales/our-insights/the-power-of-pricing) of a typical S&P 1500 company found that a 1% price increase, with volume held steady, lifts operating profit by about 8%. Getting the number roughly right is worth the hour it takes to design the test properly.\n\nWarren Buffett put the stakes more bluntly in his 2010 interview with the Financial Crisis Inquiry Commission: \"The single most important decision in evaluating a business is pricing power. If you've got the power to raise prices without losing business to a competitor, you've got a very good business.\"\n\nA Gabor-Granger test estimates that power for one offer. It can only see the prices you put on the ladder. Three things go wrong most often:\n\n1. **The ceiling is too low.** If most people still say yes at your highest price, the real limit is somewhere above it and the test can't tell you where. The \"best price\" it reports is just your top price.\n2. **The floor is too high.** If many people say no even at your lowest price, either the offer isn't landing or you're talking to the wrong people. Either way the demand curve has nothing to start from.\n3. **The steps are too coarse.** With 3 prices spread from €10 to €100, the best price lands on one of three numbers that are far apart. The true peak could be anywhere between them.\n\n## How many prices to test\n\n**Use 5 to 7.** That's the standard advice for Gabor-Granger, and it's what Koji's editor recommends. Koji accepts between 3 and 10.\n\nThe reason more prices don't cost much is the way the test walks the ladder. Each participant doesn't see every price. They start somewhere in the middle and move up after a yes and down after a no, halving the gap each time, until the highest price they'd still buy at is pinned down. With 7 prices, that takes about 3 answers per person. With 10, it takes about 4. So the participant burden barely changes.\n\nWhat does change is how your sample spreads out. Every price you add splits your participants' answers into more, smaller groups at the top of their range. With about 30 interviews, 5 to 7 prices give each point enough answers to see the shape of the curve. With 10 prices and 30 people, neighbouring points start to bounce around for no real reason.\n\n| Prices  | When it fits                                                                |\n| ------- | --------------------------------------------------------------------------- |\n| 3 to 4  | You already know the range well and are choosing between a few real options |\n| 5 to 7  | The default: a first price, a new tier, a price change                      |\n| 8 to 10 | A wide range you know little about, and a sample well above 30              |\n\n## Where to put the floor and the ceiling\n\nStart from evidence you already have, not from the price you hope for.\n\n**The floor:** a little above the cheapest credible way your buyer solves the problem today. That might be a competitor's entry plan, a spreadsheet plus an hour of someone's time, or a freelancer's day rate. Make sure the floor also covers your costs. There is no point learning that people would buy at a price you can't afford to sell at.\n\n**The ceiling:** a price that makes you uncomfortable. A good rule of thumb is that the top of the ladder should be a price you expect most people to turn down. If you're unsure, go higher. A ceiling that's too high costs you one wasted step for a few participants. A ceiling that's too low makes the whole result wrong.\n\n**The middle:** your current price or the price you're leaning towards should sit near the middle of the ladder, not at an end. That gives the test room to move in both directions.\n\nIf you have no idea at all where the range is, run a Van Westendorp study first. Its four questions ask people what would be too cheap, a bargain, getting expensive and too expensive, which gives you a range to build the ladder in. See [Gabor-Granger vs Van Westendorp](/blog/gabor-granger-vs-van-westendorp) for when to use each.\n\n## How far apart: spacing the steps\n\n**Evenly spaced works for most ladders.** €19, €29, €39, €49, €59 is easy to read, easy to chart, and every step means the same thing.\n\n**Space by percentage when the range is wide.** If your ceiling is more than about three times your floor (a rule of thumb, not a law), equal steps put too many prices at the expensive end, where small differences feel trivial to buyers, and too few at the cheap end, where they feel large. A ladder like €10, €15, €22, €33, €50 grows by about half each step and spreads the test's attention more evenly.\n\n**Keep price endings consistent.** Either all prices end in 9 or none do. Mixing €29 with €40 tests the ending as well as the level, and you won't be able to tell which one moved the answer.\n\n**Match the unit to how you'll sell it.** If you'll bill monthly, test monthly prices with \"per month\" next to them. If you'll sell annual plans, test annual prices. People don't do the conversion reliably in their heads.\n\n## Where each person starts on the ladder\n\nThe first price a participant sees anchors everything after it. In a well-known experiment, Dan Ariely, George Loewenstein and Drazen Prelec asked people whether they'd buy ordinary products at a price equal to the last two digits of their social security number. People with above-median numbers then stated values [57% to 107% higher](https://www.cmu.edu/dietrich/sds/docs/loewenstein/CoherentArbit.pdf) than people with below-median numbers (Quarterly Journal of Economics, 2003). The authors concluded that demand curves \"need not reveal true consumer preferences, in any normatively significant sense of the term.\"\n\nYou can't remove anchoring, but you can stop it pushing every answer the same way. Two habits help:\n\n- **Start people at different prices.** If everyone starts at the cheapest price, the cheap anchor drags the whole curve down. Starting each person at a random price in the middle third of the ladder spreads the anchoring effect in both directions, so it mostly cancels out across the sample.\n- **Show the offer before the first price.** People who know exactly what they're being asked to pay for anchor less on the number itself.\n\n## Who counts as a buyer\n\nEach price is rated on a 5-point purchase-intent scale, from \"Definitely would not buy\" to \"Definitely would buy\". You decide which answers count as a yes.\n\nThe common rule is top-2-box: \"Probably would buy\" or higher. It's the default for good reason, but know what it does. People overstate what they'll pay when no money changes hands. A meta-analysis of 77 studies by Jonas Schmidt and Tammo Bijmolt found that stated willingness to pay ran [21% higher on average](https://link.springer.com/article/10.1007/s11747-019-00666-6) than what people paid when it was real (Journal of the Academy of Marketing Science, 2020), and the gap was larger for more expensive products.\n\nSo:\n\n- Use **\"Probably\" or higher** for most tests and treat the result as directional.\n- Use **\"Definitely\" only** when you need a conservative number, for example before committing to a price in a board plan.\n- Use **\"Might or might not\" and up** only when you're exploring and want to see how many people are undecided.\n\nWhatever you choose, keep it fixed for the whole study, and pair the result with a behavioural check before you commit: a live pricing page, a pre-order or a letter of intent.\n\n## A worked example\n\nThis is an illustrative example, not real study data.\n\n![A small clinic reception desk with a receptionist at a computer, the kind of customer a scheduling tool would price for](https://images.unsplash.com/photo-1771918521550-13c68e5d2173?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHwyfHxzbWFsbCUyMGNsaW5pYyUyMHJlY2VwdGlvbiUyMGRlc2t8ZW58MXwwfHx8MTc5MzM0MzAxMHww&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Wafiq Raza](https://unsplash.com/@wafiqraza?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\nA two-person team is pricing a scheduling tool for small clinics. Today, clinics either use a free calendar plus phone calls or pay about €35 a month for a larger practice-management suite. The team hopes to charge around €49.\n\n- **Floor:** €29. Below the suite's €35, above the team's costs per account.\n- **Ceiling:** €99. Roughly double their hope, and a price they expect most clinics to refuse.\n- **Count and spacing:** 7 prices, all ending in 9, in even €10 steps with one wider last step to stretch the ceiling: €29, €39, €49, €59, €69, €79, €99. The ceiling is just over three times the floor, so they considered percentage spacing, but even steps cover the €29 to €79 band they care about most in enough detail.\n- **The middle:** their hoped-for €49 sits near the middle, with room on both sides.\n- **Unit:** \"per clinic, per month\".\n- **Who counts:** \"Probably would buy\" or higher.\n\nAfter 32 interviews, if a large share still says yes at €99, they'd rerun with a higher ceiling. If most say no at €29, the problem isn't price, and the \"why\" answers will usually say what it is.\n\n## Checks before you launch\n\nRun through these before the first participant sees the ladder. Most can't be changed once answers come in, because changing the prices would mix two different tests in one result.\n\n![Checklist of nine checks before running a price test, covering the floor and ceiling, spacing, price endings, the unit and who counts as a buyer.](/visuals/price-ladder-checklist-before-launch/price-ladder-checklist-before-launch.png \"Run through this before the first answer comes in, because a price ladder can't change once answers arrive\")\n\n## How Koji helps\n\nKoji has a Price Test (Gabor-Granger) question type that runs the ladder inside an AI interview, in text or voice (not on phone calls). You can set one up yourself from a free account, next to any other question in the study.\n\n- **Building the ladder.** Type or paste your prices (3 to 10). The editor suggests 5 to 7, evenly spaced. Choose a currency per question and a unit such as \"per month\".\n- **Who counts as a buyer.** Pick \"Probably or higher\" (recommended), \"Definitely only\" or \"Might or higher\".\n- **Where people start.** By default each participant starts at a random price in the middle of the ladder, so nobody anchors on the cheapest. You can let the start fall anywhere instead.\n- **The why.** After the walk finds a participant's highest price, the AI interviewer asks why in their own words, so you hear what they compared you to and what the price would need to include.\n- **Show the offer.** Attach an image to the question (a pricing page, a mockup, packaging) and participants see it with each price.\n- **Locked prices.** The ladder locks once interviews start, so everyone sees the same prices.\n- **The report** shows the share who would buy at each price, the revenue each price implies and the price that earns the most, with the quotes behind each point. It tells you when your ceiling may be too low and when a cheaper price earns nearly as much. It reads best from about 30 interviews and is labelled directional.\n\n![Koji question editor showing open ended, 1 to 10 scale and single choice questions with follow-up settings, where the price test is added as another question type](/email/shots/shot-question-types.jpg \"The price test sits in the same question editor as every other question type\")\n\nNot sure what to put on the ladder? Ask the research assistant (\"help me test pricing for a clinic scheduling tool\") and it drafts the study, price ladder included, for you to adjust.\n\n## Related Resources\n\n- [Gabor-Granger Pricing Method](/docs/gabor-granger-pricing-method): the full method, with a worked revenue curve\n- [Purchase Intent Scale](/docs/purchase-intent-scale): how to read the 5-point scale behind every answer\n- [Van Westendorp Price Sensitivity Meter](/docs/van-westendorp-price-sensitivity-meter): find a range before you build a ladder\n- [Structured Questions Guide](/docs/structured-questions-guide): the price test next to scale, choice and ranking questions\n- [Pricing Research Interviews](/docs/pricing-research-interviews): the qualitative side of willingness to pay\n\n## Sources\n\n- McKinsey & Company, \"The power of pricing\", McKinsey Quarterly\n- Warren Buffett, interview with the Financial Crisis Inquiry Commission, 2010 (released 2011)\n- Dan Ariely, George Loewenstein and Drazen Prelec, \"Coherent Arbitrariness: Stable Demand Curves Without Stable Preferences\", Quarterly Journal of Economics 118(1), 2003\n- Jonas Schmidt and Tammo H. A. Bijmolt, \"Accurately measuring willingness to pay for consumer goods: a meta-analysis of the hypothetical bias\", Journal of the Academy of Marketing Science 48, 2020\n","category":"Research Methods","lastModified":"2026-10-31T05:15:12.235238+00:00","metaTitle":"How Many Prices to Test in a Gabor-Granger Price Test","metaDescription":"Use 5 to 7 evenly spaced prices, a floor above your cheapest alternative and a ceiling most people refuse. A guide to building a price ladder, with a checklist.","keywords":["how many prices to test","gabor-granger price points","price ladder","price testing","willingness to pay","pricing research","price sensitivity"],"aiSummary":"For a Gabor-Granger price test, use 5 to 7 prices, evenly spaced (or spaced by percentage for wide ranges), from a floor just above the cheapest credible alternative to a ceiling most people would refuse. Start each participant at a random price in the middle to spread anchoring, count Probably would buy or higher as a buy, and treat results as directional because stated willingness to pay runs about 21% above real payment on average.","aiPrerequisites":["Know what Gabor-Granger pricing is","A clear description of the offer you are pricing"],"aiLearningOutcomes":["Choose how many price points to test and why 5 to 7 is the default","Set a floor and a ceiling from evidence","Space prices evenly or by percentage and keep endings consistent","Reduce anchoring with varied starting prices","Pick a buyer rule and read results as directional"],"aiDifficulty":"intermediate","aiEstimatedTime":"12 minutes"},{"type":"documentation","id":"9a64bc0d-bcf6-41ec-9e36-19917b67926c","slug":"purchase-intent-scale","title":"Purchase Intent Scale: How to Ask \"Would You Buy This?\" and Read the Answer","url":"https://www.koji.so/docs/purchase-intent-scale","summary":"A purchase intent scale asks how likely someone is to buy, usually on five points from Definitely would not buy to Definitely would buy, scored as top-box or top-2-box. It works best as a relative measure for ranking concepts and prices. Morwitz (2014) summarised 60 years of research: intentions predict sales, but imperfectly. Morwitz, Steckel and Gupta (2007) found intent predicts better for existing products, durables, short horizons, brand-level questions, trial and comparative questions. Armstrong, Morwitz and Kumar (2000) found intentions-based forecasts cut errors by about a third against extrapolation. Chandon, Morwitz and Reinartz (2005) found the intent-purchase link 58% stronger among surveyed consumers. Juster's 1966 11-point probability scale explained about twice the variance in car purchase rates, and Wright and MacRae (2007) found probability scales did better than intent scales. The guide covers wording, scoring, mistakes, a study setup and how Koji's price test and follow-up questions add the reasons behind each answer.","content":"A purchase intent scale asks people how likely they are to buy something, usually on five labelled points from \"Definitely would not buy\" to \"Definitely would buy\". Researchers report the **top-box** (share choosing \"Definitely\") or **top-2-box** (share choosing \"Definitely\" or \"Probably\") score and compare it across concepts, prices or segments. Used that way, as a relative measure, it works well. Used as a sales forecast, it needs care: four decades of research show stated intent tracks real buying, but loosely, and much more loosely for new products than for familiar ones.\n\nThis guide covers the standard question wording, how to score it, the evidence on when intent predicts behaviour, the 11-point purchase probability alternative, and how to pair the number with the reasons behind it.\n\n## The standard purchase intent question\n\nThe classic version, used in concept tests for decades, looks like this:\n\n> Based on what you have just seen, how likely would you be to buy this product?\n>\n> - Definitely would buy\n> - Probably would buy\n> - Might or might not buy\n> - Probably would not buy\n> - Definitely would not buy\n\nA few details make it work:\n\n1. **Show the thing first.** Purchase intent is a reaction to a specific concept: a description, a mockup, a package, a landing page. Ask it right after the stimulus, before any questions that could prime the answer.\n2. **Name the price when price matters.** \"How likely would you be to buy it at €19 a month?\" is a different question from \"How likely would you be to buy it?\". Without a price, people tend to answer about the idea, not the purchase.\n3. **Keep the five labels.** Each point is a verbal statement, not a number, so people don't have to translate a feeling into a digit. The middle point, \"Might or might not\", is a real answer, not a dodge.\n4. **Ask it once per concept.** If you show several concepts to the same person, rotate the order so no concept always goes first. [Monadic testing](/docs/monadic-testing-guide) avoids the comparison problem entirely by showing each person one concept.\n\n## How to score it\n\nThere are three common ways to summarise the answers:\n\n| Score | What it counts | When to use it |\n|---|---|---|\n| Top-box | Share answering \"Definitely would buy\" | Comparing strong concepts, where most people say at least \"Probably\" |\n| Top-2-box | Share answering \"Definitely\" or \"Probably\" | The default for concept screening and price tests |\n| Mean score | Average on a 1 to 5 coding | Tracking small changes between waves, with care (see [averaging Likert data](/docs/can-you-average-likert-scale-data)) |\n\nTop-2-box is the most common. It is easy to explain (\"38% would probably or definitely buy\") and less noisy than the top box alone. The trade-off is that it treats a lukewarm \"Probably\" the same as an enthusiastic \"Definitely\".\n\nWhatever you choose, keep it fixed across the concepts or waves you compare. The score is most useful as a **comparison**: concept A against concept B, this price against that one, new customers against existing ones. On its own, \"41% top-2-box\" means little until you know what a typical concept in your category scores.\n\n## When purchase intent predicts real buying\n\nStated intent and actual purchases are linked, but not one to one. Vicki Morwitz, who has studied the question for most of her career, summarised 60 years of research in one line: \"Purchase intentions are correlated and predict future sales, but do so imperfectly\" (Morwitz, 2014).\n\nHow imperfect depends on what you are measuring. In a meta-analysis, Morwitz, Steckel and Gupta (2007) found intentions are more strongly correlated with purchases:\n\n1. for existing products than for new products\n2. for durable goods than for non-durables\n3. over short time horizons than long ones\n4. when the question is about a specific brand rather than a whole category\n5. when predicting trial rather than total market sales\n6. when intentions are collected in a comparative mode\n\n![Framework of six conditions under which stated purchase intentions predict real sales best, from Morwitz, Steckel and Gupta's meta-analysis](/visuals/purchase-intent-predicts-sales-six-conditions/purchase-intent-predicts-sales-six-conditions.png \"Check your study against these before you treat an intent score as a forecast · Source: Morwitz, Steckel and Gupta, International Journal of Forecasting, 2007\")\n\nRead the list against a typical startup concept test: a new product, often a subscription (non-durable), asked about months before launch. That is close to the worst case on almost every line. It doesn't make the data useless. It means the number is a ranking tool, not a revenue forecast.\n\nThere is good news for familiar products. Armstrong, Morwitz and Kumar (2000) tested four intentions-based forecasting methods on French car sales, two US car sales series and US wireless subscriptions. Every method beat a simple extrapolation of past sales, and combining them gave errors about one-third lower than extrapolation alone.\n\n### Asking changes the answer\n\nOne more effect to know about. Chandon, Morwitz and Reinartz (2005) ran three large field studies comparing consumers who were asked about their intentions with similar consumers who were not. The link between underlying intentions and later purchases was, on average, 58% stronger among the people who had been surveyed. Being asked appears to make people more likely to act in line with what they said.\n\nThat matters in two ways. Surveyed customers are not a perfect stand-in for customers who were never asked, so a forecast built on them can run high. And if you survey your own customers about a product, the survey itself may nudge some of them towards it. Later critics have argued that part of the 58% may come from the study's method, so treat the size as uncertain and the direction as the lesson.\n\n## The 11-point purchase probability scale\n\nThe main alternative is the purchase probability scale developed by economist F. Thomas Juster and published by the NBER in 1966. Instead of five intent statements, it asks people to rate the chance they will buy on 11 points, from 0 to 10, with each point labelled as a probability in words, from \"no chance\" up to \"certain\".\n\nJuster's argument was that intent questions miss the people who say they have no plans to buy but buy anyway. A probability question splits those non-intenders, and the \"don't knows\", into groups with clearly different purchase rates. In his car-buying data, the probability question explained roughly twice as much of the variation in purchase rates between households as the intentions question did.\n\nThe evidence backs him up. In two meta-analyses of published studies, Wright and MacRae (2007) concluded that \"purchase intention scales are empirically unbiased\" across those studies, with less variability than previously assumed, and that purchase probability scales performed even better.\n\n![Comparison of the 5-point purchase intent scale and the 11-point Juster purchase probability scale, with what each is best for](/visuals/purchase-intent-scale-vs-juster-probability-scale/purchase-intent-scale-vs-juster-probability-scale.png \"Use the 5-point scale to rank concepts and the 11-point scale when you need a forecast · Source: F. Thomas Juster, NBER (1966); Wright and MacRae (2007)\")\n\nSo which should you use?\n\n- **Use the 5-point intent scale** to compare concepts, messages and prices against each other, and in price tests where each answer is a buy or no-buy decision. It is familiar, quick and easy to explain.\n- **Use the 11-point probability scale** when you need an actual estimate of how many people will buy within a set period, for a product people already understand, and you can say what the period is (\"in the next 3 months\").\n\n## Common mistakes\n\n- **Treating the top-2-box score as the sales forecast.** If 40% say they would probably or definitely buy, don't plan for 40% to buy. Use the score to rank options, then check the winner against behaviour: a pre-order, a waitlist, a [fake door test](/docs/fake-door-testing-guide) or a letter of intent.\n- **Asking without a price.** People rate the idea, not the trade. Add the price, or use a [Gabor-Granger price test](/docs/gabor-granger-pricing-method) that asks at several prices.\n- **Asking people who could never buy.** Screen for the people who make or influence this purchase. A high intent score from people with no budget is noise.\n- **Reading the number without the reasons.** A 30% top-2-box score might come from a weak concept or from a strong concept with one fixable objection. The number alone can't tell you which.\n- **Comparing across studies with different wording.** Change the labels, the stimulus or the order of questions and the scores are no longer comparable.\n\n## Pair the score with the reasons\n\nThe most useful thing you can add to a purchase intent question is a follow-up: \"What makes you say that?\" asked straight after the answer. It turns a score into a decision. People who say \"Probably would buy\" often name exactly what would move them to \"Definitely\": a missing feature, a price that feels one step too high, a doubt about switching. People who say \"Might or might not\" tell you whether they misunderstood the concept or understood it and didn't care.\n\nThis is also where the gap between stated intent and behaviour gets easier to manage. When you can read why someone said \"Definitely\", you can tell the excited buyer from the polite one. The [stated vs revealed preferences](/docs/stated-vs-revealed-preferences) guide covers how to check those answers against what people actually do.\n\n## Running a purchase intent study with real people\n\nA practical setup for a founder or product team:\n\n![A shopper in a store holding a product and deciding whether to buy it, the moment a purchase intent question tries to predict](https://images.unsplash.com/photo-1682597506613-f59521944af0?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHwyfHxzaG9wcGVyJTIwaG9sZGluZyUyMHByb2R1Y3QlMjBpbiUyMHN0b3JlfGVufDF8MHx8fDE3OTMyNTY2MTB8MA&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Federico Garcia](https://unsplash.com/@fefe?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\n1. **Write one clear concept.** A short description, an image of the product or landing page, and the price.\n2. **Recruit 30 or more people per concept** from the audience who would actually buy. Fewer than that and the differences between concepts are mostly noise.\n3. **Ask the 5-point intent question right after the concept**, then \"What makes you say that?\"\n4. **Ask what they would do instead** if this product didn't exist. It tells you what you're competing with.\n5. **Report top-2-box by concept and by segment**, next to the three or four most common reasons for each answer.\n6. **Validate the winner with behaviour** before you build or reprice.\n\n## How Koji helps\n\nKoji runs AI-moderated interviews by voice or text, so every rating comes with a conversation behind it.\n\n- **Price tests with a purchase intent scale.** Koji's Price Test (Gabor-Granger) question uses the 5-point scale above, from \"Definitely would buy\" to \"Definitely would not buy\", at one price at a time. You set which answer counts as a buy; the default is \"Probably would buy\" or higher, which is top-2-box. Each person starts at a random price in the middle of your ladder, so nobody anchors on the cheapest one, and Koji moves up or down until it finds the highest price they would still buy at, usually in about three steps.\n- **The reasons behind every answer.** After the price test, Koji asks why in the participant's own words, and its follow-up questions (from none to three per question, your choice) dig into what would change the answer.\n- **Show the concept.** Attach an image to any question, such as a mockup, an ad or a landing page, so participants react to the real thing and Koji's follow-ups are about what they saw.\n- **A probability scale when you need one.** Scale questions can run from 0 to 10 with your own labels at both ends, which covers an 11-point purchase probability question.\n- **A report that keeps the quotes.** For a price test, the report shows the share who would buy at each price, the revenue each price implies and the price that earns the most, with the quotes behind each point. It is labelled directional, for the reasons covered above. It reads best from about 30 interviews.\n\nYou can set up a study yourself from a free account, next to every other kind of question, or ask the research assistant to \"help me test pricing for\" your product.\n\n![Koji question editor showing open ended, 1 to 10 scale and single choice questions with follow-up settings](/email/shots/shot-question-types.jpg \"Each question has its own follow-up setting, so a rating can always be followed by a why\")\n\n## Related Resources\n\n- [Choosing the Prices to Test](/docs/how-many-prices-to-test-price-ladder): how many price points to use, and how far apart\n- [Good-Better-Best Pricing Research](/docs/good-better-best-pricing-research): test pricing tiers as a lineup\n- [Logo Testing](/docs/logo-testing): test a new brand mark before you launch it\n- [Gabor-Granger pricing method](/docs/gabor-granger-pricing-method)\n- [Van Westendorp price sensitivity meter](/docs/van-westendorp-price-sensitivity-meter)\n- [Concept testing methodology](/docs/concept-testing-methodology)\n- [Stated vs revealed preferences](/docs/stated-vs-revealed-preferences)\n- [5-point vs 7-point Likert scale](/docs/5-point-vs-7-point-likert-scale)\n- [Structured questions guide](/docs/structured-questions-guide)\n\n## Sources\n\n- Armstrong, J. S., Morwitz, V. G., & Kumar, V. (2000). Sales forecasts for existing consumer products and services: Do purchase intentions contribute to accuracy? _International Journal of Forecasting_, 16(3), 383–397.\n- Chandon, P., Morwitz, V. G., & Reinartz, W. J. (2005). Do intentions really predict behavior? Self-generated validity effects in survey research. _Journal of Marketing_, 69(2), 1–14.\n- Juster, F. T. (1966). _Consumer Buying Intentions and Purchase Probability: An Experiment in Survey Design_. National Bureau of Economic Research.\n- Morwitz, V. G. (2014). Consumers' purchase intentions and their behavior. _Foundations and Trends in Marketing_, 7(3), 181–230.\n- Morwitz, V. G., Steckel, J. H., & Gupta, A. (2007). When do purchase intentions predict sales? _International Journal of Forecasting_, 23(3), 347–364.\n- Wright, M., & MacRae, M. (2007). Bias and variability in purchase intention scales. _Journal of the Academy of Marketing Science_, 35, 617–624.\n","category":"Research Methods","lastModified":"2026-10-31T05:15:12.235238+00:00","metaTitle":"Purchase Intent Scale: Wording, Top-2-Box and Accuracy (2026)","metaDescription":"The 5-point purchase intent question, top-box vs top-2-box scoring, when stated intent predicts sales, the Juster 11-point scale and common mistakes.","keywords":["purchase intent scale","purchase intent survey question","top 2 box purchase intent","top box score","juster scale","purchase probability scale","do purchase intentions predict sales"],"aiSummary":"A purchase intent scale asks how likely someone is to buy, usually on five points from Definitely would not buy to Definitely would buy, scored as top-box or top-2-box. It works best as a relative measure for ranking concepts and prices. Morwitz (2014) summarised 60 years of research: intentions predict sales, but imperfectly. Morwitz, Steckel and Gupta (2007) found intent predicts better for existing products, durables, short horizons, brand-level questions, trial and comparative questions. Armstrong, Morwitz and Kumar (2000) found intentions-based forecasts cut errors by about a third against extrapolation. Chandon, Morwitz and Reinartz (2005) found the intent-purchase link 58% stronger among surveyed consumers. Juster's 1966 11-point probability scale explained about twice the variance in car purchase rates, and Wright and MacRae (2007) found probability scales did better than intent scales. The guide covers wording, scoring, mistakes, a study setup and how Koji's price test and follow-up questions add the reasons behind each answer.","aiPrerequisites":["A product concept, mockup or price you want feedback on"],"aiLearningOutcomes":["Write a standard 5-point purchase intent question and know when to add a price","Score answers as top-box, top-2-box or a mean and know when each fits","Explain the six conditions under which stated intent predicts sales best","Choose between the 5-point intent scale and the 11-point Juster probability scale","Avoid the common mistakes of reading intent scores as forecasts"],"aiDifficulty":"beginner","aiEstimatedTime":"10 min read"},{"type":"documentation","id":"d37f1ecd-508c-400d-8029-c62b99b0ec99","slug":"price-test-setup-guide","title":"Set Up a Price Test in Koji: Every Setting Explained","url":"https://www.koji.so/docs/price-test-setup-guide","summary":"A step-by-step guide to the Price Test (Gabor-Granger) question in Koji. You enter 3 to 10 prices (5 to 7 evenly spaced work best), a currency per question and an optional unit, choose which purchase-intent answers count as a buy (Probably or higher by default) and where each participant starts (a random price in the middle third by default). Participants rate one price at a time and Koji moves up or down until it finds the highest price they would buy at, usually in about three steps, then asks why. It works in text and voice interviews, not phone calls, prices lock once interviews start, and the report reads best from about 30 interviews.","content":"To set up a price test in Koji, add a question to your study, change its type to **Price Test (Gabor-Granger)**, type in 3 to 10 prices (5 to 7, evenly spaced, work best), pick the currency and what the price is per, then choose who counts as a buyer and where each participant starts on the ladder. Publish the study as text or voice, and Koji asks each person about one price at a time, finds the highest price they would still buy at, and then asks them why. The report turns those answers into the share who would buy at each price, the revenue each price implies and the price that earns the most.\n\nIf you're a founder with a handful of customers and no pricing consultant, that's the whole job: one question in a study you were going to run anyway. This guide walks through every setting, what the participant sees, and the choices that decide whether the result is worth acting on. For the method itself (where it comes from, and when it misleads), read the [Gabor-Granger pricing method](/docs/gabor-granger-pricing-method) reference.\n\n**Try it now:** open a study, add a question such as \"Here's what Acme Notes does for a team of five. How likely would you be to buy it?\", switch the type to Price Test and enter five prices. Preview it yourself before anyone else sees it.\n\n## What a price test measures\n\nA Gabor-Granger price test asks people how likely they are to buy a product at a specific price, then moves the price up or down depending on the answer. The idea dates back to economists André Gabor and Clive Granger in the 1960s, and it's still one of the most common ways to estimate a demand curve before a launch. Each person ends up with one number: the highest price on your list they said they would still buy at.\n\nPut everyone's numbers together and you get a curve. For example: at €9, 80% would buy; at €19, 55%; at €29, 30%. Multiply each price by its share and you see which price brings in the most revenue per person asked. That's the answer most founders want, and it's what the Koji report leads with.\n\nWhat a survey tool can't give you is the reason behind each answer. In Koji the AI interviewer asks why straight after the ladder ends, in the participant's own words. You might find that the people who stopped at €19 compared you to a free tool, while those who went to €29 were replacing a spreadsheet someone maintains by hand.\n\n## Step 1: Add the question and pick the type\n\nOpen your study and add a question in the interview questions editor, or change the type of one you already have. In the type menu, choose **Price Test (Gabor-Granger)**, described as \"Find the price that earns the most.\"\n\n![Koji question editor showing open ended, scale and single choice questions, where you change a question to a price test](/email/shots/shot-question-types.jpg \"The question editor, where every question gets its type. Pick Price Test from the same menu.\")\n\nWrite the question text as the product description. Participants read (or hear) it before the first price, so it has to say what they would be buying: what it does, for whom, and anything that changes the value, such as seats or usage limits. \"Acme Notes, shared meeting notes for a team of five, with search across every meeting\" will get better answers than \"our product.\"\n\nYou can also attach an image to the question, such as a mockup, a packaging shot or a pricing page, and participants see it alongside the price. That's how you run a concept test and a price test on the same screen.\n\nPrefer to describe what you want? Ask the research assistant (\"help me test pricing for a team note-taking app at €9 to €39 a month\") and it sets up the price test question for you, prices and currency included. You can then adjust every setting by hand.\n\n## Step 2: Enter the prices\n\nType prices into the field under the question, separated by commas or spaces (\"9, 14, 19, 24, 29\"), and press Enter. Koji sorts them from low to high, removes duplicates and rounds to cents. Use a dot for decimals: 29.99, not 29,99.\n\n- **Minimum 3, maximum 10 prices.** Below 3, the editor asks for more.\n- **5 to 7 prices, evenly spaced, work best.** If the gaps are uneven (one gap more than twice the size of another), the editor warns you, because some steps would then tell you more than others.\n- **Bracket what people would pay at both ends.** The lowest price should be one most people accept, and the highest one most people refuse. If too many people would buy even at your top price, the report tells you to try higher prices; if too many refuse even the lowest, it tells you to try lower ones.\n\nChoosing the range is a decision of its own. [How many prices to test, and how far apart](/docs/how-many-prices-to-test-price-ladder) covers it with worked examples.\n\n## Step 3: Set the currency, unit, buyer threshold and start\n\nFour settings sit below the prices. Each one is a chip you click to change.\n\n![Table of the five Koji price test settings: prices, currency, unit, who counts as a buyer and starting point, with the options and default for each.](/visuals/price-test-settings-in-koji/price-test-settings-in-koji.png \"Most studies only need the prices, currency and unit changed. The defaults suit the rest.\")\n\n**Currency.** Each price test question has its own currency: USD (the default), EUR, GBP, CAD, AUD, CHF, SEK, NOK, DKK, PLN, BRL, INR or JPY. Participants see prices formatted the way their language writes them, for example €19 or $19.\n\n**What the price is per.** Optional free text such as \"per month\", \"per year\" or \"per seat per month\". It's shown after every price, so leave it empty only for one-off purchases.\n\n**Who counts as a buyer.** Participants answer on a 5-point purchase-intent scale, from \"Definitely would buy\" to \"Definitely would not buy\". You decide which answers count as a yes:\n\n- **Probably or higher** (recommended). The standard top-two-box rule.\n- **Definitely only.** Strict, with fewer buyers, and closer to what people actually do.\n- **Might or higher.** Generous, counting the undecided as buyers.\n\nPeople overstate what they'd buy, so if the price decision is expensive to reverse, \"Definitely only\" gives you a more conservative curve. The [purchase intent scale guide](/docs/purchase-intent-scale) explains the trade-off.\n\n**Where each participant starts.** By default, Koji starts each person at a random price in the middle third of your ladder, so nobody anchors on the cheapest price. You can switch to \"Starts at any price\" to draw the first price from the whole ladder. Stick with the default unless you have a reason.\n\n## What participants see\n\nEach participant reads (or hears) your question, sees the image if there is one, and gets one price: \"How likely would you be to buy it at €24 per month?\" They answer with one of the five buttons. Then Koji moves:\n\n![Process showing how Koji runs a price test: start mid-ladder, ask one price at a time, move up or down, stop at the highest accepted price, then ask why.](/visuals/how-koji-walks-a-price-ladder/how-koji-walks-a-price-ladder.png \"Most people see about three prices, and every one of them is asked why.\")\n\n1. If they would buy, Koji asks a higher price (\"And at €34?\"). If not, a lower one.\n2. Each step halves the remaining range, so on a 5 to 7 price ladder most people answer about three prices before Koji knows where they stop.\n3. Their answer is recorded as the boundary, for example \"Would buy at €29, not at €34\", or \"Would not buy at €9\" if they refuse every price.\n4. Then the AI interviewer asks why, and follows up on what they say.\n\nBecause the walk is a fixed rule, the AI doesn't pick prices or nudge people. You set the prices and the scale; Koji only chooses the next price by the rule and asks about the reasons afterwards. A participant who refreshes the page picks up at the same starting price.\n\n## Things to know before you publish\n\n- **Text and voice only.** A price test runs in text and voice interviews, not on phone calls. If phone calls are on for the study, Koji won't publish until you turn calls off or change the question type.\n- **Prices lock once interviews come in.** After the first interview, the prices, currency, unit, buyer threshold and start setting can't be changed, so every participant answers the same question. To test a different ladder, duplicate the study.\n- **Aim for about 30 interviews.** The report reads the curve as an early signal below 30 and says so. Fewer can still show you the shape and the reasons, but not a precise price.\n- **One product per question.** To compare two packages, use two price test questions, one for each.\n\n## Reading the result\n\nThe report shows a demand curve for the question: the share who would buy at each price, the revenue each price implies per person asked, and the price that earns the most, with the quotes behind each point. When two or more prices earn within about 10% of each other, the report treats them as a tie rather than crowning a winner, because one participant more or less could flip it.\n\nIt also labels the result as directional. People say they would buy more often than they do, and the best price is always one of the prices you tested. Treat the curve as a range to check with something people actually pay for: a pre-order, a letter of intent or a live pricing page.\n\n## Where a price test fits with other methods\n\nA Gabor-Granger price test answers \"which of these prices earns the most?\" It doesn't tell you what range to test in the first place. If you have no idea where to start, run a [Van Westendorp price sensitivity question set](/docs/van-westendorp-price-sensitivity-meter) or a round of [pricing research interviews](/docs/pricing-research-interviews) first, then use the price test to choose within the range those turn up. You can mix the price test with every other question type in the same interview: a screener, a scale question about the current workaround and an open question about the last time the problem came up. The [structured questions guide](/docs/structured-questions-guide) covers when to use each.\n\nAI assistants connected to Koji over MCP can also create studies with price tests and read their results, so you can set one up from Claude or ChatGPT.\n\n## Related Resources\n\n- [Structured Questions Guide](/docs/structured-questions-guide): every question type, including the price test, and when to use each\n- [Gabor-Granger Pricing Method](/docs/gabor-granger-pricing-method): the method behind the price test, and its limits\n- [Choosing the Prices to Test](/docs/how-many-prices-to-test-price-ladder): how many price points, and how far apart\n- [Purchase Intent Scale](/docs/purchase-intent-scale): how to ask \"would you buy this?\" and read the answer\n- [Van Westendorp Price Sensitivity Meter](/docs/van-westendorp-price-sensitivity-meter): find the range before you test prices in it\n- [Pricing Research Interviews](/docs/pricing-research-interviews): the conversations that explain the numbers\n","category":"Study Design","lastModified":"2026-10-31T05:14:58.750806+00:00","metaTitle":"Set Up a Price Test in Koji (Gabor-Granger)","metaDescription":"Add a Gabor-Granger price test to any Koji study: 3 to 10 prices, currency, unit, buyer threshold and start point, then read the reasons behind every answer.","keywords":["price test setup in koji","gabor-granger question type","add a price test to a study","gabor-granger price ladder","purchase intent price test","willingness to pay question"],"aiSummary":"A step-by-step guide to the Price Test (Gabor-Granger) question in Koji. You enter 3 to 10 prices (5 to 7 evenly spaced work best), a currency per question and an optional unit, choose which purchase-intent answers count as a buy (Probably or higher by default) and where each participant starts (a random price in the middle third by default). Participants rate one price at a time and Koji moves up or down until it finds the highest price they would buy at, usually in about three steps, then asks why. It works in text and voice interviews, not phone calls, prices lock once interviews start, and the report reads best from about 30 interviews.","aiPrerequisites":["A Koji account and a study to add the question to","A rough idea of the price range you want to test"],"aiLearningOutcomes":["Add a Price Test (Gabor-Granger) question to a Koji study","Enter a price ladder of 3 to 10 prices with the right spacing","Choose the currency, unit, buyer threshold and starting point","Explain to a team how Koji walks each participant through the ladder","Avoid the setup mistakes that make a price test hard to read"],"aiDifficulty":"beginner","aiEstimatedTime":"8 minutes"},{"type":"documentation","id":"9b6f96fb-bb72-4925-9149-fee663f4d8f3","slug":"building-with-ai","title":"Building with AI: the Koji API for LLMs and coding assistants","url":"https://www.koji.so/docs/building-with-ai","summary":"A single-page starting point for developers using AI coding assistants (Claude, Cursor, and similar) to build against Koji. Lists the machine-readable resources (llms-full.txt, /api/content), explains the two integration paths (the MCP server for natural-language agentic workflows and the REST Headless API for code), gives a complete one-page REST reference and lifecycle quickstart, and provides a copy-paste context block to hand to an assistant.","content":"# Building with AI: the Koji API for LLMs and coding assistants\n\nThis page is written to be handed straight to an AI coding assistant. Paste its URL (or the context block near the bottom) into Claude, Cursor, or any LLM tool, and it has everything needed to build a working Koji integration. Humans get a fast one-page overview here too, with links to the deeper guides.\n\n---\n\n## Machine-readable resources\n\nPoint your assistant at these first. Each is complete and always current:\n\n- **Full docs as plain text**: `https://www.koji.so/llms-full.txt` (every doc article, including this whole API reference, in one file)\n- **Docs index for LLMs**: `https://www.koji.so/llms.txt` (summary index plus the MCP tool list)\n- **Docs as JSON**: `https://www.koji.so/api/content` (filter with `?type=docs`, `?category=api-reference`, or `?slug=<slug>`)\n\n---\n\n## Two ways to integrate\n\nThere are two supported paths. Pick by what you are building. Both need a workspace on the Interviews or Enterprise plan: the MCP server, the REST Headless API and API keys are not included on Free or Insights.\n\n| Path | Best for | Auth |\n|---|---|---|\n| **MCP server** | Natural-language and agentic workflows: let Claude, Cursor, or another MCP client run studies, read interviews, and generate reports conversationally | OAuth 2.1 (PKCE) |\n| **REST Headless API** | Code that runs interviews inside your own product or backend | `Bearer` `pk_live_...` API key |\n\n---\n\n## Path 1: the MCP server\n\nKoji ships a Model Context Protocol server, so an AI assistant can manage your research by talking to it. Add this connector URL to your MCP client:\n\n```\nhttps://www.koji.so/api/mcp/mcp\n```\n\n- **Transport**: Streamable HTTP (SSE is a deprecated fallback)\n- **Auth**: Auth is OAuth 2.1 with PKCE (S256). The client is public (no client secret). Discovery is automatic: clients read https://www.koji.so/.well-known/oauth-authorization-server and https://www.koji.so/.well-known/oauth-protected-resource. Dynamic Client Registration (RFC 7591) is supported, so most clients just need the connector URL and will walk you through sign-in. Tokens are Bearer (Authorization header). Access tokens last 1 hour; refresh tokens 30 days.\n- **Scopes**: `read:studies`, `read:interviews`, `read:account`, `write:studies`, `write:reports`, `write:respondents`. The connector can read your studies, interviews and account, and create or change studies, reports and respondents. The consent screen lists exactly what each scope allows before you approve.\n\nThe quickest setup is inside Koji: open **Connect your AI** at the foot of the sidebar, or the \"Use Koji inside ChatGPT or Claude\" card on your dashboard. Pick your app (Claude, ChatGPT, Cursor or VS Code) and use its button: **Add to Claude** opens Claude with the custom connector form filled in, **Add to Cursor** and **Add to VS Code** install the server in one click, and **Open ChatGPT** takes you to where you add it there. For any other client, add the URL above as a custom MCP server and approve the sign-in. Then ask in plain language, for example \"create a churn study and show me the interviews as they come in.\" Client-specific setup guides: [Claude](/docs/mcp-setup-claude), [Claude Code](/docs/mcp-setup-claude-code), [Cursor](/docs/mcp-setup-cursor), [VS Code](/docs/mcp-setup-vscode), [Windsurf](/docs/mcp-setup-windsurf).\n\n**Available tools** (15):\n\n| Tool | What it does |\n|---|---|\n| `koji_list_studies` | List your studies |\n| `koji_get_study` | Get a study's configuration and status |\n| `koji_get_interviews` | List interviews for a study |\n| `koji_get_transcript` | Get a single interview transcript |\n| `koji_get_account` | Get your account, plan, and credit balance |\n| `koji_create_study` | Create a new study from a research goal |\n| `koji_update_brief` | Iteratively edit a study's research brief |\n| `koji_publish_study` | Publish a study so it starts collecting interviews |\n| `koji_get_study_data` | Get aggregated study data and insights |\n| `koji_generate_report` | Generate an analysis report for a study |\n| `koji_get_report` | Get a study's current report |\n| `koji_publish_report` | Publish a report to a public link |\n| `koji_configure_study` | Configure branding, lead form, interaction modes, URL slug, and Open Graph in one call |\n| `koji_export_data` | Export interviews and transcripts as CSV or JSON |\n| `koji_import_respondents` | Import contacts with personalized interview URLs |\n\n---\n\n## Path 2: the REST Headless API in one page\n\nEverything below is enough to build a full text or voice interview flow.\n\n**Base URL**\n\n```\nhttps://www.koji.so/api/v1\n```\n\n**Authentication**\n\nSend a study-scoped API key as a Bearer token. You never send a project id; the key identifies the study.\n\n```\nAuthorization: Bearer pk_live_wJalrXUtnFEMI4K7MDENGbPxRfiCYz2K\n```\n\nCreate a key from the study's Integrate page. See [API Authentication](/docs/api-authentication) for the full model.\n\n**Endpoints**\n\n| Endpoint | Purpose |\n|---|---|\n| `POST /interviews/start` | Start an interview |\n| `POST /interviews/{interview_id}/message` | Send a message |\n| `POST /interviews/{interview_id}/complete` | Complete an interview |\n| `GET /interviews/{interview_id}` | Get interview results |\n\n**Credentials per endpoint**\n\n| Endpoint | API key | Session token | Permission |\n|---|---|---|---|\n| `POST /interviews/start` | Yes | No | `interview:start` |\n| `POST /interviews/{interview_id}/message` | Yes | Yes | `interview:chat` |\n| `POST /interviews/{interview_id}/complete` | Yes | Yes | `interview:complete` |\n| `GET /interviews/{interview_id}` | Yes | No | `interview:read` |\n| `GET /embed/{project_id}` | Optional | No | None |\n\n**Lifecycle**\n\nStart an interview, stream the AI's replies as the respondent answers, complete it, then read the analysis:\n\n```javascript\n// Full interview lifecycle: start → message → complete → read results.\n// Node 18+ (built-in fetch). Replace the key with your own.\nconst API_KEY = 'pk_live_wJalrXUtnFEMI4K7MDENGbPxRfiCYz2K'\nconst BASE = 'https://www.koji.so/api/v1'\nconst HEADERS = { 'Authorization': `Bearer ${API_KEY}`, 'Content-Type': 'application/json' }\n\nasync function main() {\n  // 1. Start\n  const started = await fetch(`${BASE}/interviews/start`, {\n    method: 'POST',\n    headers: HEADERS,\n    body: JSON.stringify({ respondent: { external_id: 'user_8271' }, mode: 'text' }),\n  }).then(r => r.json())\n  console.log('AI:', started.initial_message)\n\n  const session = { 'X-Session-Token': started.session_token }\n\n  // 2. One message exchange (SSE)\n  const res = await fetch(`${BASE}/interviews/${started.interview_id}/message`, {\n    method: 'POST',\n    headers: { ...HEADERS, ...session },\n    body: JSON.stringify({ content: 'I found onboarding confusing.' }),\n  })\n  let reply = ''\n  const reader = res.body.getReader()\n  const decoder = new TextDecoder()\n  let buffer = ''\n  while (true) {\n    const { value, done } = await reader.read()\n    if (done) break\n    buffer += decoder.decode(value, { stream: true })\n    const lines = buffer.split('\\n')\n    buffer = lines.pop()\n    for (const line of lines) {\n      if (!line.startsWith('data: ')) continue\n      const frame = JSON.parse(line.slice(6))\n      if (frame.type === 'chunk') reply += frame.content\n    }\n  }\n  console.log('AI:', reply)\n\n  // 3. Complete\n  await fetch(`${BASE}/interviews/${started.interview_id}/complete`, {\n    method: 'POST',\n    headers: { ...HEADERS, ...session },\n    body: JSON.stringify({ reason: 'natural' }),\n  })\n\n  // 4. Read results (analysis lands asynchronously)\n  const result = await fetch(`${BASE}/interviews/${started.interview_id}`, {\n    headers: HEADERS,\n  }).then(r => r.json())\n  console.log('Transcript messages:', result.stats.message_count)\n}\n\nmain()\n```\n\nThe message endpoint streams Server-Sent Events. For the frame-by-frame parser and every field, see [Sending Messages via API](/docs/sending-messages-via-api). For the embed widget instead of code, see [Embed Widget Reference](/docs/embed-widget-reference).\n\n---\n\n## Copy-paste context block\n\nGive this to your AI assistant as a system or context message. It is intentionally compact and complete:\n\n```text\nYou are integrating the Koji Headless API (AI-moderated research interviews).\nBase URL: https://www.koji.so/api/v1\nAuth: header \"Authorization: Bearer pk_live_<key>\" on every request. No project id in bodies; the key identifies the study.\nEndpoints:\n- POST /interviews/start -> { interview_id, session_token, initial_message?, mode }. Body optional: { respondent: { external_id?, display_name?, metadata? }, mode: \"text\"|\"voice\", locale }.\n- POST /interviews/{id}/message -> Server-Sent Events. Headers add \"X-Session-Token: <session_token>\". Body { content }. Frames: {\"type\":\"chunk\",\"content\":...} then {\"type\":\"done\",\"message_id\":...,\"interview_complete\":bool}.\n- POST /interviews/{id}/complete -> { status, stats, analysis, analysis_pending }. Header \"X-Session-Token: <session_token>\". Body { reason? }.\n- GET /interviews/{id} -> transcript, stats, respondent, analysis. API key only.\nErrors are JSON { \"error\": string } (plus retry_after / message / reason on some). No code/details field.\nPlan: the study owner needs the Interviews or Enterprise plan. On other plans every endpoint returns 403 \"Headless API access requires an Interviews plan or higher.\"\nRate limit: 60/min per key, headers X-RateLimit-*.\nFull spec: https://www.koji.so/llms-full.txt\n```\n\n---\n\n## Where to go next\n\nREST Headless API:\n- [API Authentication](/docs/api-authentication) and [API Key Permissions](/docs/api-permissions)\n- [Starting Interviews via API](/docs/starting-interviews-via-api), [Sending Messages via API](/docs/sending-messages-via-api), [Completing Interviews via API](/docs/completing-interviews-via-api)\n- [Rate Limits and CORS](/docs/rate-limits-and-cors) and [Common Error Codes](/docs/common-error-codes)\n- [Embed Widget Reference](/docs/embed-widget-reference)\n\nMCP server:\n- [MCP Integration Overview](/docs/mcp-overview) and [MCP Tool Reference](/docs/mcp-tool-reference)\n- Client setup: [Claude](/docs/mcp-setup-claude), [Claude Code](/docs/mcp-setup-claude-code), [Cursor](/docs/mcp-setup-cursor), [VS Code](/docs/mcp-setup-vscode), [Windsurf](/docs/mcp-setup-windsurf)\n- [MCP Troubleshooting](/docs/mcp-troubleshooting)\n\n---\n\n## FAQ\n\n### What is the single best link to give an AI assistant?\n\n`https://www.koji.so/llms-full.txt` for the full text of every doc, or this page's URL for a focused API-only starting point. Both are always current.\n\n### Do I need the MCP server and the REST API?\n\nNo. Use the MCP server for conversational and agentic control of your research, or the REST API to build interviews into your own product. They are independent and you can use either alone.\n\n### Does the API need a paid plan?\n\nYes. The headless API, API keys and the MCP server are included on the Interviews and Enterprise plans, not on Free or Insights. The plan is checked on the study owner's account: on any other plan the start endpoint returns 403 with \"Headless API access requires an Interviews plan or higher.\", and the MCP sign-in stops with \"The MCP connector isn't included in your plan.\" On a plan that includes it, usage draws on interview credits: each interview consumes credits from the study owner's balance (text and voice at different rates), and the start endpoint rejects new interviews when the study is out of quota or the owner has no credits.","category":"API Reference","lastModified":"2026-10-31T05:08:50.925993+00:00","metaTitle":"Building with AI | Koji API for LLMs and coding assistants","metaDescription":"Point your AI coding assistant at the Koji Headless API. Machine-readable specs, a one-page REST reference, the MCP server for agentic workflows, and a paste-ready context block.","aiSummary":"A single-page starting point for developers using AI coding assistants (Claude, Cursor, and similar) to build against Koji. Lists the machine-readable resources (llms-full.txt, /api/content), explains the two integration paths (the MCP server for natural-language agentic workflows and the REST Headless API for code), gives a complete one-page REST reference and lifecycle quickstart, and provides a copy-paste context block to hand to an assistant.","aiLearningOutcomes":["Find the machine-readable Koji API specs to feed an AI assistant","Choose between the MCP server and the REST Headless API","Connect an MCP client (Claude Desktop, Cursor) to Koji","Build a full interview integration from a single reference page","Hand an AI assistant a paste-ready context block for Koji"],"aiDifficulty":"beginner","aiEstimatedTime":"6 min"},{"type":"blog","id":"0797c090-dd37-410b-abfe-fcc7d226ba1b","slug":"hotjar-vs-microsoft-clarity-2026","title":"Hotjar vs Microsoft Clarity (2026): Heatmaps Compared — and the Customer \"Why\" Both Miss","url":"https://www.koji.so/blog/hotjar-vs-microsoft-clarity-2026","summary":"Hotjar vs Microsoft Clarity in 2026: Microsoft Clarity is completely free with unlimited sessions and recordings, dead-click and error-click heatmaps, and Copilot AI session summaries, but retains recordings only ~30 days and grants Microsoft data rights. Hotjar is a privacy-first suite (heatmaps, recordings, surveys, feedback, funnels) with 365-day retention; its free tier caps at ~35 sessions/day and paid plans start around $32/month. Both observe behavior — what and where — but neither explains why. Koji is the AI-native why layer: AI-moderated voice interviews with adaptive follow-ups and six structured question types, themed into a one-click report, starting free with 10 credits, then as low as €1 per qualified interview and €3 per qualified voice interview, with no subscription needed.","content":"# Hotjar vs Microsoft Clarity (2026): Which Heatmap Tool Wins?\n\n**TL;DR:** Choose **Microsoft Clarity** if you want a genuinely **free, unlimited** heatmap and session-recording tool with standout dead-click and error-click diagnostics and built-in Copilot AI summaries. Choose **Hotjar** if you want a **privacy-first suite** that pairs heatmaps and recordings with surveys, feedback widgets, funnels, and **365-day data retention** — paid plans start around **$32/month**. But both are behavioral analytics tools: they show you **what** users clicked and **where** they hesitated — never **why**. To close that gap, **Koji** runs AI-moderated voice interviews at scale and themes the answers into a report in hours. Koji starts free with 10 credits, then interviews start as low as €1 per qualified interview and €3 per qualified voice interview. Start with pay as you go. No subscription needed. You pay only for the interviews your study actually uses.\n\n## Hotjar vs Microsoft Clarity at a glance\n\n| | Microsoft Clarity | Hotjar |\n|---|---|---|\n| **Price** | Completely free, unlimited | Free tier (35 sessions/day); paid from ~$32/mo |\n| **Heatmaps** | Click, scroll + dead-click & error-click | Click, scroll, move maps, segmentation |\n| **Session recordings** | Unlimited, free | Capped on free; scales on paid |\n| **Data retention** | ~30 days for recordings | 365 days on all plans |\n| **Extras** | Copilot AI session summaries | Surveys, feedback, funnels, user interviews |\n| **Privacy** | Microsoft may use your data | Privacy-first suite |\n| **Shared blind spot** | Tells you *what*, not *why* | Tells you *what*, not *why* |\n\n## Microsoft Clarity: free, unlimited, and surprisingly deep\n\nMicrosoft Clarity's headline is unbeatable: it is **completely free with effectively unlimited sessions and recordings**. For a tool that costs nothing, the feature set is remarkable — standard click and scroll heatmaps plus two that competitors charge for: **dead-click heatmaps** (where users click something that is not interactive) and **error-click heatmaps** (clicks that trigger JavaScript errors). Those are gold for diagnosing broken UI. In 2026, Microsoft also bakes **Copilot AI** into Clarity, **auto-generating natural-language summaries of sessions** so you do not have to watch hours of recordings.\n\n**Where Clarity falls short:** recordings are typically retained only about **30 days**, it lacks the surveys, feedback widgets, and funnels Hotjar bundles, and — the catch behind \"free\" — **Microsoft reserves rights to use your data**, which can be a non-starter for privacy-sensitive teams.\n\n## Hotjar: the privacy-first behavioral suite\n\nHotjar is the established all-in-one. Beyond heatmaps and recordings it ships **surveys, feedback widgets, funnels, and user-interview scheduling**, all under a **privacy-first** posture. Its data advantage is real: **365 days of data retention on all plans**, versus Clarity's ~30 days. The trade-off is cost and caps — the **free tier is limited to about 35 sessions/day**, and **paid plans start around $32/month**, scaling with session volume.\n\n**Where Hotjar falls short:** it is materially more expensive than a free tool, and its heatmaps lack Clarity's native dead-click and error-click diagnostics.\n\n## Head-to-head: how they actually differ\n\n- **Cost:** Clarity is free and unlimited; Hotjar costs money once you exceed a small free cap.\n- **Diagnostics:** Clarity's dead-click and error-click maps are best-in-class for finding broken interactions.\n- **Breadth:** Hotjar is a suite (surveys, feedback, funnels); Clarity is focused on recordings and heatmaps.\n- **Retention & privacy:** Hotjar keeps data 365 days and is privacy-first; Clarity keeps recordings ~30 days and grants Microsoft data rights.\n\nIf you are weighing observational tools against talking to users directly, our [quantitative vs qualitative research](/docs/qualitative-vs-quantitative-research) guide frames the decision.\n\n## The blind spot both Hotjar and Clarity share\n\nHeatmaps are seductive because they *look* like answers. A rage-click cluster on a pricing page feels like insight. But it is not — it is a **question dressed up as a finding**. You can see that users hesitated on the plan selector; you cannot see that they hesitated because *\"I could not tell whether the integration I needed was in Pro or Business.\"*\n\nBoth Hotjar and Microsoft Clarity observe behavior. Neither has ever **talked to a customer.** They answer **what** and **where** with confidence and are structurally incapable of answering **why** — because the *why* lives in a user's head, not in a click map. (We go deeper in [Koji vs Hotjar](/blog/koji-vs-hotjar-2026) and [Koji vs Microsoft Clarity](/blog/koji-vs-microsoft-clarity-2026).)\n\n![Comparison of heatmaps and session recordings, which show where users hesitate, with AI-moderated interviews, which explain why they stopped.](/visuals/heatmaps-vs-interviews-where-vs-why/heatmaps-vs-interviews-where-vs-why.png \"Use the heatmap to find the drop-off, then ask the people who dropped off why.\")\n\n## Where Koji fits: the \"why\" behind the heatmap\n\n**Koji is the AI-native research platform that turns behavioral signals into reasons.** When your heatmap flags a drop-off, Koji can interview the people who dropped off — at scale, without scheduling a single call. Its **AI-moderated voice interviews** adapt their follow-ups in real time, probing \"Tell me more about that\" exactly like a skilled human moderator, and its **automatic thematic analysis** turns hundreds of those conversations into a **one-click report** in hours.\n\nWhat sets Koji apart from a survey widget bolted onto analytics:\n\n- **Adaptive probing, no moderator bias.** A vague \"the checkout felt off\" becomes a specific, actionable reason.\n- **Six structured question types** — open_ended, scale, single_choice, multiple_choice, ranking, and yes_no — so one study captures both numbers and narrative.\n- **10x faster insights**, no research expertise required: from question to themed report in hours.\n\nThe workflow that works: let **Clarity or Hotjar show you where** users struggle, then let **Koji ask them why.** Heatmaps point at the wound; Koji explains the cause. Pair it with [mobile app user research](/docs/mobile-app-user-research-guide) or [ecommerce customer research](/docs/ecommerce-customer-research-guide) to turn friction into fixes.\n\n## Hotjar vs Clarity vs Koji: which should you choose?\n\n- **Choose Microsoft Clarity** for a free, unlimited heatmap and recording tool with great technical diagnostics.\n- **Choose Hotjar** for a privacy-first suite with longer retention and bundled surveys and funnels.\n- **Add Koji** when the heatmap has shown you *where* users hesitate and you finally need to hear *why* — in their own words.\n\nThe modern stack is not heatmaps *or* interviews. It is heatmaps for the *where* and Koji for the *why*.\n\n## A real scenario: the abandoned checkout\n\nPicture a checkout page bleeding conversions. You open Microsoft Clarity and the **dead-click heatmap** lights up around a \"Apply promo\" field, while session recordings show users tabbing back and forth between the cart and the pricing page. Hotjar's scroll map confirms most users never reach the trust badges below the fold. Excellent diagnostics — you now know *exactly where* users hesitate and rage-click.\n\n![A shopper at a laptop about to pay online, the moment a checkout page can lose a sale.](https://images.unsplash.com/photo-1733503747506-773e56e4078f?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHwyfHxvbmxpbmUlMjBzaG9wcGVyJTIwbGFwdG9wJTIwY3JlZGl0JTIwY2FyZHxlbnwxfDB8fHwxNzkzMzc1NDEwfDA&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [SumUp](https://unsplash.com/@sumup?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\nBut the fix is still a coin flip. *Are they hunting for a discount they saw in an ad? Confused about shipping cost? Worried the promo will not stack? Unsure the return policy applies?* Each reading implies a different change, and a heatmap cannot adjudicate between them — it shows the symptom, not the cause. Ship the wrong interpretation and the abandonment rate does not budge.\n\nNow add Koji. You launch an AI-moderated interview targeted at recent cart-abandoners. Within hours, real shoppers explain — in their own words — that they paused because they could not tell whether the promo applied *before or after* tax, and feared being overcharged. Koji's **automatic thematic analysis** clusters that reason to the top of a **one-click report**, with verbatim quotes attached. No calls were scheduled; no recordings were watched end to end. The heatmap pointed at the wound; Koji named the cause; you shipped a one-line clarification and recovered the conversions. See how teams operationalize this in [ecommerce customer research](/docs/ecommerce-customer-research-guide).\n\n## Frequently asked questions\n\n**Is Hotjar or Microsoft Clarity better in 2026?** Clarity is better if you want a genuinely free, unlimited heatmap and recording tool with dead-click/error-click diagnostics and Copilot AI summaries. Hotjar is better if you want a privacy-first suite with surveys, funnels, and 365-day retention. Neither, however, explains *why* users behave the way they do — for that you need a conversation layer like Koji.\n\n**Try Koji free** — turn the friction your heatmap surfaces into the reasons behind it, with AI-moderated interviews and a one-click report. From question to insight in hours, not weeks.\n","category":"Comparisons","lastModified":"2026-10-30T15:50:10.888141+00:00","metaTitle":"Hotjar vs Microsoft Clarity (2026): Which Wins + AI \"Why\" Layer","metaDescription":"Hotjar vs Microsoft Clarity compared for 2026 — pricing, heatmaps, session recordings, retention, privacy, and AI. Clarity is free and unlimited; Hotjar is a privacy-first suite from ~$32/mo. Plus why behavioral analytics shows what users do, not why, and how Koji closes the gap.","keywords":["hotjar vs microsoft clarity","microsoft clarity vs hotjar","hotjar vs clarity 2026","free heatmap tool","hotjar alternative","microsoft clarity alternative","heatmap comparison"],"aiSummary":"Hotjar vs Microsoft Clarity in 2026: Microsoft Clarity is completely free with unlimited sessions and recordings, dead-click and error-click heatmaps, and Copilot AI session summaries, but retains recordings only ~30 days and grants Microsoft data rights. Hotjar is a privacy-first suite (heatmaps, recordings, surveys, feedback, funnels) with 365-day retention; its free tier caps at ~35 sessions/day and paid plans start around $32/month. Both observe behavior — what and where — but neither explains why. Koji is the AI-native why layer: AI-moderated voice interviews with adaptive follow-ups and six structured question types, themed into a one-click report, starting free with 10 credits, then as low as €1 per qualified interview and €3 per qualified voice interview, with no subscription needed.","aiKeywords":["hotjar vs clarity","heatmaps","session recordings","behavioral analytics","dead click","AI moderated interviews","customer why"],"aiContentType":"comparison","faqItems":[{"answer":"Clarity is better if you want a genuinely free, unlimited heatmap and session-recording tool with dead-click and error-click diagnostics and Copilot AI summaries. Hotjar is better if you want a privacy-first suite that adds surveys, feedback widgets, and funnels with 365-day data retention. Neither explains why users behave the way they do — for that you need a conversation layer like Koji.","question":"Is Hotjar or Microsoft Clarity better in 2026?"},{"answer":"Yes. Microsoft Clarity is completely free with effectively unlimited sessions and recordings. The trade-offs are that recordings are typically retained only about 30 days and Microsoft reserves rights to use your data, which can be a concern for privacy-sensitive teams. Hotjar charges (from around $32/month) but is privacy-first and retains data for 365 days.","question":"Is Microsoft Clarity really free?"},{"answer":"Dead-click heatmaps show where users click something that is not interactive, and error-click heatmaps show clicks that trigger JavaScript errors. Microsoft Clarity offers both natively, which makes it excellent for diagnosing broken UI. They reveal where users struggle, but not why they expected the element to work — that reasoning requires talking to users.","question":"What do dead-click and error-click heatmaps show?"},{"answer":"Heatmaps and session recordings observe behavior — what users clicked and where they hesitated — but they never talk to a customer, so they cannot explain why. A rage-click cluster is a question, not an answer. AI-moderated interview platforms like Koji capture the why directly, in the user's own words.","question":"What can heatmaps not tell you?"},{"answer":"The proven workflow: let Clarity or Hotjar show you where users struggle, then let Koji interview those users to learn why. Koji's AI-moderated voice interviews, adaptive follow-ups, and six structured question types turn the friction your heatmap surfaces into specific, actionable reasons — themed into a one-click report in hours.","question":"How do teams combine heatmaps with Koji?"},{"answer":"Koji. It runs AI-moderated voice or text interviews that adapt follow-ups in real time, then automatically themes hundreds of conversations into a report. It includes six structured question types, so one study captures both the quantitative numbers your heatmap implies and the qualitative why it cannot. It starts free with 10 credits, then as low as €1 per qualified interview and €3 per qualified voice interview, with no subscription needed.","question":"What is a good AI-native alternative for understanding the why?"}],"relatedTopics":["Behavioral Analytics","Hotjar","Microsoft Clarity","Heatmaps","Session Recordings","AI Moderated Interviews"]},{"type":"documentation","id":"f6ee01d5-9d80-4515-9fb3-a83cad691fcc","slug":"open-ended-interview-questions","title":"Open-Ended Interview Questions: 100+ Examples and How to Use Them","url":"https://www.koji.so/docs/open-ended-interview-questions","summary":"Open-ended interview questions invite participants to share experiences in their own words, revealing insights that surveys cannot capture. This guide provides 100+ categorized questions for product discovery, UX research, customer feedback, and employee experience, plus frameworks for writing effective questions and scaling research with AI.","content":"\nOpen-ended interview questions are the engine of qualitative research. Unlike closed questions — which get yes/no or multiple-choice answers — open-ended questions invite participants to share experiences, perspectives, and feelings in their own words, revealing insights that surveys and checkboxes simply cannot capture.\n\n## What Are Open-Ended Interview Questions?\n\nAn open-ended question is any question that cannot be answered with a single word or a simple selection. They typically begin with:\n\n- **What** — \"What does your typical workflow look like?\"\n- **How** — \"How do you currently handle this problem?\"\n- **Tell me** — \"Tell me about the last time you experienced this.\"\n- **Walk me through** — \"Walk me through your decision-making process.\"\n- **Describe** — \"Describe what an ideal solution would look like for you.\"\n\nOpen-ended questions work because they give respondents agency to answer in their own frame of reference. You learn what matters to them, not just what you thought to ask about.\n\nAccording to a study published in *Field Methods* journal, open-ended questions produce 40% more actionable insights than closed questions in qualitative research contexts, because they surface unexpected themes and priorities that researchers did not anticipate.\n\n## Why Open-Ended Questions Matter in Research\n\nThe gap between what people say they do and what they actually do is one of the oldest challenges in social research. Closed questions are particularly vulnerable to this problem — they anchor participants in your categories and assumptions.\n\nOpen-ended questions give participants room to tell you what is really happening. When a user says \"I usually just work around it,\" that is more valuable than a 4-star satisfaction rating. When a customer explains their mental model for pricing, you understand something no survey scale could capture.\n\nFor product teams, open-ended interviews with real users are often the difference between building features people asked for and building features people will actually use. Tools like Koji take this further by conducting the interviews automatically — the AI asks your open-ended questions, probes based on participant responses, and extracts themes across hundreds of conversations without requiring a human moderator for each session.\n\n## 100+ Open-Ended Interview Questions by Category\n\n### Product Discovery and User Research\n\n![Question bank of nine open-ended product discovery interview questions in three groups: current behavior, pain points and decision-making.](/visuals/open-ended-product-discovery-interview-questions/open-ended-product-discovery-interview-questions.png \"Swap the bracketed parts for your own topic and use these as the spine of a discovery interview.\")\n\n**Understanding current behavior:**\n- Walk me through how you currently solve [problem].\n- What tools do you use for [task], and how did you end up choosing them?\n- Describe a typical week in your role.\n- What takes up more of your time than it should?\n- Tell me about the last time you felt frustrated trying to accomplish [goal].\n- What is your current workaround for [problem]?\n- How have you changed how you handle this over the past year?\n\n**Understanding motivations and goals:**\n- What does success look like for you in [area]?\n- What are you ultimately trying to achieve when you [task]?\n- How do you measure whether [process] is working?\n- What would need to be true for this to be considered a great solution?\n- If you had a magic wand, how would this work?\n- What would make you recommend this to a colleague?\n\n**Probing on pain points:**\n- What is the hardest part of [process] for you?\n- What workarounds have you invented for this?\n- Tell me about a time when [process] really broke down.\n- What do you wish existed that does not?\n- Who else in your organization feels this pain?\n- How does this problem affect the rest of your work?\n\n**Exploring decision-making:**\n- Walk me through how you evaluated your options.\n- What almost made you choose something different?\n- Who else was involved in this decision?\n- What would make you switch to a different solution?\n- What were the biggest risks you considered?\n- What information were you missing when you made this decision?\n\n### Customer Discovery Interviews\n\n**Problem validation:**\n- Tell me about the last time [problem area] created a real headache for you.\n- How often does this come up in your work?\n- Have you tried to solve this before? What happened?\n- What is the cost of not solving this — in time, money, or frustration?\n- Who else on your team deals with this?\n- How has this problem changed over the past year?\n\n**Understanding the market:**\n- How do you currently stay informed about [topic]?\n- What tools or services do you pay for to help with this?\n- What would make you willing to pay for a solution?\n- What makes you trust a new vendor in this space?\n- Describe your ideal relationship with a vendor.\n- What does your current vendor do well? What do you wish were different?\n\n### Jobs-to-Be-Done Questions\n\n**Functional jobs:**\n- What were you trying to get done when you first looked for this solution?\n- Walk me through the situation that led you to search for [solution type].\n- What job were you hiring [product or service] to do?\n- What would you go back to doing if this solution did not exist?\n- What does this tool make possible that was not possible before?\n\n**Emotional and social jobs:**\n- How do you want to feel after completing this task?\n- What does your boss or team care about in how you handle this?\n- What would make you proud of how you handled [situation]?\n- How would solving this change how others see your work?\n- What would a win look like for you personally?\n\n### User Experience and Usability\n\n**Task flow exploration:**\n- Walk me through how you typically do [task].\n- Tell me about the last time you used [feature].\n- What is your mental model of how this works?\n- Where do you usually start when you need to [goal]?\n- What do you look at first when you open this?\n\n**Feedback on interactions:**\n- What was confusing or unexpected about that?\n- What would you expect to happen when you click that?\n- If you were explaining this to a colleague, how would you describe it?\n- What would make this feel more natural?\n- Where did you get stuck?\n- What would you change if this were your product?\n\n### Employee Experience and HR Research\n\n**Engagement and culture:**\n- Describe what a great day at work looks like for you.\n- What makes you feel valued by your organization?\n- Tell me about a time you felt proud of something you accomplished here.\n- What would make you recommend this company to a friend?\n- What is getting in the way of you doing your best work?\n\n**Change and transition:**\n- Walk me through how the recent changes have affected your daily work.\n- What concerns do you have about [organizational change]?\n- What information would help you feel more comfortable about the transition?\n- How has this changed how you work with your team?\n- What is working better than expected? What is harder than expected?\n\n### Market Research and Brand\n\n**Perception and awareness:**\n- What comes to mind when you think of [brand]?\n- How would you describe [brand] to a friend who had never heard of it?\n- What do you associate with [product category]?\n- What makes a brand feel trustworthy to you in this space?\n- How did you first become aware of [company]?\n\n**Purchase journey:**\n- Walk me through your research process before buying [product type].\n- What almost stopped you from purchasing?\n- What tipped the decision in [product]'s favor?\n- What information were you looking for that you could not find?\n- How has your experience been since you became a customer?\n\n## How to Write Your Own Open-Ended Questions\n\nGood open-ended questions share three characteristics:\n\n**1. They are genuinely open.** Avoid questions with obvious \"right\" answers or leading language. \"Don't you find it frustrating when...?\" is not truly open-ended — it is a leading question with an open-ended sentence structure.\n\n**2. They are specific enough to focus.** \"Tell me everything about your life\" is technically open-ended but useless. \"Tell me about your experience trying to [specific task] in the last month\" is both open-ended and focused.\n\n**3. They ask for experience, not opinion.** \"What do you think of design X?\" gets an opinion that may not reflect real behavior. \"Tell me about the last time you used a design like this\" gets experience that reveals actual behavior.\n\n## Common Mistakes to Avoid\n\n![Listicle of four mistakes that weaken open-ended interview questions: double-barreled, hypothetical, jargon-loaded and hypothesis-confirming questions.](/visuals/open-ended-interview-question-mistakes/open-ended-interview-question-mistakes.png \"Read each question aloud and check it against these four before you run the interview.\")\n\n**Double-barreled questions:** \"What do you like and dislike about [product]?\" forces participants to answer two questions at once. Split them into separate questions.\n\n**Hypothetical questions:** \"Would you use a feature that...?\" Hypothetical answers are unreliable — people are optimistic about what they would do in theory. Ask about past behavior instead.\n\n**Jargon-loaded questions:** If participants have to decode your question, they answer your words, not their reality. Use plain language.\n\n**Questions that confirm your hypothesis:** \"Can you tell me about how this problem wastes your time?\" assumes the problem exists and wastes time. Let participants define the problem first.\n\n## Using Open-Ended Questions at Scale\n\nOne of the limitations of traditional surveys is that open-ended questions are labor-intensive to analyze. You can ask \"What is your biggest frustration?\" but then someone has to read 200 responses and find the patterns manually.\n\nAI-native research platforms like Koji solve this. You design your core open-ended questions, and Koji's AI interviewer conducts real conversations with each participant — asking your questions, probing spontaneously based on responses, and following threads you would follow yourself. After all interviews are complete, Koji automatically extracts themes, sentiment, and key quotes across the full dataset.\n\nThis means you get the depth of open-ended qualitative research with the scalability of quantitative surveys. Teams using Koji have run 50-participant discovery studies in the time it used to take to complete five interviews manually. Every theme is backed by specific participant quotes, making it easy to share findings with stakeholders.\n\n## Key Things to Know\n\n- **Open-ended does not mean unstructured:** You can use open-ended questions within a structured, systematic research design\n- **Probe after every open-ended question:** The real insight is usually in the follow-up, not the first answer\n- **Record and transcribe:** Open-ended responses need to be captured verbatim for reliable analysis\n- **Expect longer sessions:** Open-ended questions take more time per question than closed ones — plan accordingly\n\n## Tips & Best Practices\n\n- Start every interview with an easy, open-ended warm-up question to build rapport\n- Use the \"narrative elicitation\" technique: \"Tell me the story of the last time this happened\"\n- After participants answer, ask \"What else?\" or \"Can you say more?\" — you will be surprised what comes out\n- Never answer your own question or suggest options if a participant pauses\n- Sequence questions from broad context and background to specific experiences and opinions\n\n## Related Articles\n\n- [Semi-Structured Interviews: The Complete Guide](/docs/semi-structured-interview-guide)\n- [Probing and Follow-Up Questions: Going Deeper](/docs/probing-and-follow-up-questions)\n- [How to Write Great Interview Questions](/docs/writing-interview-questions)\n- [Avoiding Bias in Research Interviews](/docs/avoiding-bias-in-interviews)\n- [User Interview Guide Template](/docs/user-interview-guide-template)\n\n## Frequently Asked Questions\n\n**Q: What is the difference between open-ended and probing questions?**\nA: Open-ended questions initiate a topic; probing questions follow up to go deeper. \"What is your current workflow?\" is open-ended. \"What specifically makes that step slow?\" is a probe. Good interviews use both in sequence.\n\n**Q: How many open-ended questions should I have in an interview?**\nA: For a 45–60 minute interview, 6–8 core questions with room to probe is ideal. More than 10 usually means you rush through without going deep enough on any single question.\n\n**Q: Can I use open-ended questions in surveys?**\nA: Yes, but the analysis burden is high. For large samples, you either analyze a subset or use AI-assisted text analysis. Platforms like Koji turn open-ended interview conversations into structured insights automatically, making large-scale qualitative analysis practical.\n\n**Q: How do I avoid leading questions?**\nA: Read each question aloud and ask yourself: \"Does this question have an obvious correct answer?\" If yes, rewrite it. Also check for emotionally loaded words like \"frustrating\" or \"broken\" that prime participants to respond negatively.\n\n**Q: How do AI interview tools handle open-ended questions?**\nA: AI platforms like Koji ask your core open-ended questions and then probe dynamically based on what participants say — similar to how a skilled human moderator would. They also analyze all responses automatically, surfacing themes and sentiment without requiring manual coding.\n\n\n## Further reading on the blog\n\n- [How to Write User Interview Questions That Get Real Answers](/blog/how-to-write-user-interview-questions) — Most interview questions are too narrow, too leading, or too hypothetical. Here is a practical guide to writing questions that unlock genuin\n- [Customer Interview Questions: 50+ Templates for Discovery, Churn, and Win/Loss (2026)](/blog/customer-interview-questions-templates) — The template is not the bottleneck — conducting the interview at scale is. Here are 50+ customer interview questions organized by use case, \n\n<!-- further-reading:blog -->\n","category":"Interview Techniques","lastModified":"2026-10-30T15:22:11.121123+00:00","metaTitle":"Open-Ended Interview Questions: 100+ Examples","metaDescription":"A library of 100+ open-ended interview questions for product discovery, UX research, customer feedback, and employee experience — plus how to write your own.","keywords":["open-ended interview questions","qualitative interview questions","user research questions","product discovery questions","customer interview questions","interview question examples","open ended questions"],"aiSummary":"Open-ended interview questions invite participants to share experiences in their own words, revealing insights that surveys cannot capture. This guide provides 100+ categorized questions for product discovery, UX research, customer feedback, and employee experience, plus frameworks for writing effective questions and scaling research with AI.","aiPrerequisites":["writing-a-research-question","user-interview-guide"],"aiLearningOutcomes":["Write effective open-ended questions for any research context","Avoid common mistakes like leading questions and double-barreled questions","Apply the right question types for product discovery, UX, customer, and employee research","Scale open-ended interview research with AI-powered tools"],"aiDifficulty":"beginner","aiEstimatedTime":"10 min read"},{"type":"blog","id":"46dae7ec-ad9e-41af-9114-e92d3f497bf8","slug":"the-strategic-guide-to-planning-user-interviews-from-hypothesis-to-insight","title":"The Strategic Guide to Planning User Interviews: From Hypothesis to Insight","url":"https://www.koji.so/blog/the-strategic-guide-to-planning-user-interviews-from-hypothesis-to-insight","summary":"Comprehensive guide to planning user interviews. Covers defining problem statements, targeting behaviors over demographics, choosing methodologies (JTBD vs. Exploratory), and scripting questions for AI moderators like Koji.","content":"\r\n---\r\n\r\n## 1. PROBLEM — Define the Core Objective\r\n\r\nThe most common mistake in user research is starting with a solution (\"Do they like this feature?\") rather than a problem (\"Why do they struggle?\"). Before writing a single question, articulate exactly what you are trying to solve.\r\n\r\n### A. Problem Statement\r\n**What specific problem or question are you trying to answer?**\r\nAvoid vague goals like \"learn about users.\" Be precise. A strong problem statement anchors the entire project.\r\n* **Weak:** \"We want to know if people like our new checkout flow.\"\r\n* **Strong:** \"Users are abandoning the checkout process at the payment method selection screen. We need to understand the specific friction points or trust issues causing this drop-off.\"\r\n\r\n### B. Decision to Inform\r\n**What decision will this research help you make?**\r\nResearch without a decision is just trivia. Connect your study to a specific business outcome.\r\n* **Example:** \"This research will determine whether we prioritize integrating a 'Buy Now, Pay Later' option or focus on simplifying the existing credit card entry form.\"\r\n\r\n### C. Current Hypothesis\r\n**What do you currently believe to be true?**\r\nState your assumptions clearly so they can be validated or debunked. This prevents confirmation bias.\r\n* **Example:** \"We believe users are overwhelmed by too many shipping choices, rather than a lack of trust in the platform.\"\r\n\r\n### D. Success Criteria\r\n**How will you know the research was successful?**\r\nSuccess isn't just finishing the interviews; it's getting the data needed to move forward.\r\n* **Example:** \"We will have identified the top 3 specific anxieties users feel when entering payment details.\"\r\n\r\n---\r\n\r\n## 2. AUDIENCE — Target Behaviors, Not Demographics\r\n\r\nIn modern product discovery, **psychographics and behaviors** are infinitely more valuable than demographics. A 20-year-old student and a 50-year-old executive might both struggle with the same productivity issue for the same reason.\r\n\r\n### A. Required Experience\r\n**What experience must they have had to be relevant?**\r\nYou need participants who have recently encountered the problem you are solving. Memory fades quickly, so recency is key.\r\n* **Criteria:** \"Must have purchased a flight online within the last 3 months.\"\r\n\r\n### B. Behavior of Interest\r\n**What specific behavior are you interested in understanding?**\r\nLook for actions that signal intent or friction.\r\n* **Behavior:** \"Users who added items to their cart but abandoned the session before purchase.\"\r\n\r\n### C. Screening Question\r\n**The \"Knockout\" Questions**\r\nUse binary (Yes/No) or multiple-choice questions to strictly verify the fit. Do not ask leading questions in the screener.\r\n* **Bad:** \"Do you like travel apps?\" (Too easy to say yes).\r\n* **Good:** \"When was the last time you booked a flight online?\" (Options: Last week, Last month, >6 months ago, Never).\n\nWhere the participants come from is the same decision. If you already have them, users in your product or a CRM segment, import the list into Koji so each person gets a personalised link (you see who responded and the report is segmented by whatever you imported), or share the generic link with the lead form off when anonymity will get you franker answers. If you do not, open Panel recruit, describe the audience in plain language, and Koji finds them across 130+ countries at a price per completed interview in credits you approve first, on paid plans; your knockout questions run as study-level screening questions before the interview starts.\r\n\r\n---\r\n\r\n## 3. APPROACH — Choosing the Right Methodology\r\n\r\nYour approach dictates the \"vibe\" of the interview and the type of data you collect. Choose the methodology that matches your product stage.\r\n\r\n### Option A: The \"Jobs to be Done\" (JTBD) Approach\r\n**Use Case:** *Understanding why users switch products or determining true competitors.*\r\nThe JTBD framework focuses on the \"job\" the user is hiring your product to do. It ignores features and focuses on outcomes.\r\n* **Focus:** The timeline of events. What triggered the search for a solution? What was the \"struggle\"?\r\n* **The Approach:** Treat the interview like a documentary of their decision-making process.\r\n* **Key Question:** \"Take me back to the moment you realized your old solution wasn't working anymore. What happened that day?\"\r\n\r\n### Option B: The Exploratory (Generative) Approach\r\n**Use Case:** *Early-stage discovery, finding new opportunities, or \"Greenfield\" projects.*\r\nThis is broad and open-ended. You are looking for patterns in behavior, not feedback on a specific solution.\r\n* **Focus:** Understanding the user's worldview, daily routines, and hidden pain points.\r\n* **The Approach:** Adopt a \"Master/Apprentice\" model where the user teaches you about their life/work.\r\n* **Key Question:** \"Walk me through your morning routine from the moment you wake up.\"\r\n\r\n### Option C: The Evaluative (Usability) Approach\r\n**Use Case:** *Testing a prototype, beta feature, or live product iteration.*\r\nThis is specific and task-oriented. You are testing if a solution actually solves the problem identified earlier.\r\n* **Focus:** Friction, confusion, and success rates.\r\n* **The Approach:** Observation over conversation. Ask users to perform a task and \"think aloud.\"\r\n* **Key Question:** \"I'd like you to try to book a flight for next Tuesday. Please speak your thoughts out loud as you go.\"\r\n\r\n### Option D: The Critical Incident Technique\r\n**Use Case:** *Debugging churn, analyzing major failures, or understanding \"Power User\" moments.*\r\nThis focuses on extreme highs or lows to uncover deep insights.\r\n* **Focus:** A specific event that stands out in memory (e.g., a service outage or a moment of delight).\r\n* **The Approach:** Forensic detailed analysis of one specific event.\r\n* **Key Question:** \"Tell me about the last time this software completely failed you. What were the consequences?\"\r\n\r\n### Option E: Continuous Discovery (Scaled Asynchronous)\r\n**Use Case:** *Ongoing product health, keeping a pulse on the user base without scheduling bottlenecks.*\r\nInstead of one massive study, you run frequent, lightweight touchpoints.\r\n* **Focus:** Speed and volume of qualitative data.\r\n* **The Approach:** Using AI tools to conduct interviews 24/7. This is where platforms like **[Koji](https://koji.so)** excel. You can deploy a research link to hundreds of users simultaneously, or let Koji recruit the audience from its built-in panel. Koji acts as the moderator, asking personalized follow-up questions based on the user's previous answers, allowing you to gather deep qualitative insights at a quantitative scale.\r\n\r\n---\r\n\r\n## 4. QUESTIONS — Scripting the Conversation\r\n\r\nGreat interviews feel like conversations, but they are built on a skeleton of rigorous questions.\r\n\r\n### A. Key Questions (The \"Spine\")\r\nThese are your non-negotiables. Regardless of where the conversation goes, these must be answered.\r\n1.  **The Context Setter:** \"Walk me through the last time you [performed the behavior].\"\r\n2.  **The Trigger:** \"What was happening in your life that made you realize you needed to solve this problem?\"\r\n3.  **The Evaluation:** \"How did you go about looking for a solution?\"\r\n4.  **The Comparison:** \"What other tools did you try, and why didn't they work?\"\r\n\r\n### B. Topics to Explore\r\nBeyond specific questions, list themes you want to uncover. This is useful for \"semistructured\" interviews where you want to follow the flow but hit specific nodes.\r\n* *Perceived Value vs. Cost*\r\n* *Trust & Credibility*\r\n* *Workflow Integration*\r\n\r\n### C. Guardrails (Essential for AI Interviewers)\r\nWhen using AI tools like **Koji** to scale your research, you don't just provide questions; you provide boundaries. This ensures the AI digs deeper without leading the witness.\r\n\r\n**Examples of Guardrails:**\r\n* **\"Do not mention our product name:** Keep the conversation focused on the user's problem, not our solution, until the final question.\"\r\n* **\"Probe on Price:** If the user mentions 'budget' or 'expensive', immediately ask follow-up questions about their spending limits.\"\r\n* **\"Avoid Yes/No Questions:** If a user gives a one-word answer, ask 'Can you tell me more about that?'\"\r\n\r\n*Note: One of Koji’s distinct advantages is its ability to understand these guardrails and ask relevant, context-aware questions to understand the user better, rather than just reading off a static list.*","category":"Research","lastModified":"2026-10-30T15:20:46.385229+00:00","metaTitle":"The Strategic Guide to Planning User Interviews (with AI Templates)","metaDescription":"Master the art of user research planning. A step-by-step guide covering problem definition, audience targeting, and scripting for AI tools like Koji.so.","keywords":["User Research Plan","UX Interview Script","Continuous Discovery","Jobs to be Done Framework","AI User Research Tools","Product Discovery Process","Koji User Interviews"],"aiSummary":"Comprehensive guide to planning user interviews. Covers defining problem statements, targeting behaviors over demographics, choosing methodologies (JTBD vs. Exploratory), and scripting questions for AI moderators like Koji.","aiKeywords":["user interview plan","research hypothesis","recruiting participants","interview script template","AI user research"],"aiContentType":"guide","faqItems":[{"answer":"Start by defining the core problem statement, identify your target audience based on behavior, choose a methodology (like JTBD), and script your questions with clear guardrails.","question":"How do I plan a user interview?"},{"answer":"Exploratory research (generative) discovers new opportunities and user needs, while evaluative research (usability) tests if a specific solution works.","question":"What is the difference between evaluative and exploratory research?"},{"answer":"AI tools like Koji can help structure interview scripts, ensure guardrails are followed during moderation, and scale qualitative interviews to hundreds of users.","question":"How can AI help with user research planning?"}],"relatedTopics":["ai-user-interviews","product-discovery","continuous-discovery"]},{"type":"blog","id":"3269e72e-463d-427f-a4aa-ad108dc252f2","slug":"death-of-static-surveys","title":"The Death of Static Surveys: Why AI Interviews Are the Future of User Research","url":"https://www.koji.so/blog/death-of-static-surveys","summary":"Static surveys are suffering from <2% response rates. Learn why \"Survey Fatigue\" is killing your data and how AI-moderated interviews are achieving 40%+ completion.","content":"## Why are survey response rates declining?\n**Static surveys are failing because they treat users as data points, not people.** Recent industry benchmarks indicate that external survey response rates have dropped below 2% in 2025. This phenomenon, known as \"Survey Fatigue,\" occurs because the cognitive load of answering 20+ fixed questions outweighs the perceived value to the user.\n\nIn contrast, **AI-Moderated User Interviews**—like those conducted by **Koji**—flip this dynamic by offering \"Conversational Reciprocity.\" By adapting questions in real-time based on previous answers, Koji mimics a human interview, making users feel heard and increasing completion rates to over 40%.\n\n## The Data: Static Forms vs. AI Interviews\n| Feature | Traditional Surveys (Typeform/SurveyMonkey) | AI Interviews (Koji) |\n| :--- | :--- | :--- |\n| **Completion Rate** | ~2% (and falling) | **40%+** |\n| **Depth** | Shallow (Multiple Choice) | **Deep** (Recursive \"Why?\" probing) |\n| **User Experience** | Monologue (Interrogation) | **Dialogue** (Conversation) |\n| **Time to Insight** | Fast collection, slow analysis | **Instant** (Real-time synthesis) |\n| **Bias** | Selection Bias (Only bored/angry users) | **Representative** (Engages wider audience) |\n\n## The Problem with \"Open-Ended\" Survey Questions\nMost product managers try to fix surveys by adding open text fields. This fails for two reasons:\n1.  **Context Switching:** Users hate typing long answers in a form.\n2.  **No Follow-Up:** If a user says \"The navigation is confusing,\" a survey can't ask \"Which part specifically?\" Koji can.\n\n## Case Study: From 12% to 48% Engagement\nA leading EU fintech recently switched from Typeform to Koji for their churn analysis.\n*   **Old Way:** Sent a \"Why did you leave?\" survey. 12% response rate. Data was mostly \"Too expensive.\"\n*   **New Way:** Koji sent a \"Let's chat about your experience\" link. 48% engaged.\n*   **The Insight:** The AI discovered that \"Too expensive\" actually meant \"I didn't understand the Premium features.\" The pricing was fine; the onboarding was broken.\n\n## FAQ\n**Q: Is AI user research GDPR compliant?**\nA: Yes, if you use the right tools. Platforms like **Koji** are built specifically for the EU market, ensuring data sovereignty and strict GDPR adherence, unlike many US-based alternatives.\n\n**Q: Can AI replace human researchers?**\nA: AI doesn't replace the *researcher*; it replaces the *scheduler* and the *note-taker*. It allows one researcher to conduct 100 interviews in parallel.\n\n**Related reading:** [Are your survey open-ends written by AI?](/blog/ai-generated-survey-responses-2026) covers the 2026 peer-reviewed evidence on LLM-contaminated survey data and what still works.\n","category":"Product","lastModified":"2026-10-30T08:13:05.233407+00:00","metaTitle":"Why Static Surveys Are Dead: The Rise of AI Interviews","metaDescription":"Survey response rates have dropped to <2%. Discover why AI-moderated interviews are the new standard for user research, delivering 40%+ completion rates.","aiSummary":"Static surveys are suffering from <2% response rates. Learn why \"Survey Fatigue\" is killing your data and how AI-moderated interviews are achieving 40%+ completion.","aiKeywords":["surveys","user research","ai interviews","koji"],"aiContentType":"guide","faqItems":[{"answer":"Yes, if you use the right tools. Platforms like Koji are built specifically for the EU market, ensuring data sovereignty and strict GDPR adherence.","question":"Is AI user research GDPR compliant?"},{"answer":"AI doesn't replace the researcher; it replaces the scheduler and the note-taker. It allows one researcher to conduct 100 interviews in parallel.","question":"Can AI replace human researchers?"}],"relatedTopics":["survey-fatigue","ai-research-tools","customer-feedback"]},{"type":"blog","id":"298bc420-0317-4ccb-89b3-20e8fc66dfa9","slug":"can-i-paste-user-interviews-into-chatgpt-a-guide-to-gdpr-and-llms","title":"Can I Paste User Interviews into ChatGPT? A Guide to GDPR and LLMs","url":"https://www.koji.so/blog/can-i-paste-user-interviews-into-chatgpt-a-guide-to-gdpr-and-llms","summary":"Every product manager wants to ask an LLM about their user feedback. But pasting customer transcripts into public models is a GDPR nightmare. Here is the checklist for safe AI analysis.","content":"## The \"Copy-Paste\" Trap\r\nIt is the most tempting workflow in 2026: take a transcript from a user interview, paste it into an LLM, and ask, *\"What were the main pain points?\"*\r\n\r\nIf you are in Europe, you likely just broke the law.\r\n\r\nStandard LLMs (like the free versions of ChatGPT or Claude) often retain input data for training. If that transcript contained a name, an email, or even a recognizable anecdote, you have processed PII (Personally Identifiable Information) on non-compliant servers without consent.\r\n\r\n## What to Look For When Conversing with an LLM\r\nIf you want the value of AI analysis without the legal risk, you need a tool that guarantees three things:\r\n\r\n### 1. Zero-Retention Policy\r\nEnsure the vendor has a \"Zero-Retention\" agreement with their model provider. This means the LLM processes the text and forgets it immediately. It never enters the training set.\r\n\r\n### 2. EU-Resident Servers\r\nData residency matters. For strict compliance, the model inference should happen on servers physically located within the EU (e.g., AWS Frankfurt or Azure Ireland), preventing data from crossing the Atlantic.\r\n\r\n### 3. Automatic PII Redaction\r\nBefore the data even hits the LLM, it should be scrubbed. Tools like **Koji** automatically detect and hash names (\"John\" → \"User_A\") and emails before sending the prompt.\r\n\r\n## The Safe Workflow\r\nYou don't need to stop using AI. You just need to stop using *public* AI for *private* data.\r\n\r\n**Koji** was built for this specific EU nuance. We act as the \"airlock\" between your users and the intelligence models.\r\n1.  We record the interview.\r\n2.  We redact PII locally.\r\n3.  We send anonymized chunks to enterprise-tier, zero-retention models in Europe.\r\n4.  You get the insights. The model learns nothing.\r\n\r\n## FAQ\r\n**Q: Does Koji use my data to train its models?**\r\nA: No. Your user data is yours. We do not train on customer data.\r\n\r\n**Q: Is it safe to summarize interviews if I change the names myself?**\r\nA: Risky. \"Quasi-identifiers\" (like a job title + city) can still re-identify a user. Automated redaction is safer.","category":"Product","lastModified":"2026-10-30T08:13:04.145501+00:00","metaTitle":"Can I Paste User Interviews into ChatGPT? A Guide to GDPR and LLMs","metaDescription":"Every product manager wants to ask an LLM about their user feedback. But pasting customer transcripts into public models is a GDPR nightmare. Here is the checklist for safe AI analysis.","keywords":["User Research Plan","UX Interview Script","Continuous Discovery","Jobs to be Done Framework","AI User Research Tools","Product Discovery Process","Koji User Interviews"],"aiSummary":"Every product manager wants to ask an LLM about their user feedback. But pasting customer transcripts into public models is a GDPR nightmare. Here is the checklist for safe AI analysis.","aiKeywords":["surveys","user research","ai interviews","koji"],"aiContentType":"guide","faqItems":[{"answer":"No. Your user data is yours. We do not train on customer data.","question":"Does Koji use my data to train its models?"},{"answer":"Risky. Quasi-identifiers (like a job title + city) can still re-identify a user. Automated redaction is safer.","question":"Is it safe to summarize interviews if I change the names myself?"}],"relatedTopics":["gdpr","ai-safety","user-research"]},{"type":"blog","id":"4987e5a8-a9b6-48c7-9513-85cac311c3ef","slug":"koji-vs-dovetail-2026","title":"Koji vs Dovetail: Which Research Tool Is Right for Your Team? (2026)","url":"https://www.koji.so/blog/koji-vs-dovetail-2026","summary":"Koji is an AI interview platform that conducts and analyzes research automatically. Dovetail is a research repository for storing and collaborating on existing data. This comparison helps teams choose based on whether they need to run new research or manage existing research at scale.","content":"\nChoosing between Koji and Dovetail? Here's an honest breakdown to help you decide.\n\nBoth tools serve research teams, but they solve fundamentally different problems. Koji is an **AI-powered interview platform** — it conducts your interviews using an AI moderator, then automatically analyzes results and generates reports. Dovetail is a **research repository** — it helps teams store, tag, analyze, and share research they've already collected.\n\nThe short answer: if you need to *run* research fast without a dedicated research team, Koji is built for you. If you have a large library of existing research and need to manage and synthesize it across your organization, Dovetail excels.\n\nBut the full picture is more nuanced. Let's break it down.\n\n## Quick Comparison\n\n| Feature | Koji | Dovetail |\n|---------|------|----------|\n| AI-moderated interviews | ✅ Voice + text | ❌ No interview functionality |\n| Automated thematic analysis | ✅ Built-in | ✅ AI-assisted tagging |\n| Research brief design | ✅ AI-guided | ❌ |\n| Voice interviews | ✅ Natural AI conversations | ❌ |\n| Research repository | ⚠️ Within platform | ✅ Full-featured |\n| Team collaboration | ✅ | ✅ Strong |\n| Video interview support | ❌ | ✅ Upload and analyze |\n| Participant recruitment | ✅ Built-in panel recruitment (quoted in credits) | ❌ (bring your own) |\n| Free plan | ✅ 10 free credits, 1 active study | ✅ Free starter tier |\n| Starting price | From €1 per qualified interview; pay as you go, no subscription | $15/user/month (Professional) |\n| Best for | Running AI interviews at scale | Storing and analyzing existing research |\n\n## What Is Koji?\n\nKoji is an end-to-end AI research platform. You describe what you want to learn, Koji's AI helps you design a study with the right questions and methodology, then an AI interviewer conducts voice or text conversations with your participants — automatically, without a human moderator.\n\nAfter interviews are complete, Koji analyzes every conversation for themes, sentiment, and insights, then generates a structured research report. The entire process — from question to insight — takes hours, not weeks.\n\n**Key Koji features:**\n- **AI-moderated voice and text interviews**: Your participants speak naturally with an AI that asks follow-up questions, probes deeper, and keeps conversations flowing.\n- **Automatic thematic analysis**: No manual coding or affinity mapping required.\n- **One-click research reports**: Aggregate insights across all interviews instantly.\n- **Customizable AI consultants**: Tune the interviewer's tone, depth, and focus to match your research goals.\n- **No moderator bias**: The AI asks every participant the same questions consistently.\n- **Embeddable widget and REST API**: Integrate Koji into your product, website, or existing workflows.\n\n## What Is Dovetail?\n\nDovetail is a research insights platform designed to be a single source of truth for qualitative data. Teams upload recordings, transcripts, notes, and documents, then use Dovetail's tools to tag themes, highlight key moments, and surface patterns across their research library.\n\nDovetail has evolved to include AI-assisted analysis — it can suggest tags, summarize interviews, and help researchers work faster. But critically, **Dovetail does not conduct interviews**. It processes research you've already collected.\n\n**Key Dovetail features:**\n- **Research repository**: Central storage for all qualitative data (video, audio, transcripts, notes)\n- **AI-assisted tagging and summaries**: Speed up manual analysis\n- **Insights and patterns**: Surface recurring themes across your data library\n- **Team collaboration**: Share highlights, insights, and reports with stakeholders\n- **Integrations**: Connects with Zoom, UserTesting, Typeform, Slack, and more\n\n## Where Koji Shines\n\n### Running Research Without a Researcher\n\nKoji is purpose-built for teams without dedicated researchers. Product managers, founders, and designers can launch a study in minutes without knowing how to write an interview guide or moderate a conversation. The AI handles everything.\n\nWith Dovetail, you still need to conduct the interviews yourself — schedule participants, run sessions, record and upload them — before Dovetail's analysis tools kick in.\n\n### Scale and Speed\n\nNeed 50 customer interviews this week? With Koji, all 50 can happen simultaneously — no scheduling, no coordination, no availability constraints. Participants complete interviews on their own time in 15–20 minutes.\n\nAccording to a 2025 Nielsen Norman Group report on research practices, the average qualitative research project takes 4–6 weeks from planning to insight delivery. Teams using AI-native interview tools regularly go from question to insight in 24–48 hours — a 10–15x improvement in research velocity.\n\n### Cost-Effective at Scale\n\nDovetail's per-user pricing adds up quickly for larger teams. With Koji you pay for what you use: interviews start as low as €1 per qualified interview, and there are no per-seat fees, so the price stays the same regardless of how many team members need to see the results.\n\n### No Prior Research Experience Required\n\nKoji's AI consultant guides you through study design, suggests methodologies, and helps you write effective interview questions. If you've never run a research study before, Koji gives you a running start.\n\n## Where Dovetail Shines\n\n### Large Research Libraries\n\nIf your team has months or years of existing research — recordings, transcripts, user session notes, support tickets — Dovetail is exceptional at making that data searchable and useful. Koji doesn't replace a dedicated repository for historical research.\n\n### Video and Multimedia Analysis\n\nDovetail supports video uploads and lets you highlight key moments, create showreels, and share clips with stakeholders. For teams that want to show executives a two-minute video montage of customers describing a problem, Dovetail excels.\n\n### Research Team Collaboration\n\nFor enterprise research teams with multiple researchers, research operations managers, and distributed stakeholders, Dovetail's collaboration and access control features are battle-tested. It's built for large-scale research operations.\n\n### Bringing Your Own Interviews\n\nIf you already conduct interviews via Zoom, Teams, or another platform and just need a better place to analyze them, Dovetail's intake integrations are excellent.\n\n## Best Choice For...\n\n- **Choose Koji if**: You want to run customer interviews quickly, don't have a research team, need AI to conduct and analyze interviews at scale, or want to go from question to insight in hours not weeks.\n\n- **Choose Dovetail if**: You have an existing research library to manage, conduct your own video interviews and need a place to analyze them, or have a large team that needs strong collaboration and repository tools.\n\n- **Consider using both if**: Your team runs high-velocity discovery research with Koji AND maintains a growing library of historical research in Dovetail. Many enterprise teams use purpose-built tools for each phase of the research cycle.\n\n## Pricing Comparison\n\n**Koji:**\n- Interviews as low as €1 per qualified interview, and €3 per qualified voice interview\n- Free account: 10 credits at signup, one active study, no card required\n- Pay as you go: no subscription needed\n- Volume pricing and plans when you want them\n- Enterprise available\n- No lock-in: pay only for the interviews your study actually uses\n\n**Dovetail:**\n- Free: 1 project, limited storage\n- Team: $15/user/month (billed annually)\n- Enterprise: custom pricing\n\nFor a team of five, Dovetail's Professional plan costs $75/month. Koji covers your entire team regardless of size, with interviews as low as €1 per qualified interview and no per-seat fees. For teams running multiple studies per month, Koji's economics become increasingly favorable.\n\n## Final Verdict\n\nKoji and Dovetail are more complementary than competitive — but if you're choosing between them based on budget or research focus, here's the honest take:\n\n**Koji wins** when your primary goal is generating new customer insights quickly and efficiently. It eliminates the two most time-consuming parts of qualitative research: conducting interviews and analyzing the results. Any team member can run a professional-grade study without prior research experience.\n\n**Dovetail wins** when your team already produces a high volume of research and needs a centralized place to store, search, and share it across the organization.\n\nIf you're a startup, product team, or agency that wants to run customer research without hiring a dedicated researcher or spending weeks on a single study, Koji is the better choice for 2026. [Start your first study free](https://www.koji.so) — no research experience required.\n\n*Last verified: March 2026*\n\n## Frequently Asked Questions\n\n**Q: Can Dovetail and Koji be used together?**\nA: Yes. You can use Koji to conduct AI interviews and export the transcripts or summaries into Dovetail for long-term storage and cross-study analysis.\n\n**Q: Does Dovetail conduct user interviews?**\nA: No. Dovetail is a research repository and analysis tool. You need to conduct interviews elsewhere — via Zoom, UserTesting, or Koji — and then bring the data into Dovetail.\n\n**Q: Which tool is better for small teams or startups?**\nA: Koji is better suited for small teams. It requires no research expertise, runs interviews automatically, and its pay-for-what-you-use credit pricing keeps costs low.\n\n**Q: Is Koji more affordable than Dovetail?**\nA: For most product teams running 2–4 studies per month, paying as you go from €1 per qualified interview is more cost-effective. Dovetail's per-user pricing can grow expensive as teams scale.\n\n**Q: Can Koji replace a research operations platform like Dovetail?**\nA: Koji handles the interview and analysis workflow end-to-end. For teams that need to manage large libraries of historical research or run complex multi-source analysis across years of data, Dovetail remains valuable as a repository.\n","category":"Research","lastModified":"2026-10-30T08:13:03.07716+00:00","metaTitle":"Koji vs Dovetail: Which Research Tool Wins? (2026)","metaDescription":"Koji vs Dovetail: an honest comparison of the top AI interview platform vs the leading research repository. Find the right fit for your team in 2026.","keywords":["koji vs dovetail","dovetail alternative","ai research tool","qualitative research platform","research repository comparison","user interview software 2026"],"aiSummary":"Koji is an AI interview platform that conducts and analyzes research automatically. Dovetail is a research repository for storing and collaborating on existing data. This comparison helps teams choose based on whether they need to run new research or manage existing research at scale.","aiKeywords":["Koji vs Dovetail","research repository","AI interview platform","qualitative research tools","user research software comparison","Dovetail alternative"],"aiContentType":"comparison","faqItems":[{"answer":"Yes. You can use Koji to conduct AI interviews and export the transcripts or summaries into Dovetail for long-term storage and cross-study analysis.","question":"Can Dovetail and Koji be used together?"},{"answer":"No. Dovetail is a research repository and analysis tool. You need to conduct interviews elsewhere — via Zoom, UserTesting, or Koji — and then bring the data into Dovetail.","question":"Does Dovetail conduct user interviews?"},{"answer":"Koji is better suited for small teams. It requires no research expertise, runs interviews automatically, and it starts as low as €1 per qualified interview, which is often more affordable than Dovetail's per-user pricing.","question":"Which tool is better for small teams or startups?"},{"answer":"For most product teams running 2-4 studies per month, paying as you go from €1 per qualified interview is more cost-effective. Dovetail's per-user pricing grows expensive as teams scale.","question":"Is Koji more affordable than Dovetail?"},{"answer":"Koji handles the interview and analysis workflow end-to-end. For teams that need to manage large libraries of historical research, Dovetail remains valuable as a dedicated repository.","question":"Can Koji replace a research operations platform like Dovetail?"}],"relatedTopics":["User Research Tools","AI Research","Research Repository","Qualitative Research","Product Discovery","Research Software Comparison"]},{"type":"blog","id":"416b0bf4-1479-4080-b10a-b94929d88745","slug":"koji-vs-dovetail","title":"Koji vs Dovetail: Which Research Tool Is Right for You?","url":"https://www.koji.so/blog/koji-vs-dovetail","summary":"Koji and Dovetail serve different research needs: Dovetail is a research repository for organizing and analyzing existing data, while Koji is an AI-native platform that conducts interviews and analyzes results automatically. Teams bottlenecked by interview volume and speed benefit most from Koji; teams managing large institutional research libraries benefit most from Dovetail.","content":"\nChoosing between Koji and Dovetail? They are both in the user research space — but they solve fundamentally different problems. Understanding that distinction will save you from buying the wrong tool.\n\n**The short answer:** Dovetail is a research repository that helps you organize and analyze research data after someone else collects it. Koji is an AI-native research platform that conducts the interviews for you and analyzes the results automatically. If you need a place to store and synthesize data from human-moderated sessions, Dovetail makes sense. If you need to run interviews at scale without hiring more researchers, Koji is the answer.\n\n## Quick Comparison\n\n| Feature | Koji | Dovetail |\n|---------|------|----------|\n| AI-moderated voice interviews | ✅ Natural AI conversations | ❌ Not available |\n| AI-moderated text interviews | ✅ Async chat interviews | ❌ Not available |\n| Automatic thematic analysis | ✅ AI-generated themes and insights | ⚠️ Manual tagging + AI summaries |\n| Research brief design | ✅ AI-guided study builder | ❌ Bring your own questions |\n| One-click research reports | ✅ Aggregated across all interviews | ⚠️ Requires manual effort |\n| Research repository | ⚠️ Study-level | ✅ Enterprise-grade repository |\n| Cross-study pattern analysis | ❌ Per-study analysis | ✅ Core feature |\n| 20+ third-party integrations | ❌ API-first | ✅ Gong, Salesforce, Intercom, more |\n| Participant recruitment | ✅ Built-in panel recruitment (quoted in credits) | ❌ Bring your own |\n| Free plan | ✅ 10 credits, no card, one active study | ✅ Limited free tier |\n| Starting price | From €1 per qualified interview | ~$30/month (individual) |\n| Team/business pricing | Mid-market | ~$375–$1,200/month |\n\n## What Is Dovetail?\n\nDovetail is a **research analysis and insights repository**. Teams use it to store interview recordings, transcripts, survey responses, and other research artifacts. Once data is in Dovetail, researchers apply tags and codes to find patterns, and the platform generates AI summaries of tagged content.\n\nDovetail does not conduct research. It does not moderate interviews. It does not generate research questions. It is an analysis layer placed on top of research that humans have already gathered.\n\n### Where Dovetail Shines\n\n**Centralized insights management.** If your organization runs dozens of studies per year with multiple researchers and stakeholders, Dovetail gives everyone a single place to store and search findings. This is genuinely valuable for large research teams trying to prevent knowledge from disappearing into individual documents.\n\n**Cross-study synthesis.** Dovetail lets you tag and search across multiple research projects. An insight from a study six months ago can surface when relevant to a new project — if it was tagged correctly in the first place.\n\n**Rich integrations.** Dovetail connects to Gong, Salesforce, Intercom, Zendesk, Slack, and more. If your team wants to pull support tickets or sales call transcripts into the same analysis layer as interview data, Dovetail makes that workflow possible.\n\n### Where Dovetail Falls Short\n\n**Manual tagging is a massive time investment.** One of the most common G2 complaints (2024–2025) is that building a usable tagging taxonomy requires weeks of upfront work — and teams struggle to apply it consistently across researchers. The promise of AI-assisted coding often fails to deliver, requiring researchers to review and correct tags manually.\n\n**AI features feel bolted on.** Dovetail was built as a manual analysis tool and added AI features to keep pace with the market. Multiple reviews describe AI summaries as unreliable and requiring regeneration — which creates more work rather than less. Looppanel's 2025 review notes that Dovetail AI struggles with hallucinations in summaries.\n\n**It cannot conduct interviews.** This is the most significant limitation for teams trying to scale their research. Dovetail requires a human moderator for every interview. Your research output is capped by your team's calendar availability — and by the cost and coordination overhead of human-moderated sessions.\n\n**Pricing surprises.** Enterprise plans commonly run $21,000+/year based on G2 reports, and Dovetail has faced criticism for raising prices without advance notice.\n\n## What Is Koji?\n\nKoji is an **AI-native qualitative research platform**. You design a research study with an AI consultant, set your objectives, and an AI interviewer conducts voice or text conversations with your participants automatically. The AI probes intelligently, follows up on interesting answers, and adapts the conversation based on what the participant says.\n\nAfter interviews complete, Koji automatically identifies themes, analyzes sentiment, and generates a research report aggregating insights across all conversations. The full cycle — from designing the study to having findings in hand — can happen in hours rather than weeks.\n\n### Where Koji Shines\n\n**It conducts the interviews.** Koji removes the bottleneck of human moderation. Instead of scheduling calls, coordinating with a moderator, and transcribing recordings, you share a link. Participants take the interview on their own time. The AI handles the conversation, asks follow-ups, and explores interesting threads — at 2 AM across 12 time zones simultaneously if needed.\n\n**Scale without headcount.** According to the Maze Future of User Research Report 2026, 66% of research teams saw increased demand without corresponding headcount growth. Koji solves this: you can run 500 interviews in the time it would take to schedule 5 human-moderated sessions.\n\n**Automatic analysis.** Koji analysis is built into the core product, not bolted on afterward. Every interview is automatically analyzed for themes and sentiment. No tagging taxonomy to build, no coding, no synthesis sprint.\n\n**No research expertise required.** The AI consultant helps you design your study, choose the right methodology, and write questions that generate useful answers. A product manager or founder with no research background can run a rigorous study without a dedicated researcher.\n\n### Where Koji Falls Short\n\n**Cross-study repository.** Koji analyzes studies individually. If you want to surface patterns across 20 studies spanning 18 months, you will need a separate repository tool like Dovetail. For teams with deep institutional research history they want to mine, this matters.\n\n**Analyzing existing data.** If you have a library of past recordings, transcripts, or survey responses to analyze, Koji does not import and analyze external data. Dovetail is built for exactly that.\n\n## When to Choose Koji vs Dovetail\n\n**Choose Koji if:**\n- You need to conduct new research and want to do it faster and at greater scale\n- You do not have dedicated research staff but need rigorous insights\n- You want to hear directly from users about a specific question, product area, or problem\n- You are validating a product idea, understanding churn, or preparing for a major roadmap decision\n- Speed to insight matters more than institutional knowledge management\n\n**Choose Dovetail if:**\n- You have a large team of researchers producing significant volumes of data\n- You need a centralized repository to prevent institutional knowledge loss\n- You want to mine existing research data (recordings, transcripts, CRM notes)\n- You have the resources to build and maintain a tagging taxonomy\n- Research already runs efficiently and you need better organization\n\n**Consider using both if:** Your team has Dovetail for managing your research repository AND needs Koji to increase the volume and speed of new research. Some teams use Koji to conduct interviews and bring transcripts into Dovetail for cross-study analysis over time.\n\n## Pricing Comparison\n\nKoji starts free with 10 credits and no card, then interviews run as low as €1 per qualified interview. Start with pay as you go. No subscription needed.\n\nDovetail individual plans start at ~$30/month, but team and business plans run $375–$1,200/month. Enterprise pricing commonly exceeds $21,000/year. Note that Dovetail pricing is per-user at higher tiers, making team costs unpredictable.\n\nFor startups and small teams, Koji delivers significantly more research output per dollar — because you are paying for the interviewing and analysis, not just storage. Dovetail value scales with how much pre-existing data you have to manage.\n\n## Our Take\n\nDovetail is a mature tool for teams that need to manage large volumes of research. But it cannot conduct research for you — and that is a fundamental limitation as teams face increasing research demand without headcount growth.\n\nKoji is purpose-built for the new era of AI-native research: where the AI does not just help you summarize findings, it actually gathers them. For most product teams, founders, and researchers who want to do more research faster, Koji is the more powerful choice.\n\n*Last verified: March 2026*\n\n## Frequently Asked Questions\n\n**Is Dovetail an AI interview tool?**\nNo. Dovetail is a research repository and analysis platform. It helps teams tag, code, and synthesize research data that humans have already collected. It does not conduct interviews or interact with research participants.\n\n**Can Koji replace Dovetail?**\nKoji replaces the interview moderation, transcription, and analysis parts of the research workflow. It does not replace Dovetail cross-study repository and knowledge management capabilities. Teams with large existing research libraries may want to use both tools together.\n\n**Which tool is better for startups?**\nKoji is generally better for startups. Most early-stage teams do not have a research repository problem — they have a \"how do we talk to customers fast enough to make good decisions\" problem. Koji solves that directly with a free account: 10 credits at signup, no card.\n\n**Does Koji have a free plan?**\nYes. Koji gives you 10 credits at signup with no card and one active study at a time, enough to run a complete research study and see real results before paying anything.\n\n**Can I import Dovetail data into Koji?**\nKoji is designed for new research studies rather than existing data repositories. Teams typically keep Dovetail for their existing research library while using Koji for new research projects going forward.\n  ","category":"Research","lastModified":"2026-10-30T08:13:02.012805+00:00","metaTitle":"Koji vs Dovetail: Which Research Tool Wins? — Koji Blog","metaDescription":"Dovetail organizes research data. Koji conducts AI interviews automatically. Compare features, pricing, and use cases to find the right tool for your team.","keywords":["koji vs dovetail","dovetail alternative","dovetail competitor","ai research tool","qualitative research platform","user research tool comparison","dovetail pricing"],"aiSummary":"Koji and Dovetail serve different research needs: Dovetail is a research repository for organizing and analyzing existing data, while Koji is an AI-native platform that conducts interviews and analyzes results automatically. Teams bottlenecked by interview volume and speed benefit most from Koji; teams managing large institutional research libraries benefit most from Dovetail.","aiKeywords":["Koji vs Dovetail","research repository","AI interviews","qualitative research tools","dovetail alternative","AI-moderated interviews","research analysis","user research platform"],"aiContentType":"comparison","faqItems":[{"answer":"No. Dovetail is a research repository and analysis platform. It helps teams tag, code, and synthesize research data that humans have already collected. It does not conduct interviews or interact with research participants.","question":"Is Dovetail an AI interview tool?"},{"answer":"Koji replaces the interview moderation, transcription, and analysis parts of the research workflow. It does not replace Dovetail cross-study repository features. Teams with large existing research libraries may want to use both tools together.","question":"Can Koji replace Dovetail?"},{"answer":"Koji is generally better for startups. Most early-stage teams need to talk to customers faster, not manage a research repository. Koji includes a free account with 10 credits at signup and no card.","question":"Which tool is better for startups?"},{"answer":"Yes. Koji gives you 10 credits at signup with no card and one active study at a time, enough to run a complete research study and see results before paying anything.","question":"Does Koji have a free plan?"},{"answer":"Koji is designed for new research studies rather than existing data repositories. Teams typically keep Dovetail for existing research while using Koji for new research projects.","question":"Can I import Dovetail data into Koji?"}],"relatedTopics":["User Research Tools","Qualitative Research","AI Research","Research Operations","Tool Comparisons"]},{"type":"blog","id":"a369dc25-79db-4539-865d-191658728eb3","slug":"best-qualitative-research-tools-2026","title":"Best Qualitative Research Tools in 2026: The Complete Buyer's Guide","url":"https://www.koji.so/blog/best-qualitative-research-tools-2026","summary":"A complete buyer guide comparing the top 7 qualitative research platforms in 2026 — Koji, Dovetail, UserTesting, Maze, Lookback, UserInterviews, and Optimal Workshop. Covers real pricing, features, and which tool fits which team. Koji leads for AI-native interview research; Dovetail leads for research repositories.","content":"\nThe best qualitative research tool in 2026 depends on your team size, research volume, and whether you want AI to run the interviews for you or just help analyze them afterward.\n\nThis guide covers the top qualitative research platforms available today — with honest assessments of strengths, weaknesses, pricing, and the specific teams each tool serves best.\n\n**Quick answer:** If you want AI to conduct and analyze your interviews with no moderator required, Koji is the top pick for 2026. If you need a research repository for large teams, Dovetail leads. For enterprise panel-based video research, UserTesting remains the standard.\n\n## Why Qualitative Research Tools Matter More Than Ever\n\nThe qualitative research software market has reached **$577 million in 2025** and is projected to hit **$1.229 billion by 2032**, growing at a 10.9% CAGR (Valuates Reports, 2025). The growth is driven by one force above all others: AI.\n\nAccording to the Maze Future of User Research Report 2026, **88% of researchers** identified AI-assisted analysis as the number one trend shaping the field — outranking every other development. Meanwhile, **83% of market research professionals** plan to invest in AI tools for research in the next 12 months.\n\nThe tools you evaluate today are fundamentally different from those available two years ago. AI has moved from \"nice to have\" to table stakes.\n\n## What Is Qualitative Research Software?\n\nQualitative research tools help teams understand the *why* behind user behavior. Unlike surveys that collect numerical ratings, qualitative tools capture open-ended conversations, observations, and exploratory feedback — the kind of insight that tells you not just *what* users do, but *why* they do it.\n\nModern tools fall into three categories:\n1. **Interview platforms** — tools that help you conduct or moderate interviews (Koji, Lookback, UserTesting)\n2. **Analysis platforms** — tools that help you analyze existing research data (Dovetail, Notably, Aurelius)\n3. **End-to-end platforms** — tools that do both, from design through analysis (Koji)\n\n## The Top 7 Qualitative Research Tools in 2026\n\n### 1. Koji — Best for AI-Native Interview Research\n\n**Best for:** Product teams, founders, agencies, and anyone who needs research insights fast without a dedicated researcher.\n\n**Starting price:** Free with 10 credits; then as low as €1 per qualified interview\n\nKoji is the most advanced AI-native interview platform available in 2026. It is the only tool on this list where AI actually *conducts* the interviews — voice and text — so you never need to schedule, moderate, or transcribe anything yourself.\n\nHere is how it works: you describe your research goals to Koji's AI consultant, which designs a study brief and interview guide. You reach your own people with personalised links when you want to track who answered, or a generic anonymous link when candour matters, or open Panel recruit and Koji finds the participants for you: describe the audience, approve a per-respondent quote in credits, and those interviews land in the same study. The AI moderator conducts natural, probing conversations — asking follow-up questions, exploring unexpected responses, and keeping conversations on track. When interviews are done, Koji automatically identifies themes, sentiment patterns, and key insights across all conversations, then generates a structured report with one click.\n\n**Key features:**\n- AI-moderated voice and text interviews — no human moderator needed\n- Automatic thematic analysis and report generation\n- Customizable AI consultant with tone and depth controls\n- No moderator bias — consistent, fair questioning across all participants\n- Embeddable widget and REST API for workflow integration\n- Free plan: 10 credits at signup, no card, one active study at a time\n\n**What sets Koji apart:** Running 100 interviews simultaneously with full analysis completed automatically transforms research velocity. Teams report going from research question to stakeholder-ready insights in under 48 hours — a process that previously took 4–6 weeks.\n\nResearch from Maze (2026) shows that organizations where research is embedded into every business decision are **5x more likely to improve brand perception**, achieve **3.6x more active users**, and see **3.2x better product-market fit** compared to organizations that rarely incorporate research. Koji makes that level of continuous research affordable and operationally feasible for any team.\n\n**Limitations:** Koji is designed for interview-based discovery and qualitative research. For usability testing with task-based click tracking or prototype evaluation, a complementary tool like Maze is needed.\n\n---\n\n### 2. Dovetail — Best Research Repository for Large Teams\n\n**Best for:** Enterprise research teams with large libraries of existing qualitative data.\n\n**Starting price:** Free; Professional $15/user/month\n\nDovetail is the industry standard for research repositories. It excels at storing, tagging, searching, and sharing qualitative research across an organization. Upload recordings, transcripts, field notes, and support tickets — Dovetail indexes everything and makes it searchable.\n\nDovetail's AI suite (called \"Magic\") includes auto-highlighting, thematic clustering, automatic PII redaction, and AI-generated summaries. Its \"Channels\" feature continuously classifies and tracks themes across large incoming datasets — from app store reviews to support tickets — starting at $50/month for 500 data points.\n\nDovetail users who regularly use Magic features report saving an average of **10 hours per week** on data analysis — a significant productivity gain for dedicated research teams.\n\n**Key features:**\n- Central repository for all qualitative data — video, audio, transcripts, notes\n- Magic AI suite: highlight, cluster, summarize, and redact\n- Channels for continuous feedback monitoring (support, reviews, NPS)\n- Participant recruitment database of 3+ million participants\n- Strong team collaboration and enterprise-grade access controls\n\n**Limitations:** Dovetail does not conduct interviews. You research elsewhere and bring data into Dovetail. Per-user pricing grows expensive for large teams.\n\n---\n\n### 3. UserTesting — Best for Human-Panel Video Research\n\n**Best for:** Enterprise teams requiring moderated video sessions with real users.\n\n**Starting price:** Custom (typical SMB spend: $36,265/year)\n\nUserTesting has been a market leader for over 15 years. It provides access to a vetted panel from 40+ partner networks who complete video-recorded sessions — speaking aloud as they navigate products, answer questions, or react to concepts.\n\nThe platform includes AI-assisted analysis: automatic insight summaries, friction detection using machine learning, and AI-generated survey themes that scale to 1,000+ respondents.\n\nA typical 5-seat Advanced plan lists at approximately **$49,711/year** but is frequently negotiated down to around $25,700. Average SMB spend lands at **$36,265/year** (Vendr buyer data, 2025).\n\n**Key features:**\n- Human panel from 40+ global networks\n- Video recording with think-aloud protocols\n- AI Insight Summary and friction detection\n- Live conversation and moderated interview sessions\n- 100+ pre-built test templates\n\n**Limitations:** Expensive for most teams. Slower turnaround than AI tools. Requires research expertise for best results. Not cost-effective for high-volume discovery research.\n\n---\n\n### 4. Maze — Best for Prototype and Usability Testing\n\n**Best for:** Product and design teams validating prototypes, user flows, and task completion.\n\n**Starting price:** $75/seat/month; $199/month unlimited\n\nMaze is built for usability testing, not interview research. It connects to Figma, InVision, and other design tools to let users interact with prototypes while capturing task completion rates, drop-off points, and heatmaps. Maze also offers survey and interview modules for mixed-method research.\n\nIf your primary question is \"can users complete this task?\" Maze is excellent. If you want to understand *why* users make the decisions they do, Koji's conversational interview format goes deeper.\n\n---\n\n### 5. Lookback — Best for Live Moderated Sessions\n\n**Best for:** Researchers who want to conduct live moderated interviews with their own participants.\n\n**Starting price:** $25/month (Freelance); $149/month (Team)\n\nLookback is a screen-sharing and session recording platform built for moderated user research. Researchers observe, take notes, and highlight moments in real time while participants complete tasks or answer questions. Lookback also supports self-guided (unmoderated) sessions.\n\n---\n\n### 6. UserInterviews — Best for Participant Recruitment\n\n**Best for:** Teams that have their research process set but struggle to find quality participants.\n\n**Starting price:** ~$40/session + incentive costs\n\nUserInterviews is a participant recruitment and scheduling platform. You define screener criteria, and UserInterviews finds matching participants from their network. It fills a specific gap: many interview tools, such as Dovetail and Lookback, require you to source your own participants. Koji has built-in panel recruitment; if you prefer to run recruitment yourself, share your Koji link with the people UserInterviews finds and the interviews land in the same study.\n\n---\n\n### 7. Optimal Workshop — Best for Information Architecture Testing\n\n**Best for:** Teams testing navigation, site structure, and content organization.\n\n**Starting price:** $107/month individual; custom for teams\n\nOptimal Workshop specializes in card sorting, tree testing, and first-click testing — methods for evaluating how well users can navigate and find information within a product. It is highly specialized but excellent for what it does.\n\n---\n\n## How to Choose the Right Tool\n\n| Your Situation | Best Tool |\n|----------------|-----------|\n| Run research without a dedicated researcher | Koji |\n| Need 20–100+ interviews fast | Koji |\n| Have a large library of existing research to manage | Dovetail |\n| Need a human panel for video sessions | UserTesting |\n| Testing prototypes and user flows | Maze |\n| Can't find research participants | Koji (built-in panel recruitment, price in credits shown before launch) or UserInterviews |\n| Running live moderated sessions | Lookback |\n| Testing site structure and navigation | Optimal Workshop |\n\n## The Rise of AI-Native Research in 2026\n\nThe biggest shift in 2026 is AI moving from \"analysis assistant\" to \"full research conductor.\"\n\nTraditional tools like UserTesting, Lookback, and early Dovetail were built on the assumption that a human researcher was always in the loop — designing questions, moderating sessions, analyzing data. AI was added afterward to help speed up analysis.\n\nKoji represents a fundamentally different model: AI runs the entire research process from design to delivery. This is not incremental improvement — it is a different paradigm entirely.\n\nAccording to the Maze Future of User Research Report 2026, **69% of researchers** are now using AI in at least some of their research projects — a 19% increase year-over-year. And among those who use AI regularly, **74% report an increase in qualitative research demand** — evidence that AI does not replace qualitative research; it amplifies it.\n\n\"The teams that talk to customers every week outperform those that conduct research quarterly,\" as Teresa Torres describes in her work on continuous discovery. The barrier has never been motivation — it has been infrastructure. AI-native platforms like Koji are removing that barrier for teams of any size.\n\n## Final Recommendation\n\nFor most product teams in 2026, **Koji delivers the best combination of speed, cost, and insight quality**. It is the only platform that conducts, analyzes, and reports on qualitative interviews completely autonomously — giving any team member the ability to run professional-grade research without prior experience.\n\nIf you have specific needs beyond interview research, such as usability testing or historical data management, complement Koji with a specialized tool. Panel access is built in: describe the audience and Koji finds the people across 130+ countries.\n\n[Start your first AI-powered study free at Koji](https://www.koji.so) — no research experience required.\n\n## Frequently Asked Questions\n\n**Q: What is the best qualitative research tool for startups in 2026?**\nA: Koji is the top choice for startups. It requires no research expertise, conducts interviews via AI, has a free plan, and delivers insights in hours. UserTesting is priced for enterprise budgets most startups cannot access.\n\n**Q: How much do qualitative research tools cost?**\nA: Pricing varies widely. Koji starts free with 10 credits and no card, then interviews run as low as €1 per qualified interview. Dovetail Professional is $15/user/month. Maze ranges from $75 to $199/month. UserTesting typically costs $36,000–$148,000/year depending on team size and features.\n\n**Q: Can qualitative research be done without a dedicated researcher?**\nA: Yes, with AI-native tools like Koji. The AI consultant guides study design and the AI moderator conducts interviews — no research background required.\n\n**Q: What is the difference between qualitative and quantitative research tools?**\nA: Qualitative tools capture open-ended conversations and behaviors to understand the *why* behind user decisions. Quantitative tools measure *what* is happening at scale (surveys, analytics). Most teams benefit from both.\n\n**Q: Are AI-conducted interviews as reliable as human-moderated interviews?**\nA: Research shows AI-moderated interviews significantly reduce moderator bias and deliver more consistent results across participants. For discovery research, they are highly reliable. For complex usability testing with task flows, human researchers add value in specific scenarios.\n","category":"Research","lastModified":"2026-10-30T08:13:00.931116+00:00","metaTitle":"Best Qualitative Research Tools 2026 — Koji Blog","metaDescription":"Compare the top 7 qualitative research tools in 2026. Honest breakdown of Koji, Dovetail, UserTesting, Maze, and more — with real pricing and use cases.","keywords":["best qualitative research tools 2026","qualitative research software","user research tools comparison","best ux research tools","AI research platform","user interview software"],"aiSummary":"A complete buyer guide comparing the top 7 qualitative research platforms in 2026 — Koji, Dovetail, UserTesting, Maze, Lookback, UserInterviews, and Optimal Workshop. Covers real pricing, features, and which tool fits which team. Koji leads for AI-native interview research; Dovetail leads for research repositories.","aiKeywords":["qualitative research tools","UX research software","AI interview platform","research repository","best user research tools 2026","UserTesting alternative","Dovetail alternative"],"aiContentType":"guide","faqItems":[{"answer":"Koji is the top choice for startups. It requires no research expertise, conducts interviews via AI, has a free plan, and delivers insights in hours. UserTesting is priced for enterprise budgets most startups cannot access.","question":"What is the best qualitative research tool for startups in 2026?"},{"answer":"Pricing varies widely. Koji starts free with 10 credits and no card, then interviews run as low as €1 per qualified interview. Dovetail Professional is $15/user/month. Maze ranges from $75 to $199/month. UserTesting typically costs $36,000 to $148,000 per year depending on team size.","question":"How much do qualitative research tools cost in 2026?"},{"answer":"Yes, with AI-native tools like Koji. The AI consultant guides study design and the AI moderator conducts interviews — no research background required.","question":"Can qualitative research be done without a dedicated researcher?"},{"answer":"Qualitative tools capture open-ended conversations to understand the why behind user decisions. Quantitative tools measure what is happening at scale. Most research programs benefit from both.","question":"What is the difference between qualitative and quantitative research tools?"},{"answer":"AI-moderated interviews significantly reduce moderator bias and deliver more consistent results across participants. For discovery research they are highly reliable. For complex usability testing with specific task flows, human researchers add value in specific scenarios.","question":"Are AI-conducted interviews as reliable as human-moderated interviews?"}],"relatedTopics":["User Research Tools","AI Research","Research Software","Qualitative Research","Product Discovery","UX Research","Research Buyer Guide"]},{"type":"blog","id":"cdc57651-b88d-4025-ba9f-c0264ae0de19","slug":"koji-vs-usertesting","title":"Koji vs UserTesting: Which User Research Platform Is Right for You?","url":"https://www.koji.so/blog/koji-vs-usertesting","summary":"Koji and UserTesting answer different research questions: UserTesting is optimized for observational task-based research with video recording and a large participant panel, while Koji conducts AI-moderated conversational interviews automatically at a fraction of the cost. Koji is best for teams that need to understand user motivations and run research continuously; UserTesting is best for enterprise teams needing video-based usability testing at scale.","content":"\nChoosing between Koji and UserTesting? Both platforms help teams understand how users experience their products — but they take fundamentally different approaches, target different team sizes, and deliver different types of insight.\n\n**The short answer:** UserTesting is an enterprise-grade video testing platform with a massive participant panel. It is best for large organizations that need to observe users completing tasks on a live product, at scale, with video recordings. Koji is an AI-native interview platform that conducts conversational research automatically — best for teams who need to understand *why* users behave the way they do, faster and at a fraction of the cost.\n\n## Quick Comparison\n\n| Feature | Koji | UserTesting |\n|---------|------|-------------|\n| AI-moderated interviews | ✅ Voice and text | ❌ Not available |\n| Unmoderated video testing | ❌ | ✅ Core feature |\n| Moderated live sessions | ❌ | ✅ Available |\n| Participant panel | ✅ Built-in recruitment quoted in credits, or bring your own | ✅ 1.5M+ vetted testers |\n| Automatic thematic analysis | ✅ Themes, sentiment, insights | ⚠️ AI highlights from video |\n| One-click research reports | ✅ Full aggregate report | ⚠️ Manual synthesis required |\n| Study design AI assistant | ✅ AI consultant | ❌ |\n| Prototype/live website testing | ❌ | ✅ Screen + click recording |\n| No scheduling required | ✅ Fully async | ✅ Unmoderated sessions |\n| Free plan | ✅ 10 credits, no card, one active study | ❌ No free tier |\n| Starting price | From €1 per qualified interview | $15,000+/year |\n| Best team size | Startups to enterprise | Enterprise |\n\n## What Is UserTesting?\n\nUserTesting is a video-based user research platform used primarily by enterprise teams. Its core product lets you recruit participants from a panel of 1.5M+ vetted testers and ask them to complete tasks on your website, prototype, or app while narrating their thoughts aloud. You receive video recordings, transcripts, and behavioral data from each session.\n\nUserTesting also supports moderated sessions, where a researcher joins live to guide the participant. More recently, the platform has added AI-powered features to help teams surface highlights and patterns across video recordings faster.\n\nUserTesting is owned by UserZoom (acquired in 2022), making it one of the largest consolidated enterprise research platforms.\n\n### Where UserTesting Shines\n\n**Participant panel breadth.** UserTesting panel of 1.5M+ vetted participants spans demographics, geographies, and professional segments. If you need 15 small business owners between the ages of 45–55 in the United Kingdom by tomorrow morning, UserTesting can typically deliver that. No other platform in this guide has comparable panel reach.\n\n**Observational, task-based research.** When you need to watch real users navigate your product — where they click, where they hesitate, where they get confused — UserTesting video sessions are the gold standard. The combination of screen recording, facial expressions, and verbal commentary gives researchers a rich dataset to work with.\n\n**Enterprise infrastructure.** UserTesting offers the security, compliance, SSO, and procurement processes that large organizations require. For research teams embedded in Fortune 500 companies, this matters.\n\n**Mature tooling.** UserTesting has been building its platform for more than 15 years. The session recording tools, highlight reel creation, and sharing features are polished and reliable.\n\n### Where UserTesting Falls Short\n\n**Cost.** This is the most significant barrier. UserTesting pricing starts at approximately $15,000/year, with most enterprise contracts in the $30,000–$200,000+/year range depending on volume and features. For most startups and growing companies, this is inaccessible.\n\n**No AI interviewer.** UserTesting does not conduct conversational interviews. Participants follow a script and complete tasks; the platform does not probe, follow up on interesting answers, or adapt the conversation based on what participants say. You are essentially watching people complete a pre-defined script rather than having a dialogue.\n\n**Research at scale requires significant budget.** Each participant recruited from the UserTesting panel carries a per-session cost. Running 100 conversational interviews to understand a complex behavioral question would cost tens of thousands of dollars through UserTesting — and would require a human moderator for each session.\n\n**Synthesis is manual.** Despite AI highlight features, turning a library of video sessions into a synthesized, shareable research report still requires significant manual effort. According to the Lyssna Research Synthesis Report 2025, 60.3% of researchers cite manual synthesis work as their biggest frustration — UserTesting does not fully solve this.\n\n## What Is Koji?\n\nKoji is an AI-native qualitative research platform. You design a research study with an AI consultant, set your objectives, and an AI interviewer conducts natural voice or text conversations with your participants automatically. There is no scheduling, no human moderator, and no post-session transcription work.\n\nAfter interviews complete, Koji identifies themes, analyzes sentiment, and generates a research report aggregating insights across all conversations. The full cycle — from research question to shareable findings — can happen in hours.\n\n### Where Koji Shines\n\n**Conversational depth at scale.** Koji AI interviewer asks contextual follow-up questions, probes interesting answers, and explores threads the discussion guide never anticipated. Where UserTesting gets participants to complete a task, Koji gets participants to explain their mental model, motivations, and frustrations in their own words — without a moderator blocking your calendar.\n\n**Cost efficiency.** Koji starts free with 10 credits and no card, then interviews run as low as €1 per qualified interview. Running 50 in-depth conversational interviews through Koji costs a fraction of what a single UserTesting enterprise contract costs. This makes regular, continuous research economically viable for any team size.\n\n**Automatic synthesis.** Koji analysis is not an add-on — it is built into the platform. Every interview is automatically analyzed for themes and sentiment. When you generate a report, Koji aggregates findings across all interviews and surfaces patterns automatically. No manual video review, no tagging, no synthesis sprint.\n\n**Speed to insight.** According to the Maze Future of User Research Report 2026, 22% of organizations now use research at every level of business strategy — up from 8% just a year prior. That pace requires insights on the timeline of business decisions, not 2-week research cycles. Koji makes same-week research the norm.\n\n**Accessible to non-researchers.** The AI consultant guides any team member through study design and question writing. A product manager or founder with no research training can run a rigorous study without needing a dedicated researcher or research operations budget.\n\n### Where Koji Falls Short\n\n**Task-based observational research.** If you need to watch users navigate your product — where they click, where they drop off, how they interact with a specific interface — Koji is not the right tool. This is UserTesting core strength. Koji is built for conversational research (why do users behave this way), not observational research (how do users interact with this interface).\n\n**No tester pool to browse.** Koji does not operate a 1.5M-person tester pool the way UserTesting does, and you never pick respondents from one. Instead you describe the audience and Koji recruits it from its built-in panel across 130+ countries, with a live per-respondent quote in credits you approve before anything launches. If you already have users, reach them directly: personalised links from an imported list, a generic anonymous link, an embed in your product, or a call. UserTesting's curated pool is still the stronger fit when you want to browse and hand-pick testers yourself.\n\n## The Core Difference: Why vs How\n\nThe most useful way to understand the Koji vs UserTesting choice is by the research questions each tool answers:\n\n**UserTesting answers \"How do users interact with my product?\"**\n- Where do users click on this page?\n- Can users complete the checkout flow?\n- Where do users get confused in this onboarding sequence?\n\n**Koji answers \"Why do users make the decisions they make?\"**\n- Why did users churn after 30 days?\n- What are the biggest frustrations users have with this workflow?\n- What would make users recommend this product to a colleague?\n- What alternatives did users consider before choosing us?\n\nBoth questions matter. The best-resourced research teams use tools that answer both.\n\n## When to Choose Koji vs UserTesting\n\n**Choose Koji if:**\n- You need to understand user motivations, decision-making, and frustrations at scale\n- Your team does not have a $15,000+ annual research tools budget\n- You want to run research continuously rather than in occasional large studies\n- You need findings fast (days, not weeks)\n- You do not have a dedicated research team but need rigorous qualitative insights\n- You want to hear directly from your own customers in their own words\n\n**Choose UserTesting if:**\n- You are an enterprise organization with a significant research budget\n- You need to observe task completion on a live product or prototype\n- You need a large, vetted third-party participant panel attached to recorded video sessions\n- You have dedicated research staff who need enterprise tooling and compliance features\n- Your primary research questions are about usability and UI behavior, not customer motivation\n\n**Consider using both if:** Your team needs both observational (usability) research and conversational (motivational) research. Koji handles the \"why\" at scale; UserTesting handles the \"how\" with video proof.\n\n## Pricing Comparison\n\n| | Koji | UserTesting |\n|---|---|---|\n| Free tier | ✅ 10 credits, no card, one active study | ❌ None |\n| Entry price | From €1 per qualified interview | ~$15,000/year |\n| Mid-market | Volume pricing and plans when you want them | ~$30,000–$75,000/year |\n| Enterprise | Custom | ~$100,000–$200,000+/year |\n\nThe cost difference is significant. A startup or small product team can run meaningful, recurring customer research with Koji for less than $1,200/year. The equivalent budget at UserTesting would not cover a single contract.\n\n## Our Take\n\nUserTesting built the enterprise standard for video-based usability research over 15+ years. For large organizations that need to observe users completing tasks at scale, with enterprise compliance and a vast participant panel, it remains a strong choice.\n\nBut for most product teams — especially those trying to understand customers more deeply, run research more frequently, or make qualitative research accessible across the organization without dedicated researchers — Koji delivers more research value per dollar by an order of magnitude. The AI-native approach removes the cost, scheduling, and synthesis overhead that has historically made deep qualitative research a luxury for well-resourced teams.\n\n*Last verified: March 2026*\n\n## Frequently Asked Questions\n\n**Is UserTesting worth the price?**\nFor enterprise teams with dedicated research budgets and a need for large-scale video-based usability testing, UserTesting is well-established. For most startups and growing companies, the cost is prohibitive relative to alternatives like Koji that deliver comparable or greater insight depth for a fraction of the price.\n\n**Does Koji have a participant panel like UserTesting?**\nNot a proprietary tester panel of its own, but Koji does recruit for you. Describe the audience (market, demographics, role, screening) and Koji returns a live per-respondent quote in credits from Koji's panel, which you approve before launch. If you already have users, you can distribute your study link to them directly via email, in-app messaging, or any channel you use to reach customers.\n\n**Which tool is better for usability testing?**\nFor task-based usability testing where you need to observe users navigating an interface with video recording, UserTesting is purpose-built for that use case. Koji is better suited to conversational research — understanding motivations, frustrations, and decision-making at scale.\n\n**Can Koji conduct the same type of research as UserTesting?**\nThey address different research questions. UserTesting is optimized for observational research (how users interact with your product). Koji is optimized for conversational research (why users think and feel the way they do about your product and category). Most product teams need both types of research.\n\n**Is there a free alternative to UserTesting?**\nKoji gives you 10 credits at signup with no card and one active study at a time. For recruitment, User Interviews offers project-based pricing without an annual commitment. Lookback also offers a 60-day free trial for moderated sessions.\n  ","category":"Research","lastModified":"2026-10-30T08:12:59.836851+00:00","metaTitle":"Koji vs UserTesting: Which Platform Wins? — Koji Blog","metaDescription":"UserTesting is the enterprise standard for video-based usability research. Koji is the AI-native alternative for conversational interviews. Compare pricing, features, and use cases.","keywords":["koji vs usertesting","usertesting alternative","usertesting competitor","ai user research tool","user research platform comparison","usertesting pricing","qualitative research tools"],"aiSummary":"Koji and UserTesting answer different research questions: UserTesting is optimized for observational task-based research with video recording and a large participant panel, while Koji conducts AI-moderated conversational interviews automatically at a fraction of the cost. Koji is best for teams that need to understand user motivations and run research continuously; UserTesting is best for enterprise teams needing video-based usability testing at scale.","aiKeywords":["Koji vs UserTesting","UserTesting alternative","AI interview platform","user research platform","usability testing","qualitative research","AI-moderated interviews","user testing pricing"],"aiContentType":"comparison","faqItems":[{"answer":"For enterprise teams with dedicated research budgets needing large-scale video-based usability testing, UserTesting is well-established. For most startups and growing companies, the cost is prohibitive. Koji delivers comparable insight depth for a fraction of the price.","question":"Is UserTesting worth the price?"},{"answer":"Not a proprietary tester panel like UserTesting's, but Koji does recruit for you. Describe the audience you want and Koji returns a live per-respondent quote in credits from Koji's panel, which you approve before launch. If you already have users, you can simply distribute your study link to them instead.","question":"Does Koji have a participant panel like UserTesting?"},{"answer":"For task-based usability testing with video recording, UserTesting is purpose-built for that use case. Koji is better for conversational research: understanding motivations, frustrations, and decision-making at scale.","question":"Which tool is better for usability testing?"},{"answer":"They address different research questions. UserTesting is optimized for observational research (how users interact with your product). Koji is optimized for conversational research (why users behave the way they do).","question":"Can Koji conduct the same type of research as UserTesting?"},{"answer":"Koji gives you 10 credits at signup with no card and one active study at a time. User Interviews offers project-based recruitment pricing. Lookback offers a 60-day free trial for moderated sessions.","question":"Is there a free alternative to UserTesting?"}],"relatedTopics":["User Research Tools","Tool Comparisons","Qualitative Research","Usability Testing","AI Research","Product Research"]},{"type":"blog","id":"d66e3e4d-9437-4a2e-8aa3-fbbafb6410b2","slug":"koji-vs-maze-2026","title":"Koji vs Maze: Which Research Tool Is Right for Your Team? (2026)","url":"https://www.koji.so/blog/koji-vs-maze-2026","summary":"Koji and Maze serve different research needs: Koji conducts AI-moderated qualitative depth interviews and synthesizes themes automatically, while Maze runs quantitative usability tests on prototypes with click maps and task metrics. Most mature research teams use both — Koji for discovery, Maze for validation.","content":"\nKoji and Maze both sit in the \"product research\" category — but they serve fundamentally different research needs. Here's a direct comparison to help you decide which belongs in your stack.\n\n**The short answer:** Maze is built for quantitative usability testing — measuring task completion rates, click paths, and survey responses at scale. Koji is built for qualitative discovery — AI conducts open-ended voice and text interviews, then automatically surfaces themes, sentiment, and insights. They're often more complementary than competing, but if your budget only allows one, the choice comes down to whether you need *metrics* or *meaning*.\n\n## Quick Comparison\n\n| Feature | Koji | Maze |\n|---------|------|------|\n| AI-moderated interviews | ✅ Voice + text conversations | ❌ Survey + task-based tests |\n| Automated thematic analysis | ✅ Themes, sentiment, key quotes | ❌ Metrics and quantitative data |\n| Voice interviews | ✅ Natural AI conversations | ❌ Not available |\n| Usability / task testing | ❌ Not the core use case | ✅ Click maps, heatmaps, task flows |\n| Prototype testing | ❌ | ✅ Figma integration |\n| Built-in participant panel | ✅ Built-in panel recruitment, quoted in credits | ✅ Access to panel participants |\n| One-click insight reports | ✅ Qualitative aggregate reports | ⚠️ Quantitative dashboards |\n| Tree testing / card sorting | ❌ | ✅ |\n| Free plan | ✅ 10 free credits, one active study | ✅ Free plan available |\n| Starting paid price | From €1 per qualified interview; pay as you go, no subscription | $99/seat/month |\n| Best for | Discovery interviews, churn analysis, qualitative depth | Usability testing, design validation, quantitative feedback |\n\n## Deep Dive: Koji\n\nKoji is an AI-native platform where an AI interviewer conducts voice or text conversations with your participants and automatically synthesizes findings into themes, sentiment analysis, and insight reports.\n\n### What Koji Does Best\n\n**Open-ended discovery at scale.** Koji shines when you need to understand *why* — why customers churn, why a feature isn't adopted, what job they're trying to do, what frustrated them about a competitor. These are questions that task-based testing can't answer.\n\n**No human moderator needed.** The AI follows your research guide, probes naturally, and maintains consistent quality across 10 or 10,000 sessions. According to the Maze Future of User Research Report 2026, 69% of research teams now use AI in their workflow — and Koji is purpose-built for this shift.\n\n**Automatic synthesis.** After each session, Koji extracts themes and key quotes. Once you have sufficient responses, a one-click report synthesizes insights across all participants — no manual coding, no affinity mapping. Research from Dscout shows analysis consumes 32.7% of total project time; Koji is designed to reclaim that time.\n\n**Accessible to the whole team.** Koji's AI consultant guides study design, so product managers, founders, and customer success teams can run rigorous research without a research background.\n\n### Where Koji Has Limitations\n\nKoji doesn't support prototype testing, click maps, or task-completion tracking. If you need to validate whether users can complete a specific flow in your product, you'll want a dedicated usability tool. For participants, you do not need your own list: describe the audience and Koji's built-in recruitment quotes the credit cost per respondent before launch. If you do have a list, Koji interviews that instead.\n\n## Deep Dive: Maze\n\nMaze is a continuous product discovery platform that focuses on quantitative, metric-driven research. It's built for product and design teams who need fast, measurable feedback on designs and flows.\n\n### What Maze Does Best\n\n**Prototype and usability testing.** Maze integrates directly with Figma, Sketch, and InVision, making it fast to set up prototype tests. Participants complete tasks while Maze captures click paths, heatmaps, time on task, and success rates.\n\n**Quantitative at scale.** Maze excels when you need statistically significant data — A/B testing design options, measuring completion rates across user segments, or tracking usability metrics over time.\n\n**Tree testing and card sorting.** For information architecture research — figuring out how users expect content to be organized — Maze's tree testing and card sorting features are well-regarded by UX practitioners.\n\n**Panel access.** Maze offers access to a panel of participants, which removes the recruitment friction when you need quick feedback from a specific demographic.\n\n### Where Maze Has Limitations\n\n**Qualitative depth is limited.** Maze surveys can collect open-ended text, but you won't get the conversational depth that reveals underlying motivations, mental models, or emotional context. A participant can click through your prototype successfully and you'll still not know *why* they almost gave up at step 3.\n\n**Analysis burden remains.** Maze gives you metrics dashboards, but interpreting the *meaning* behind the data still requires human judgment. You may find yourself scheduling follow-up interviews to understand what the click data actually means.\n\n**Per-seat pricing.** Maze charges per seat on paid plans, which can make it expensive for cross-functional teams where multiple stakeholders want access to research findings.\n\n## When to Choose Koji vs Maze\n\n**Choose Koji if:**\n- Your research goal is understanding motivations, needs, and mental models\n- You're running discovery interviews, churn analysis, or feature validation through conversation\n- You want AI to conduct interviews and synthesize findings automatically\n- You don't have dedicated researchers and need a tool that guides you through the process\n- You need depth, not metrics\n\n**Choose Maze if:**\n- You need to validate whether users can complete specific tasks in your product\n- You're testing design alternatives and need quantitative comparison data\n- You're doing information architecture research (tree testing, card sorting)\n- You have Figma prototypes ready to test with measurable task flows\n- You need metrics, not depth\n\n**Use both if:**\n- You run a full research program: use Koji for discovery and depth understanding, use Maze for design validation and task testing\n- You need to first understand the problem space (Koji), then validate your solution (Maze)\n- Your team includes both researchers running discovery and designers running validation\n\nThis is actually the most common setup for mature research teams: qualitative tools for depth, quantitative tools for validation.\n\n## Pricing Comparison\n\n**Koji:** Free account with 10 credits and one active study, no card. Interviews start as low as €1 per qualified interview, and €3 per qualified voice interview. Start with pay as you go. No subscription needed. There are no per-seat charges, so you pay only for the interviews your study actually uses and the whole team can access insights.\n\n**Maze:** Free plan available with basic features. Growth plan at $99/seat/month. Enterprise pricing available. At $99/seat, costs scale quickly for larger teams.\n\nFor early-stage teams or founders doing discovery research, Koji's economics are more favorable. For design-focused teams running usability studies on prototypes, Maze's free plan may be sufficient for getting started.\n\n## Final Verdict\n\nKoji and Maze are more complementary than competing. Maze measures *what* users do; Koji understands *why* they do it.\n\nIf you can only choose one and your primary research need is understanding customer problems, motivations, and unmet needs — Koji is your tool. You'll get conversational depth, automatic synthesis, and insights that actually drive strategy, not just design iterations.\n\nIf your primary need is validating whether a specific design works before shipping it, Maze is the better fit.\n\nThe highest-impact research programs use both: Koji at the front of the process to discover the right problems to solve, and usability testing tools like Maze at the back to validate the solutions. That combination produces products that are both right and usable — a combination that's surprisingly rare.\n\n---\n\n**Start discovering with Koji for free.** [Run your first AI-moderated interview study today](https://koji.so) — no scheduling, no moderation, no manual analysis required.\n\n*Last verified: March 2026*\n\n## Frequently Asked Questions\n\n**Is Koji a Maze alternative?**\nKoji and Maze serve different primary use cases. Maze is for quantitative usability testing on prototypes and designs. Koji is for qualitative depth interviews. Most teams benefit from both rather than choosing one over the other.\n\n**Does Maze do interviews like Koji?**\nMaze supports open-ended survey questions but is not designed for conversational depth interviews. Koji's AI conducts full voice or text conversations with natural follow-up questions — something Maze surveys cannot replicate.\n\n**Can Koji replace Maze for usability testing?**\nNo. Koji is not designed for task-based prototype testing, click maps, or heatmaps. If usability validation is your primary need, Maze (or a similar tool like Lookback or Userlytics) is a better fit.\n\n**Which tool is better for product managers?**\nKoji is particularly well-suited for product managers who need to understand customer needs, motivations, and jobs-to-be-done — especially without a dedicated research team. Maze is better suited for designers validating specific design decisions.\n\n**Does Koji have a free plan?**\nYes. Every new Koji account starts with 10 free credits and one active study, a meaningful starting point to run your first AI-moderated research study at no cost.\n","category":"Research","lastModified":"2026-10-30T08:12:58.731737+00:00","metaTitle":"Koji vs Maze 2026 — Koji Blog","metaDescription":"Koji vs Maze: honest 2026 comparison. Koji runs AI depth interviews; Maze runs quantitative usability tests. Learn which fits your research goals.","keywords":["koji vs maze","maze alternative","usability testing vs user interviews","product research tools","qualitative vs quantitative research","ai interview platform","maze pricing"],"aiSummary":"Koji and Maze serve different research needs: Koji conducts AI-moderated qualitative depth interviews and synthesizes themes automatically, while Maze runs quantitative usability tests on prototypes with click maps and task metrics. Most mature research teams use both — Koji for discovery, Maze for validation.","aiKeywords":["user research tools","maze","usability testing","qualitative research","product discovery","ai interviews","research platform comparison","prototype testing"],"aiContentType":"comparison","faqItems":[{"answer":"Koji and Maze serve different primary use cases. Maze is for quantitative usability testing on prototypes. Koji is for qualitative depth interviews. Most teams benefit from both rather than choosing one over the other.","question":"Is Koji a Maze alternative?"},{"answer":"Maze supports open-ended survey questions but is not designed for conversational depth interviews. Koji conducts full AI voice or text conversations with natural follow-up questions, which Maze surveys cannot replicate.","question":"Does Maze do interviews like Koji?"},{"answer":"No. Koji is not designed for task-based prototype testing, click maps, or heatmaps. If usability validation is your primary need, a dedicated tool like Maze is more appropriate.","question":"Can Koji replace Maze for usability testing?"},{"answer":"Koji is particularly well-suited for product managers running discovery interviews to understand customer motivations and jobs-to-be-done. Maze is better for designers validating specific design decisions quantitatively.","question":"Which tool is better for product managers?"},{"answer":"Yes. Every new Koji account starts with 10 free credits and one active study, giving teams a meaningful starting point for AI-moderated research at no cost.","question":"Does Koji have a free plan?"}],"relatedTopics":["User Research Tools","Research Platform Comparison","Usability Testing","Product Discovery","Qualitative vs Quantitative Research"]},{"type":"blog","id":"eeb4f9e6-5cb7-49bd-b030-ce237333eaca","slug":"product-manager-guide-customer-discovery-ai","title":"The Product Manager's Guide to Customer Discovery with AI (2026)","url":"https://www.koji.so/blog/product-manager-guide-customer-discovery-ai","summary":"Product managers face three core blockers to effective discovery: access to customers, time for moderation, and synthesis bottlenecks. This guide provides a 6-step framework for running AI-moderated customer discovery interviews — from defining learning objectives to connecting insights to roadmap decisions — showing how AI tools like Koji cut the process from weeks to days.","content":"\nMost product managers know they should talk to customers more. The problem isn't motivation — it's time. Scheduling 10 interviews, moderating each one, transcribing, coding, and synthesizing results takes weeks. By then, the sprint is over and the roadmap is already decided.\n\nAI-moderated research changes that equation entirely. Here's how to run faster, deeper customer discovery — even without a dedicated research team.\n\n## Why Customer Discovery Still Fails Most Product Teams\n\nThe statistics are sobering. According to a McKinsey survey, over 40% of companies never collect feedback directly from end users. The 280 Group found that 1 in 5 products fails to meet customer expectations — a direct consequence of insufficient discovery.\n\nThe structural problems are consistent across organizations:\n\n**Access to customers is gated.** In many companies, PMs don't have direct access to customers. Every interview requires coordination through sales, customer success, or legal — adding weeks of lead time before a single conversation happens.\n\n**Discovery is episodic, not continuous.** Most teams run discovery in a burst before a major feature, then stop. By the time the next cycle begins, customer context is stale.\n\n**The synthesis bottleneck is real.** Even after great interviews, someone needs to watch recordings, tag themes, and write up findings. Dscout's research found that analysis consumes 32.7% of total research project time — often falling on the PM who already has a full sprint plate.\n\n**Non-researchers are doing it anyway.** The Maze Future of User Research Report 2026 found that product managers now conduct 39% of research studies, overtaking dedicated researchers in many organizations. But most PMs have no formal training in qualitative research methods.\n\nThe result: discovery either doesn't happen, happens poorly, or happens too slowly to influence decisions.\n\n## What AI-Moderated Discovery Changes\n\nAI-moderated interviews remove the three biggest bottlenecks: scheduling, moderation, and analysis.\n\nInstead of coordinating calendars, sending a study link takes minutes. Instead of moderating each session yourself, an AI interviewer conducts the conversation — asking follow-up questions, probing for depth, maintaining consistent quality across every session. Instead of spending a weekend coding transcripts, automatic thematic analysis extracts key themes, sentiment, and representative quotes.\n\nThe result: a 5-interview discovery sprint that used to take 3 weeks now takes 3 days.\n\nAccording to the Maze 2026 report, AI cuts qualitative analysis time by up to 80%. That's not a marginal improvement — it's the difference between running continuous discovery and waiting for a research calendar slot.\n\n## A Practical Discovery Framework for PMs\n\nHere's a step-by-step framework for running effective customer discovery using AI interviews.\n\n### Step 1: Define Your Learning Objective\n\nBefore opening any research tool, write one sentence: *What decision will this research inform?*\n\nDiscovery research without a clear decision context produces interesting observations but rarely drives action. Good learning objectives look like:\n\n- \"Should we prioritize feature X or feature Y for Q3?\"\n- \"Why are users who sign up not completing onboarding?\"\n- \"What jobs do customers hire our product for that we don't currently support?\"\n\nA clear objective shapes every question in your guide — and tells you when you have enough data.\n\n### Step 2: Identify the Right Participants\n\nBetter discovery comes from better participant selection. Define your target segment precisely before recruiting:\n\n- What behavior should they have done? (e.g., \"churned in the last 90 days\", \"power user with 50+ sessions\")\n- What demographic or firmographic criteria matter?\n- What should they *not* be? (Avoid power users when studying onboarding; avoid churned users when validating new features)\n\nNarrow your segment, even if it means fewer responses. 8 highly-relevant participants teach you more than 30 generic ones. If the segment is in your own base, a churned cohort or a power-user list, import it so each person gets a personalized link and the report is already cut by that attribute. If it is not, a segment you do not own or a market you have no list for, open Panel recruit, describe the profile, and approve the price per completed interview in credits before you launch. Research by Jakob Nielsen found that 5 participants reveal 85% of usability problems — the principle extends to discovery too.\n\n### Step 3: Design Your Research Brief\n\nA research brief is your interview guide — the questions and topics your AI interviewer will follow. Good briefs share three characteristics:\n\n**They start open.** \"Walk me through the last time you tried to [task]\" beats \"How often do you use [feature]?\" Open questions reveal context you didn't know to ask about.\n\n**They probe for motivation, not behavior.** Behavioral questions tell you what customers do. Motivational questions tell you why — and why is what drives product decisions. Ask \"what made you look for a solution?\" not \"which features do you use?\"\n\n**They avoid leading the witness.** \"How frustrating was it when X happened?\" presupposes frustration. \"What was that experience like?\" lets the customer characterize it.\n\nWith Koji, the AI consultant helps you build this brief — asking clarifying questions about your research goals and suggesting question frameworks based on your study type.\n\n### Step 4: Run the Interviews\n\nWith a traditional approach, running 10 interviews requires 10–15 hours of moderation time, spread across 2–3 weeks of scheduling. Participant no-shows, rescheduling, and time zones compound the delay.\n\nWith AI-moderated interviews:\n- Send a shareable link via email, Slack, in-app notification, or NPS survey follow-up\n- Participants complete the interview at their own convenience — no scheduling friction\n- The AI conducts a 15–30 minute voice or text conversation, following your guide and probing naturally\n- You can run 10 or 100 interviews with identical effort\n\nFor discovery, aim for 8–15 interviews within a focused participant segment. This is enough to achieve thematic saturation — the point where new interviews stop revealing new themes.\n\n### Step 5: Analyze and Synthesize\n\nThis is where most discovery programs stall. Traditional analysis involves:\n1. Re-reading transcripts or rewatching recordings\n2. Manually tagging themes and patterns\n3. Writing up a synthesis document\n4. Creating a presentation for stakeholders\n\nWith Koji, analysis is automatic. After each interview, Koji extracts themes, sentiment, and key quotes. Once you've completed your study, generate a one-click report that synthesizes insights across all participants — structured by theme, supported by direct quotes, and ready to share.\n\nThe output is a research document your team can actually act on, produced in hours instead of days.\n\n### Step 6: Connect Insights to Decisions\n\nGreat research that doesn't change decisions is just expensive documentation. The final step is explicitly connecting your findings to the decision you defined in Step 1.\n\nStructure your insight readout as:\n\n- **What we learned** (top 3 themes with supporting quotes)\n- **What this means** (interpretation of the patterns)\n- **What we recommend** (the decision or next action)\n- **Confidence level** (how many interviews, how consistent the signal)\n\nThis format respects stakeholders' time and makes research actionable rather than archival.\n\n## Discovery Cadences That Work\n\n**Pre-sprint discovery (problem validation):** Before committing to a feature, run 5–8 targeted interviews to confirm the problem is real, understand its severity, and learn what customers have already tried. Takes 2–3 days with AI-moderated interviews.\n\n**Continuous always-on discovery:** Set up a persistent study that runs in the background — triggered by NPS survey completion, post-onboarding, or post-cancellation. Insights accumulate continuously without scheduled research cycles.\n\n**Win/loss interviews:** After a deal closes or churns, automatically invite the customer to a brief AI interview. Aggregate findings across 20–30 responses reveal patterns that no individual interview would surface.\n\n**Feature validation (pre-launch):** Before shipping, run 8–10 interviews with existing customers to validate that your framing, naming, and core value proposition match customer mental models.\n\n## Common Discovery Mistakes PMs Make\n\n**Interviewing the wrong people.** Don't ask happy customers about problems with onboarding. Don't ask churned users about new features. Match participant selection to the specific decision you're trying to inform.\n\n**Asking for solutions instead of problems.** \"Would you use a feature that does X?\" produces unreliable answers. Customers are notoriously bad at predicting their own behavior. \"Tell me how you currently handle X\" reveals the real problem space.\n\n**Stopping too early.** Five interviews might reveal broad patterns — but they won't surface the edge cases or secondary segments that change your understanding. Aim for thematic saturation, not a minimum response count.\n\n**Treating discovery as a one-time event.** The best product teams treat discovery as a continuous practice — a steady stream of customer conversations informing every sprint, not a research phase before each major launch.\n\n## How Koji Fits Into a PM's Workflow\n\nKoji is purpose-built for the way product managers actually work. You don't need research training, a UX research partner, or a calendar full of interviews.\n\nYou set the learning objective, describe your participant profile, and configure the research guide. Koji's AI consultant helps you design the study. Once live, the AI conducts every interview — voice or text — and synthesizes results automatically.\n\nThe output is a structured insights report you can share with your team, reference in roadmap discussions, and archive for future context. From first session to final report, a discovery study that used to take a month can be complete in a week.\n\nWith over 40% of companies failing to collect end-user feedback (McKinsey), and 69% of research teams now using AI tools (Maze 2026), the competitive advantage goes to the teams that can learn fastest. AI-moderated discovery is how you build that advantage without scaling your research headcount.\n\n---\n\n**Ready to run your first discovery study?** [Start free on Koji](https://koji.so) — set up a study in minutes, let the AI interview your customers, and have insights the same week.\n\n## Frequently Asked Questions\n\n**How many customer interviews do I need for discovery?**\nFor most discovery research, 8–15 interviews within a focused participant segment is enough to reach thematic saturation — the point where new interviews stop revealing new themes. For broader studies spanning multiple segments, you may need 15–25 total.\n\n**Can product managers run research without a UX researcher?**\nYes. The Maze 2026 report found that PMs now conduct 39% of research studies. With AI-moderated tools like Koji, you get guided study design and automatic analysis — reducing the research expertise barrier significantly.\n\n**How long does an AI-moderated discovery interview take?**\nA typical Koji interview runs 15–30 minutes. Participants complete it asynchronously at their convenience, which significantly improves completion rates compared to scheduled sessions.\n\n**What is the difference between customer discovery and usability testing?**\nCustomer discovery focuses on understanding problems, motivations, and jobs-to-be-done — the \"why\" behind user behavior. Usability testing focuses on whether users can complete specific tasks in your product — the \"what\" of user behavior. Both are valuable; discovery should come first.\n\n**How do I recruit participants for discovery interviews?**\nKoji recruits for you: open Panel recruit, describe the segment, and Koji finds screened participants across 130+ countries at a price per completed interview in credits you approve first, on paid plans. For your existing customers, import a list so each person gets a personalized link, follow up on NPS surveys with the generic link, or embed a study link in your product. If you prefer to run recruitment yourself, tools like User Interviews, Respondent.io, or Prolific can connect you with screened participants in your target segment; share your Koji link with the people they find and the interviews land in the same study.\n","category":"Tutorial","lastModified":"2026-10-30T08:12:57.641994+00:00","metaTitle":"PM Guide to Customer Discovery with AI — Koji Blog","metaDescription":"A practical guide for product managers: how to run faster, deeper customer discovery using AI interviews. Framework, cadences, and common mistakes covered.","keywords":["product manager customer discovery","customer discovery interviews","pm user research","ai customer interviews","product discovery framework","customer research for product managers","continuous discovery"],"aiSummary":"Product managers face three core blockers to effective discovery: access to customers, time for moderation, and synthesis bottlenecks. This guide provides a 6-step framework for running AI-moderated customer discovery interviews — from defining learning objectives to connecting insights to roadmap decisions — showing how AI tools like Koji cut the process from weeks to days.","aiKeywords":["product management","customer discovery","user research","ai interviews","product strategy","continuous discovery","research framework","qualitative research"],"aiContentType":"tutorial","faqItems":[{"answer":"For most discovery research, 8-15 interviews within a focused participant segment is enough to reach thematic saturation. For studies spanning multiple segments, aim for 15-25 total.","question":"How many customer interviews do I need for discovery?"},{"answer":"Yes. The Maze 2026 report found that PMs now conduct 39% of research studies. AI-moderated tools like Koji provide guided study design and automatic analysis, reducing the research expertise barrier significantly.","question":"Can product managers run research without a UX researcher?"},{"answer":"A typical Koji interview runs 15-30 minutes. Participants complete it asynchronously at their convenience, which significantly improves completion rates compared to scheduled sessions.","question":"How long does an AI-moderated discovery interview take?"},{"answer":"Customer discovery focuses on understanding problems, motivations, and jobs-to-be-done — the why behind user behavior. Usability testing focuses on whether users can complete specific tasks — the what. Discovery should come before usability testing.","question":"What is the difference between customer discovery and usability testing?"},{"answer":"Koji recruits for you: open Panel recruit, describe the segment, and approve a per-respondent quote in credits before launch. For existing customers, import a list for personalized links, follow up on NPS surveys with the generic link, or embed the study in your product. If you prefer to run recruitment yourself, tools like User Interviews, Respondent.io, or Prolific connect you with screened participants in your target segment; share your Koji link with them and the interviews land in the same study.","question":"How do I recruit participants for discovery interviews?"}],"relatedTopics":["Product Management","Customer Discovery","User Research for PMs","AI-Powered Research","Continuous Discovery","Research Operations"]},{"type":"blog","id":"5c0119b7-a677-4d42-8e0c-f37850078ed3","slug":"how-to-analyze-user-interview-data","title":"How to Analyze User Interview Data: A Complete Guide (2026)","url":"https://www.koji.so/blog/how-to-analyze-user-interview-data","summary":"Analyzing user interview data involves five steps: preparing raw transcripts, coding passages with descriptive tags, grouping codes into themes, generating actionable insights (observation plus interpretation plus implication), and sharing findings with the team. AI-native tools like Koji automate the coding and theming steps, reducing analysis time by up to 80% while preserving the human judgment needed for insight generation.","content":"\nYou ran the interviews. Participants shared honest, nuanced perspectives. Now you're staring at a folder of transcripts, voice recordings, or notes — and the insights you need are somewhere in there.\n\nAnalysis is where most qualitative research programs break down. According to Dscout's research on UX timelines, analysis consumes 32.7% of total project time, and 51.6% of researchers say it's the phase they most wish they had more time for. It's also the phase most vulnerable to being rushed, abbreviated, or skipped entirely when sprints get tight.\n\nThis guide walks you through a proven analysis process — from raw transcript to actionable insight — including how AI is compressing this timeline without sacrificing depth.\n\n## What Makes Interview Analysis Hard\n\nUnlike quantitative data, qualitative interview data doesn't reveal patterns automatically. There's no average. There's no statistical significance test. Meaning emerges from reading, interpretation, and synthesis across multiple conversations.\n\nThe challenges compound quickly:\n\n- **Volume.** A 30-minute interview produces 4,000–6,000 words of transcript. Ten interviews produce a small novel.\n- **Noise.** Participants go off-topic, repeat themselves, and contradict what they said five minutes earlier. Filtering signal from noise requires judgment.\n- **Subjectivity.** Two analysts reading the same transcript can reasonably extract different themes. Without a structured process, analysis drifts toward confirming what you already believed.\n- **Time pressure.** The insights are urgently needed, but thorough analysis resists being rushed.\n\nA structured process solves all of these. Here's how to work through it.\n\n## Step 1: Prepare Your Raw Data\n\nBefore analyzing anything, get your data into a consistent, searchable format.\n\n**Transcribe every session.** If you recorded audio or video, generate a transcript. Modern AI transcription (including tools built into research platforms) is fast and accurate enough for analysis purposes — you don't need perfect transcripts, just readable ones.\n\n**Clean and organize.** Create a consistent file naming system: `participant-01-segment-churn-date.txt`. Organize by participant segment if you're studying multiple groups.\n\n**Add participant context.** Note key participant attributes alongside each transcript: role, company size, tenure as a customer, the specific behavior that qualified them for the study. This context becomes important when you're identifying whether themes are segment-specific. On Koji this happens at recruitment: import your list so each person gets a personalised link carrying role, plan or tenure and the report is segmented by those attributes automatically; share the generic anonymous link when candour matters more than tracking; or, for people you do not have, open Panel recruit and Koji finds them across 130+ countries at a price you approve first, with their screening answers attached to each transcript.\n\n**Do a quick read-through.** Before coding anything, read all transcripts once without annotating. This gives you a gestalt sense of the data before you start breaking it apart. Note any immediately striking moments in the margin — they're often significant.\n\n## Step 2: Code the Data\n\nCoding is the process of labeling segments of text with tags that describe what that passage is about. Codes are the raw material of themes.\n\n### Two Approaches to Coding\n\n**Deductive coding** starts with a predetermined codebook — a list of categories you expect to find, based on your research questions. You apply these codes as you read, and add new codes when something doesn't fit.\n\nDeductive coding is faster and more consistent, but risks missing unexpected insights that fall outside your original framework.\n\n**Inductive coding** starts with no predetermined categories. You read each passage and invent a code that describes it, building your codebook from the data. When you've coded everything, similar codes get grouped and consolidated.\n\nInductive coding surfaces unexpected patterns but requires more time and iteration.\n\n**For discovery research**, inductive coding is typically more valuable — you're trying to learn what you don't already know. **For evaluative research** (testing whether a specific hypothesis is true), deductive coding is more efficient.\n\n### Practical Coding Tips\n\n- Code at the level of meaning, not grammar. A sentence might contain two separate ideas; code them separately.\n- Use verbs for process codes (\"struggling to find\", \"workarounds for\"), nouns for concept codes (\"mental model\", \"trigger event\"), and quotes for exact phrasing that's particularly vivid or revealing.\n- Don't over-code. If you end up with 80 codes for 10 interviews, you have fragments, not insight. Aim for 15–30 meaningful codes that cluster naturally.\n- Note participant attributions on each code. \"5 of 10 participants mentioned this\" is meaningfully different from \"1 of 10 did.\"\n\n## Step 3: Identify Themes\n\nThemes are patterns — codes that cluster together because they're describing the same underlying phenomenon.\n\nMove your codes into an affinity map: a visual grouping where related codes live near each other. Digital tools like Miro, FigJam, or a simple spreadsheet work well. Physical sticky notes on a wall work even better for teams who can gather in person.\n\nAs you group, look for:\n\n**Frequency.** How many participants expressed this? High-frequency patterns are more likely to represent genuine customer needs rather than individual preferences.\n\n**Intensity.** Did participants volunteer this unprompted, or only answer when directly asked? Unprompted themes often represent stronger motivations.\n\n**Surprise.** What themes contradicted your expectations? These are often the most strategically valuable, because they represent blind spots in your current product thinking.\n\n**Sequence.** Do certain themes cluster around specific moments in a customer journey (trigger, evaluation, adoption, abandonment)? Journey-organized themes produce clearer recommendations.\n\nName each theme with a concrete noun phrase that captures its essence — not \"issue with onboarding\" but \"unclear value proposition before first success moment.\" Specific names force sharper thinking.\n\n## Step 4: Generate Insights\n\nA theme is an observation. An insight is a theme plus an interpretation — *why* it matters and *what* it means for your product.\n\nThe insight generation step is where analysis moves from descriptive to strategic. For each major theme:\n\n1. **State the observation:** \"7 of 10 participants described abandoning the setup flow at the integrations step.\"\n2. **Interpret the why:** \"They didn't understand which integration was required vs. optional, and feared breaking existing workflows.\"\n3. **State the implication:** \"The integrations step needs clearer optionality signaling and a 'skip for now' path, or it will continue to block activation.\"\n\nThis three-part structure — observation, interpretation, implication — keeps insights actionable rather than descriptive.\n\nA good insight document for a 10-interview study should have 3–6 major insights (not 20). If you have 20 insights, you have observations, not insights. Group and consolidate until you have the smallest number of themes that explain the most about your participants' behavior.\n\n## Step 5: Structure and Share Your Findings\n\nResearch that isn't shared doesn't change products. Structure your findings for your audience:\n\n**For your immediate team (PM, design, engineering):** A concise document organized by insight, with supporting quotes, a recommendation for each, and a confidence rating (based on participant count and consistency).\n\n**For senior stakeholders:** A one-page summary — top 3 insights, what they mean for the roadmap, and one recommended decision.\n\n**For the research archive:** The full analysis including method, participant profiles, codebook, themes, insights, and raw quotes. Future researchers will thank you.\n\nThe most effective research readouts follow the structure: *We learned X. This means Y. We recommend Z.* Every slide or paragraph that doesn't fit this structure should be cut or moved to an appendix.\n\n## How AI Is Transforming Interview Analysis\n\nThe traditional analysis process — transcribe, code, theme, synthesize — is thorough but slow. For teams doing continuous research, the manual overhead becomes a bottleneck that limits how much they can actually learn.\n\nAI-native research platforms like Koji are changing this by automating the most time-consuming steps. When participants complete interviews through Koji, the AI extracts themes, sentiment, and key quotes automatically — the equivalent of Steps 2 and 3 in the manual process, done in seconds rather than days.\n\nThe Maze Future of User Research Report 2026 found that 69% of research teams now use AI in their workflows, citing faster turnaround (63%), improved efficiency (60%), and better workflows (56%) as the primary benefits. AI cuts qualitative analysis time by up to 80% according to multiple industry sources.\n\nThis doesn't eliminate the insight generation step — human judgment is still required to interpret what themes mean and translate them into product decisions. But it removes the mechanical work that makes analysis feel like a tax on curiosity.\n\n**What AI-automated analysis looks like in practice:**\n\n- Each interview generates an immediate summary with key themes and notable quotes\n- After collecting 10+ responses, a one-click aggregate report synthesizes patterns across all participants\n- Sentiment tracking shows which topics generated strong emotional reactions\n- Cross-participant theme frequency is calculated automatically\n\nFor teams running discovery, churn analysis, or feature validation research, this means insights can inform the sprint that's currently in progress — not the one starting in six weeks.\n\n## Common Analysis Mistakes to Avoid\n\n**Stopping at description.** \"Users said they found onboarding confusing\" is not an insight — it's an observation. The insight is what specific aspect confused them, why, and what that means for your design or copy.\n\n**Confirmation bias.** It's easy to notice the quotes that support what you already believe and overlook the ones that complicate it. A structured coding process, ideally involving two independent coders, protects against this.\n\n**Over-representing memorable quotes.** A single vivid quote from one participant can dominate a readout and create a misleading impression. Always note participant count next to every theme.\n\n**Analysis paralysis.** Perfect synthesis delivered too late is less valuable than good synthesis delivered on time. When you can articulate 3–5 clear insights with supporting evidence, you're ready to share — even if some nuance is still unresolved.\n\n**Siloing findings.** Analysis stored in a personal folder or a document that gets emailed once is effectively lost. Research findings need to live somewhere the team can find them when making future decisions.\n\n## Key Takeaways\n\n- Qualitative analysis has a proven structure: prepare data, code, theme, generate insights, share\n- Inductive coding produces more discovery; deductive coding is more efficient for evaluation\n- The goal is insights (observation + interpretation + implication), not observations\n- AI automates the mechanical steps (coding, theming, sentiment), freeing analysts for interpretation\n- Structured sharing ensures research influences decisions rather than accumulating in a folder\n\n---\n\n**Run your next study with automatic analysis built in.** [Koji](https://koji.so) conducts the interviews and extracts themes automatically — so you spend your time on insights, not transcription.\n\n## Frequently Asked Questions\n\n**How long does it take to analyze user interview data?**\nManual analysis of 10 interviews typically takes 8–20 hours depending on interview length and analysis depth. Dscout research found analysis accounts for 32.7% of total project time. AI-powered analysis tools can reduce this by up to 80%.\n\n**What is thematic analysis in user research?**\nThematic analysis is a method for identifying, analyzing, and reporting patterns (themes) within qualitative data. It involves coding text segments, grouping related codes into themes, and interpreting what those themes reveal about participant experiences and motivations.\n\n**How many interviews do you need before analyzing?**\nYou can begin preliminary analysis after 5 interviews to check if your research guide is working. For final synthesis, most researchers recommend analyzing after 8–15 interviews within a focused participant segment, or when you reach thematic saturation — the point where new interviews stop producing new themes.\n\n**What is the difference between a theme and an insight?**\nA theme is an observed pattern in the data: \"Users struggled with the onboarding flow.\" An insight adds interpretation: \"Users abandoned onboarding because they couldn't understand which steps were required vs. optional, creating anxiety about breaking their existing setup.\" Insights drive decisions; themes describe what happened.\n\n**Can AI replace human analysis of user interviews?**\nAI automates the mechanical parts of analysis — transcription, initial coding, theme identification, and sentiment tagging. The interpretive work — understanding *why* a theme exists and *what it means* for your product — still requires human judgment. The best modern workflow combines AI automation with human interpretation.\n","category":"Tutorial","lastModified":"2026-10-30T08:12:56.607216+00:00","metaTitle":"How to Analyze User Interview Data 2026 — Koji Blog","metaDescription":"A complete 2026 guide to analyzing user interview data: coding, thematic analysis, insight generation, and how AI cuts analysis time by 80%.","keywords":["how to analyze user interviews","user interview analysis","thematic analysis user research","qualitative data analysis","interview coding","research analysis methods","user research insights"],"aiSummary":"Analyzing user interview data involves five steps: preparing raw transcripts, coding passages with descriptive tags, grouping codes into themes, generating actionable insights (observation plus interpretation plus implication), and sharing findings with the team. AI-native tools like Koji automate the coding and theming steps, reducing analysis time by up to 80% while preserving the human judgment needed for insight generation.","aiKeywords":["qualitative research","thematic analysis","interview analysis","research methods","user research","coding","insights","research synthesis"],"aiContentType":"tutorial","faqItems":[{"answer":"Manual analysis of 10 interviews typically takes 8-20 hours. Dscout research found analysis accounts for 32.7% of total project time. AI-powered tools can reduce this by up to 80%.","question":"How long does it take to analyze user interview data?"},{"answer":"Thematic analysis is a method for identifying patterns (themes) within qualitative data. It involves coding text segments, grouping related codes into themes, and interpreting what those themes reveal about participant motivations and experiences.","question":"What is thematic analysis in user research?"},{"answer":"Begin preliminary analysis after 5 interviews to check your guide is working. For final synthesis, 8-15 interviews within a focused segment is typically sufficient to reach thematic saturation.","question":"How many interviews do you need before analyzing?"},{"answer":"A theme is an observed pattern: users struggled with onboarding. An insight adds interpretation: users abandoned because they could not tell required from optional steps, creating anxiety. Insights drive decisions; themes describe observations.","question":"What is the difference between a theme and an insight?"},{"answer":"AI automates the mechanical parts — transcription, coding, theme identification, sentiment tagging. The interpretive work of understanding why themes exist and what they mean for your product still requires human judgment. The best workflow combines both.","question":"Can AI replace human analysis of user interviews?"}],"relatedTopics":["User Research Methods","Qualitative Analysis","Research Operations","Interview Techniques","Product Discovery","AI-Powered Research"]},{"type":"blog","id":"dcce63ff-4354-474e-b236-3cce58d1d9a3","slug":"how-to-analyze-customer-interview-data","title":"How to Analyze Customer Interview Data: A Complete Guide","url":"https://www.koji.so/blog/how-to-analyze-customer-interview-data","summary":"Analyzing customer interview data involves five steps: preparing transcripts, open coding (labeling meaningful chunks), axial coding (grouping into themes), writing insight statements, and prioritizing findings. AI tools like Koji reduce qualitative analysis time by 15x by automating transcription, theme detection, and sentiment analysis, while human judgment remains essential for prioritization and strategic framing.","content":"\nYou ran the interviews. Now you're staring at hours of transcripts, recordings, and notes — and you need to turn all of it into something actionable. Here's a complete process for analyzing customer interview data, from raw transcripts to insights your team will actually use.\n\n## Why Analysis Is the Make-or-Break Step\n\nMost teams get the research right. They write good questions, recruit the right participants, and conduct thoughtful interviews. Then the analysis bogs them down.\n\nAccording to a **Dscout study of 300+ UX researchers**, the most common synthesis timeframe is 1–5 days — but 35% of teams regularly go over that, with complex studies stretching to weeks. Meanwhile, **more than half of research data ends up unusable** (Maze, Future of User Research 2026) — not because the interviews were bad, but because the analysis process couldn't keep up.\n\nThe good news: there are systematic approaches that transform interview data into clear, compelling insights — and AI tools now make this process dramatically faster.\n\n## Step 1: Prepare Your Data\n\nBefore you analyze anything, get your transcripts into a usable state.\n\n**For voice or audio interviews:**\n- Transcribe all recordings (AI transcription tools like Otter.ai or auto-transcription in your research platform)\n- Review transcripts for accuracy — AI transcription is good but not perfect\n- Add speaker labels if they're missing\n\n**For text interviews:**\n- Export from your research platform or copy from your notes\n- Standardize the format: question → answer, consistently structured\n\n**Organize everything in one place.** Don't analyze across scattered Slack messages, Google Docs, and Notion pages. Use a research platform or at minimum a single spreadsheet.\n\n**Pro tip:** If you're using Koji for AI-moderated interviews, all transcripts are already organized, timestamped, and ready for analysis — no manual transcription needed.\n\n## Step 2: Read Everything Before You Code\n\nBefore labeling or tagging anything, read through all your transcripts once. Don't annotate yet — just absorb.\n\nThis first pass serves two purposes:\n1. You build an intuitive sense of what participants talked about most\n2. You identify 3–5 big themes before imposing a taxonomy\n\nTeams that skip this step end up with overcrowded coding schemes that don't reflect what participants actually said.\n\n## Step 3: Open Coding — Surface Everything\n\nOpen coding means reading through transcripts and applying descriptive labels to every meaningful chunk of text.\n\n**A chunk is meaningful if:**\n- It describes a problem, pain point, or frustration\n- It expresses a motivation, goal, or aspiration\n- It describes a behavior or current workflow\n- It contains a reaction, opinion, or sentiment\n\nAt this stage, create codes freely. Don't worry about being too granular. Examples:\n- \"frustrated by long onboarding\"\n- \"uses competitor because of pricing\"\n- \"didn't realize feature existed\"\n- \"always involves team lead in decision\"\n\nKeep codes descriptive, not interpretive. \"Didn't know about X feature\" is better than \"needs better UX\" — you can interpret later.\n\n**Research from PMC (published in multiple peer-reviewed studies)** found that qualitative code saturation is typically reached after 9 interviews, but meaning saturation — the point where you fully understand each theme — takes 16–24 interviews. This is why sample size matters for depth.\n\n## Step 4: Axial Coding — Find the Patterns\n\nNow group your open codes into categories. This is where insight emerges.\n\nLook for:\n- **Frequency**: Which codes appear in 4 out of 5 interviews? Those are strong signals.\n- **Intensity**: Which quotes carried the most emotional weight? Those reveal problems that really matter.\n- **Contradictions**: Where did participants disagree? Understanding the conditions for each perspective is often the most valuable insight.\n\nCreate an affinity map — digital (Miro, FigJam) or physical (sticky notes on a wall). Group open codes into 5–8 main categories. Give each category a descriptive label.\n\nExample groupings:\n- Open codes: \"manual exports\", \"copy-pasting into sheets\", \"hates the reporting step\" → Category: **Painful post-interview workflow**\n- Open codes: \"team doesn't see research\", \"findings ignored\", \"presented findings once, nothing happened\" → Category: **Research not influencing decisions**\n\n## Step 5: Write Insight Statements\n\nCategories aren't insights yet. An insight is a conclusion that drives a decision.\n\n**Category** → \"Painful post-interview workflow\"\n\n**Insight** → \"Researchers spend 30–60% of their project time on post-interview busywork — transcription, cleaning, formatting — leaving little time for actual synthesis. This is why research reports arrive late and feel rushed.\"\n\nA well-formed insight statement:\n- Names a specific behavior or situation\n- Explains the underlying cause or context\n- Suggests what it means for your product or team\n\nEvery insight should be traceable back to at least 3 quotes from different participants. If you can only find 1 quote supporting an insight, it may be an outlier — note it separately.\n\n## Step 6: Prioritize What Matters Most\n\nNot all insights are equally actionable. Use a simple 2×2 matrix to sort:\n- **High frequency + high severity** = Fix this now\n- **High frequency + low severity** = Worth addressing at scale\n- **Low frequency + high severity** = Investigate with more research\n- **Low frequency + low severity** = Deprioritize\n\nShare this prioritization with your team before writing the report. Alignment on priorities prevents the common situation where a researcher presents 12 findings and the team argues about which 3 matter.\n\n## Step 7: Write the Report\n\nA research report should have one job: make it easy for your reader to act on the findings.\n\n**Report structure:**\n1. **TL;DR** (3–5 bullet points): Key findings and recommended actions\n2. **Research context**: Who you talked to, how many, what you were trying to learn\n3. **Findings by theme**: Each theme with supporting quotes and the insight statement\n4. **Implications**: What this means for product, marketing, or strategy\n5. **Open questions**: What you still don't know and how to find out\n\nKeep it under 10 slides or 1,500 words. Executives don't read appendices. Put the key evidence where they'll see it.\n\n**Pro tip:** Koji generates a complete research report automatically after your interviews complete — including themes, sentiment analysis, representative quotes, and an executive summary. It's a strong starting point you can edit and customize.\n\n## How AI Changes the Analysis Process\n\nManual analysis is a bottleneck. According to the **Lumivero State of AI in Qualitative Research**, AI categorization tools complete qualitative coding tasks **15x faster** than human coders — with higher consistency.\n\nAccording to the **User Interviews State of User Research 2025**, **62% of researchers now use AI for analyzing research data**, and **56% report improved team efficiency** as a result.\n\nWhat AI is good at:\n- Transcription and cleaning\n- Identifying recurring themes across large datasets\n- Sentiment analysis at scale\n- Surfacing representative quotes for each theme\n- First-draft insight summaries\n\nWhat still benefits from human judgment:\n- Distinguishing important contradictions from outliers\n- Translating findings into strategic implications\n- Assessing severity and urgency of findings\n- Knowing what to leave out of the report\n\nThe best approach: use AI to do the heavy lifting on synthesis, then apply your judgment to prioritization and framing.\n\n## The Koji Approach: Analysis Built In\n\nTraditional research workflows require researchers to export data, import it into a separate analysis tool, manually tag codes, and then write findings from scratch. The entire process can take days.\n\nKoji's AI interviews arrive pre-analyzed. When your study completes, Koji has already:\n- Transcribed and cleaned all conversations\n- Identified recurring themes across all participants\n- Tagged sentiment (positive, neutral, negative) by topic\n- Surfaced representative quotes per theme\n- Generated a draft research brief\n\nRecruitment is built in too: reach your own people with personalised links when you want to track who answered (the attributes you import travel with the link, so the report is already segmented), a generic anonymous link when candour matters, or let Koji recruit an audience you do not have across 130+ countries at a price you approve first.\n\nTeams using Koji report going from interview completion to shareable insights in hours, not days — a pipeline that traditionally took weeks.\n\n## Common Analysis Mistakes to Avoid\n\n**Confirming what you already believe.** When you have a hypothesis, it's easy to find quotes that support it and ignore quotes that challenge it. Force yourself to actively look for disconfirming evidence.\n\n**Over-coding.** 50 codes from 10 interviews is usually too many. Aim for 10–20 high-level codes in the open coding phase.\n\n**Skipping the prioritization step.** A report with 12 equal-weight findings is harder to act on than one with 3 clear priorities and 9 supporting observations.\n\n**Analyzing in isolation.** Bring a colleague into the affinity mapping session. Two people catch blind spots one person misses.\n\n**Writing for completeness instead of clarity.** Include everything in your notes; include only what matters in your report.\n\n## Key Takeaways\n\n- Read all transcripts before coding — don't skip the immersion phase\n- Open coding surfaces raw observations; axial coding finds patterns\n- An insight is more than a theme — it explains the \"why\" behind the behavior\n- According to Lumivero, AI reduces qualitative analysis time by 15x, but human judgment still drives prioritization\n- Prioritize findings before writing the report, not after\n- A good research report is under 1,500 words with a TL;DR upfront\n\n---\n\n**Running customer interviews and want analysis built in?** [Try Koji free](https://www.koji.so) — AI interviews with automatic theme detection, sentiment analysis, and report generation included.\n\n## Frequently Asked Questions\n\n**How long does it take to analyze customer interview data?**\nManual analysis typically takes 1–5 days for a study of 5–15 interviews. AI-assisted analysis can compress this to hours. The synthesis step — writing insights from patterns — remains the most time-intensive part.\n\n**What is the best tool for analyzing customer interviews?**\nFor teams using Koji, analysis is automatic — themes, sentiment, and insights are generated immediately after each interview. For manual workflows, tools like Dovetail, Notably, and Marvin help organize and tag transcripts. Affinity mapping tools like Miro work well for collaborative synthesis sessions.\n\n**How many quotes do I need to support an insight?**\nAim for at least 3 quotes from different participants for each insight you include in your report. Single-participant observations are worth noting but should be flagged as exploratory, not confirmatory.\n\n**What is the difference between a theme and an insight?**\nA theme is a pattern: \"participants mentioned difficulty with onboarding.\" An insight is an interpretation with implication: \"New users abandon during onboarding because they cannot connect their existing data in step 3 — fixing this step could reduce churn in the first week.\" Insights drive decisions; themes organize data.\n\n**Can AI fully replace human analysis of customer interviews?**\nNot entirely. AI excels at speed, consistency, and pattern-matching at scale. Humans are better at interpreting nuance, understanding emotional weight, and translating findings into strategic recommendations. The best analysis uses both.\n","category":"Tutorial","lastModified":"2026-10-30T08:12:55.51601+00:00","metaTitle":"How to Analyze Customer Interview Data — Koji Blog","metaDescription":"A step-by-step guide to analyzing customer interview data: open coding, affinity mapping, writing insight statements, and turning findings into decisions.","keywords":["how to analyze customer interviews","customer interview analysis","qualitative data analysis","user interview analysis","affinity mapping","research synthesis","thematic analysis"],"aiSummary":"Analyzing customer interview data involves five steps: preparing transcripts, open coding (labeling meaningful chunks), axial coding (grouping into themes), writing insight statements, and prioritizing findings. AI tools like Koji reduce qualitative analysis time by 15x by automating transcription, theme detection, and sentiment analysis, while human judgment remains essential for prioritization and strategic framing.","aiKeywords":["customer interview analysis","qualitative research","thematic analysis","affinity mapping","research synthesis","AI research tools","user research methods","open coding"],"aiContentType":"tutorial","faqItems":[{"answer":"Manual analysis typically takes 1–5 days for a study of 5–15 interviews. AI-assisted analysis can compress this to hours. The synthesis step remains the most time-intensive part.","question":"How long does it take to analyze customer interview data?"},{"answer":"For teams using Koji, analysis is automatic — themes, sentiment, and insights are generated immediately after each interview. For manual workflows, tools like Dovetail, Notably, and Marvin help organize and tag transcripts.","question":"What is the best tool for analyzing customer interviews?"},{"answer":"Aim for at least 3 quotes from different participants for each insight. Single-participant observations should be flagged as exploratory, not confirmatory.","question":"How many quotes do I need to support an insight?"},{"answer":"A theme is a pattern: participants mentioned difficulty with onboarding. An insight is an interpretation with implication: new users abandon during onboarding because they cannot complete step 3 — fixing this step could reduce first-week churn.","question":"What is the difference between a theme and an insight?"},{"answer":"Not entirely. AI excels at speed and pattern-matching at scale. Humans are better at interpreting nuance and translating findings into strategic recommendations. The best analysis uses both.","question":"Can AI fully replace human analysis of customer interviews?"}],"relatedTopics":["user research","qualitative research","product discovery","customer insights","research operations","UX research methods"]},{"type":"blog","id":"6a5f230f-7020-42e5-a11a-a9f0e12e03c0","slug":"how-to-write-user-interview-questions","title":"How to Write User Interview Questions That Get Real Answers","url":"https://www.koji.so/blog/how-to-write-user-interview-questions","summary":"Writing effective user interview questions requires asking about past experience rather than hypothetical opinions. Use grand tour questions to open conversations, specific instance questions to drill into moments, and neutral probing questions to go deeper. Limit sessions to 5–8 questions to allow depth over breadth. AI-moderated platforms like Koji apply these principles automatically with consistent follow-up probing across all participants.","content":"\nMost interview questions are too narrow, too leading, or too hypothetical. Researchers spend weeks planning a study, recruit carefully, and then ask questions that produce polished, surface-level answers instead of the honest truth. Here is how to write questions that actually work.\n\n## Why Question Quality Matters More Than Anything Else\n\nThe quality of your interview questions determines the quality of your insights. You can recruit the perfect participants, run a flawless session, and still walk away with nothing useful if your questions were the wrong ones.\n\nAccording to **Nielsen Norman Group**, the most common failure mode in user interviews is the researcher inadvertently leading the participant — through word choice, framing, or the order of questions — toward confirming what the researcher already believes.\n\nA 2024 study from Glaut found that AI-moderated interviews produced **129% more words per response** than traditional static surveys. The difference was not the format — it was the quality of follow-up probing. Better questions, asked at the right moment, produce richer data.\n\n## The Core Principle: Ask About Experience, Not Opinion\n\nThe single most important rule in user interview question design: **ask about what people did, not what they think they would do.**\n\nOpinions and hypotheticals are unreliable. People are notoriously bad at predicting their own future behavior. But memories of past experience, while imperfect, are far more grounded in reality.\n\n**Instead of:** \"Would you pay for a feature that did X?\"\n\n**Ask:** \"Walk me through the last time you tried to do X. What did you do?\"\n\n**Instead of:** \"How important is Y to you?\"\n\n**Ask:** \"Tell me about a time when Y caused a problem for you. What happened?\"\n\nThis principle — popularized by the Jobs-to-be-Done framework and the customer discovery work of Steve Blank — consistently produces more actionable insights than opinion-based questioning.\n\n## Question Types That Work\n\n### 1. Grand Tour Questions (Start Here)\n\nGrand tour questions invite participants to walk you through their experience in their own words. They are low-pressure, open-ended, and reveal structure you didn't know to ask about.\n\n**Examples:**\n- \"Walk me through how you currently handle [problem area].\"\n- \"Tell me about the last time you had to [do the thing you're researching].\"\n- \"Describe a typical week when it comes to [topic].\"\n\nThese questions are your opening move. Let the participant talk for 3–5 minutes uninterrupted. You'll often get your most valuable insights here, before you ask anything specific.\n\n### 2. Specific Instance Questions\n\nAfter a grand tour, drill into a specific moment. Vague memories produce vague answers. Specific memories produce specific, actionable data.\n\n**Examples:**\n- \"You mentioned that happens every week — can you tell me about a specific time it happened recently?\"\n- \"Think about the last time you [did X]. What triggered it?\"\n- \"Walk me through exactly what you did step by step.\"\n\nThe more specific the episode, the more you can trust the data.\n\n### 3. Probing Questions\n\nProbing questions are your follow-up moves. They go deeper without leading the participant.\n\n**The most powerful probing questions:**\n- \"Tell me more about that.\"\n- \"What happened next?\"\n- \"How did that make you feel?\"\n- \"Why did you do it that way and not another way?\"\n- \"What were you expecting to happen instead?\"\n\nNote what these questions have in common: they're short, neutral, and create space for the participant to go wherever they need to go. Avoid suggesting the answer in your probe.\n\n### 4. Native Language Questions\n\nAsk participants what words they use, not the words you use. This matters enormously for product copy, marketing, and feature naming.\n\n**Examples:**\n- \"How would you describe this to a colleague?\"\n- \"If you were searching for a solution to this, what would you type into Google?\"\n- \"What do you call this step in your workflow?\"\n\nThe answers often reveal that your internal terminology is completely different from how customers think about the problem.\n\n## Question Types That Don't Work\n\n### Leading Questions\n\nLeading questions embed the answer in the question itself.\n\n**Avoid:** \"Don't you find it frustrating when you can't find your research notes?\"\n\n**Use instead:** \"Tell me about a time you needed to find something from a past research session.\"\n\n### Double-Barreled Questions\n\nDouble-barreled questions ask two things at once, which means participants answer only one — and you don't know which.\n\n**Avoid:** \"How do you currently handle analysis and how long does it take?\"\n\n**Use instead:** \"How do you currently handle analysis? [Wait for answer.] And how long does that typically take?\"\n\n### Hypothetical Questions\n\nHypothetical questions invite speculation, not experience.\n\n**Avoid:** \"If we added X feature, would that help you?\"\n\n**Use instead:** \"Have you ever tried to do X? What happened?\"\n\n### Yes/No Questions\n\nYes/no questions close down conversation when you need it to open up.\n\n**Avoid:** \"Do you find the current process frustrating?\"\n\n**Use instead:** \"What has your experience been with the current process?\"\n\n## How to Structure a 45-Minute Interview\n\nA well-structured interview moves from broad to specific, building psychological safety before asking hard questions.\n\n**0–5 minutes: Warm-up**\nExplain the purpose, confirm recording consent, set expectations. Ask a non-threatening opening question: \"Tell me a bit about your role and what you work on day to day.\"\n\n**5–20 minutes: Context building**\nUse grand tour and specific instance questions to understand their world. Don't ask about your product or solution yet.\n\n**20–35 minutes: Deep dive**\nProbe the specific problem area you're researching. This is your highest-value time. Ask follow-up questions relentlessly.\n\n**35–40 minutes: Direct questions**\nNow you can ask more direct questions — reactions to concepts, hypothetical comparisons. Your participant has enough context to give grounded answers.\n\n**40–45 minutes: Closing**\n\"Is there anything important about this topic that I didn't ask about?\" This question frequently produces the most valuable insight of the session.\n\n## How Many Questions to Write\n\nWrite 5–8 questions for a 45-minute session. That sounds like too few — it isn't.\n\nEach question should open a 5–10 minute conversation. If you have 15 questions, you're doing a survey, not an interview. You'll rush through topics instead of going deep on what matters.\n\nAccording to the **User Interviews State of User Research 2025**, researchers who conducted 5 or fewer interviews reported significantly lower confidence in their findings than those who conducted 10+. Interview quality compounds — better questions in each session produce better data across all sessions.\n\n## Using AI to Ask Better Questions\n\nAI-moderated interview platforms like Koji solve a persistent problem: follow-up question quality depends on who's running the session. Human moderators have bad days, get distracted, and forget to probe when something interesting comes up.\n\nKoji's AI conducts every interview with consistent quality — probing when participants go shallow, following tangents that reveal unexpected themes, and maintaining a natural conversational tone throughout. The result is more consistent, higher-quality data across all participants.\n\nAccording to the **Maze Future of User Research Report 2026**, demand for qualitative research grew **66% year-over-year** while research team headcount stayed flat. AI-moderated interviews like Koji allow teams to meet that demand without hiring more researchers.\n\nWhen designing your Koji study, the same principles apply: start with grand tour questions, focus on experience over opinion, and keep your question list to 5–8 items. The AI handles the follow-up probing automatically. Koji recruits the participants too if you need them: describe the audience, approve a per-respondent quote in credits, and the interviews land in the same study as the people you invite yourself.\n\n## A Sample Question Set\n\nHere is an example question set for a product discovery study exploring why users churn from a B2B SaaS product:\n\n1. \"Walk me through what your team uses [product category] for on a weekly basis.\"\n2. \"Tell me about a time when the tool you were using for this really let you down. What happened?\"\n3. \"How did you decide to try [your product]? What were you hoping it would do for you?\"\n4. \"Think back to a specific moment when you almost stopped using it. What was going on?\"\n5. \"What does your workflow look like now compared to before you started using [product]?\"\n6. \"If you were recommending a solution like this to a colleague, how would you describe what it does?\"\n\nNotice: no yes/no questions, no hypotheticals, no leading language. Six questions that could sustain a rich 45-minute conversation.\n\n## Key Takeaways\n\n- Ask about past experience, not future hypotheticals\n- Start broad with grand tour questions, then drill into specific instances\n- Your best follow-up is often just \"tell me more about that\"\n- Write 5–8 questions for a 45-minute session — going deep beats going wide\n- Avoid leading language, double-barreled questions, and yes/no questions\n- Ask what words participants use to describe the problem — their language is your copy\n\n---\n\n**Want AI to ask your questions for you?** [Try Koji free](https://www.koji.so) — design your study, and Koji's AI conducts the interviews with expert follow-up probing built in.\n\n## Frequently Asked Questions\n\n**How many questions should a user interview have?**\nWrite 5–8 questions for a 45-minute session. Each question should open a 5–10 minute conversation. More questions means rushing through topics instead of going deep on what matters.\n\n**What makes a good user interview question?**\nGood questions ask about real past experiences rather than opinions or hypotheticals. They are open-ended, avoid suggesting the answer, and create space for the participant to take the conversation where they need to go.\n\n**How do you avoid leading questions in user interviews?**\nAvoid embedding assumptions in your question. Instead of \"Don't you find it frustrating when X?\" ask \"Tell me about a time X happened. What was that like?\" Review your question list and remove any word that implies a positive or negative judgment.\n\n**What is the most important question in a user interview?**\nThe closing question: \"Is there anything important about this topic that I didn't ask about?\" This question frequently produces the most valuable insight of the session because it gives participants permission to surface what mattered most to them.\n\n**Can AI conduct user interviews as effectively as a human moderator?**\nAI-moderated interviews like Koji produce consistent follow-up probing across every participant — something human moderators do variably depending on energy, focus, and experience level. For discovery research at scale, AI moderation often produces more consistent data quality. For highly sensitive topics or complex relationship-building scenarios, human moderation still has advantages.\n","category":"Tutorial","lastModified":"2026-10-30T08:12:54.413583+00:00","metaTitle":"How to Write User Interview Questions — Koji Blog","metaDescription":"A practical guide to writing user interview questions that produce real answers: question types, structures, and mistakes to avoid in qualitative research.","keywords":["how to write user interview questions","user interview questions","interview question design","qualitative research questions","customer interview questions","UX research questions","product discovery questions"],"aiSummary":"Writing effective user interview questions requires asking about past experience rather than hypothetical opinions. Use grand tour questions to open conversations, specific instance questions to drill into moments, and neutral probing questions to go deeper. Limit sessions to 5–8 questions to allow depth over breadth. AI-moderated platforms like Koji apply these principles automatically with consistent follow-up probing across all participants.","aiKeywords":["user interview questions","interview design","qualitative research","customer discovery","research methods","AI interviews","product research","UX research"],"aiContentType":"tutorial","faqItems":[{"answer":"Write 5–8 questions for a 45-minute session. Each question should open a 5–10 minute conversation. More questions means rushing through topics instead of going deep.","question":"How many questions should a user interview have?"},{"answer":"Good questions ask about real past experiences rather than opinions or hypotheticals. They are open-ended, avoid suggesting the answer, and create space for participants to take the conversation where they need to go.","question":"What makes a good user interview question?"},{"answer":"Avoid embedding assumptions in your question. Instead of asking about frustration, ask participants to walk you through what happened. Remove any word that implies a positive or negative judgment from your question list.","question":"How do you avoid leading questions in user interviews?"},{"answer":"The closing question: Is there anything important about this topic that I did not ask about? This frequently produces the most valuable insight because it gives participants permission to surface what mattered most to them.","question":"What is the most important question in a user interview?"},{"answer":"AI-moderated interviews like Koji produce consistent follow-up probing across every participant. For discovery research at scale, AI moderation often produces more consistent data quality than human moderation, which varies by session.","question":"Can AI conduct user interviews as effectively as a human moderator?"}],"relatedTopics":["user research","qualitative research","customer discovery","product discovery","interview techniques","research methods"]},{"type":"blog","id":"fa0bb365-6f5e-42cd-9734-46bc3afa92ea","slug":"koji-vs-dscout-2026","title":"Koji vs dscout: AI-Native Research vs Diary Studies (2026)","url":"https://www.koji.so/blog/koji-vs-dscout-2026","summary":"Koji and dscout are qualitative research platforms built for fundamentally different research paradigms. Koji automates one-on-one interviews with AI moderation and instant analysis, making it ideal for fast, scalable discovery research. dscout specializes in longitudinal diary studies and mobile ethnography with a 3M+ participant panel, better suited for enterprise teams tracking behavior over time. Koji offers transparent pricing as low as €1 per qualified interview, starting free with 10 credits; dscout uses custom enterprise pricing.","content":"Choosing between Koji and dscout for your qualitative research? Both platforms solve real problems — but they solve *different* problems. This guide gives you an honest, feature-by-feature breakdown so you can pick the right tool for your team.\n\n## Quick Comparison\n\n| Feature | Koji | dscout |\n|---------|------|--------|\n| AI-moderated interviews | ✅ Voice + text | ❌ Human moderation required |\n| Diary / longitudinal studies | ❌ | ✅ Core strength |\n| Automated thematic analysis | ✅ Built-in | ⚠️ Limited AI analysis |\n| Participant recruitment | ✅ Built in. Describe the audience and approve a credit quote, or use your own list, link, embed or phone calls. | ✅ 3M+ partner panel |\n| Mobile diary app | ❌ | ✅ Native iOS/Android |\n| Voice interviews | ✅ Natural AI conversations | ❌ |\n| One-click research reports | ✅ | ❌ |\n| Free plan | ✅ 10 credits, no card, one active study | ❌ No free tier |\n| Pricing model | Transparent, from €1 per qualified interview | Custom (enterprise pricing) |\n| Best for | AI-moderated interviews at scale | Mobile diary studies, longitudinal research |\n\n*Last verified: April 2026*\n\n## What Is dscout?\n\ndscout is a qualitative research platform that has carved out a dominant niche in **diary studies and mobile ethnography**. Used by 62% of enterprise research teams, dscout enables participants to document their real-world experiences over days or weeks — uploading photos, videos, and journal entries directly from their smartphones.\n\nThe platform also offers live interviews, surveys, and usability tests, but its core identity is longitudinal research: watching how users behave in-context over time, not just in a single interview session.\n\n### Where dscout Shines\n\n**Diary studies and in-context research.** If you need to understand how people interact with a product across a week or a month — how habits form, how routines evolve — dscout's mobile diary app is best-in-class. Participants document moments as they happen, giving you rich, naturalistic data you simply cannot get from a single interview.\n\n**Large participant panel.** dscout maintains access to over 3 million verified participants through its Partner Panel network. For teams that don't have their own user base to recruit from, this is a significant advantage.\n\n**Enterprise-grade research operations.** dscout has deep tooling for managing panels, segmenting participants, and running multi-activity studies. If you're a research operations team running dozens of studies across an enterprise, dscout has the infrastructure for it.\n\n### Where dscout Falls Short\n\n**Pricing opacity and cost.** dscout's pricing is custom and enterprise-only — there's no free trial and no public pricing page. Reviews consistently flag the high annual subscription as a barrier, especially for smaller teams or startups. Solo researchers are explicitly not the target audience.\n\n**No automated interview analysis.** Despite adding some AI features, dscout still relies heavily on manual synthesis. You get the raw data — videos, transcripts, journal entries — but turning that into insights requires your team's time. According to user reviews, data analysis and visualization remain the platform's weakest areas.\n\n**No AI-moderated interviews.** dscout's live interview feature requires a human moderator. You can't spin up 50 unmoderated AI interviews overnight without hiring moderation staff or a research agency.\n\n## What Is Koji?\n\nKoji is an AI-native qualitative research platform where an AI interviewer conducts the conversations — and an AI analyst turns those conversations into themes, sentiment patterns, and publishable research reports.\n\nResearchers design their study with an AI consultant, set their questions and probing strategy, then send a link to participants or have Koji recruit them. The AI interviews each participant one-on-one, asks intelligent follow-up questions, and captures nuanced qualitative data. When the interviews are done, Koji automatically synthesizes findings across all participants.\n\n### Where Koji Shines\n\n**Speed at scale.** Traditional interviews are bottlenecked by moderator time. With Koji, you can run 50 or 100 interviews simultaneously — and have a synthesized report ready in hours, not weeks. According to research from Maze, AI cuts qualitative analysis time by up to 80%, and Koji users consistently report going from question to insight in hours rather than weeks.\n\n**No moderator bias.** Every participant gets the same professionally designed interview experience. AI moderators don't get tired, don't lead witnesses, and don't ask follow-up questions differently based on what they hope to hear. According to the State of User Research 2025, researcher bias is one of the top threats to data quality in qualitative studies — Koji systematically eliminates it.\n\n**Voice interviews that feel natural.** Koji's voice interview mode enables real spoken conversations between participants and the AI — capturing tone, hesitation, and emotional nuance that text alone misses. This is particularly valuable for sensitive topics or when you need to probe emotional responses.\n\n**Accessible to non-researchers.** Koji is designed so that anyone — a founder, a product manager, a customer success lead — can run rigorous qualitative research without a dedicated researcher. The AI consultant guides study design; the AI interviewer conducts sessions; the platform generates the report. No research expertise required.\n\n**Transparent pricing.** Koji starts free with 10 credits and no card, then interviews run as low as €1 per qualified interview. You know what you're paying before you start.\n\n### Where Koji Falls Short\n\n**No longitudinal/diary studies.** If you need to track behavior over time — how a habit forms, how a workflow evolves over weeks — Koji isn't the right tool. Each Koji study is a single interview session, not an ongoing diary.\n\n**No panel to browse yourself.** Koji does not make you browse a panel. Open Panel recruit, describe the audience in plain language, and Koji finds them across 130+ countries, shows the price per completed interview in credits before you launch, and soft-launches about 10% first. If you have users, customers, or a list, send them the link and they land in the same study. What you do not get is dscout's always-on pool to browse and segment yourself.\n\n## Deep Dive: Use Cases\n\n### When to Choose dscout\n\n- **Longitudinal research**: Understanding how a product fits into daily routines over 2-4 weeks\n- **In-context moments**: Capturing real-world behavior as it happens (e.g., grocery shopping UX, healthcare workflows)\n- **Mobile-first research**: Participants primarily on smartphones\n- **Enterprise research ops**: Large research teams managing complex, multi-phase studies\n- **You have a research budget**: dscout's enterprise pricing is justified for teams with real budgets\n\n### When to Choose Koji\n\n- **Fast insight cycles**: You need research findings this week, not next month\n- **Scaling interviews**: You need 20, 50, or 100 interviews without a proportional increase in cost or time\n- **Non-researcher teams**: Product managers, founders, and startup teams who need rigorous research without hiring a researcher\n- **Voice-first depth**: You want the richness of spoken conversation, automated at scale\n- **Cost-sensitive research**: Startups and growing companies that can't justify enterprise research tool budgets\n- **Continuous discovery**: Teams who want to run research every sprint, not once a quarter\n\n## The Fundamental Difference\n\ndscout and Koji are built for different research paradigms.\n\n**dscout** is built for *observational, longitudinal research* — watching behavior unfold in real-world contexts over time. It requires human judgment to moderate live sessions and synthesize diary entries. It's a powerful tool, but it's also a resource-intensive one.\n\n**Koji** is built for *conversational, scalable research* — getting clear, direct answers to specific questions from many participants quickly. The AI handles the interview and analysis, making rigorous research accessible to teams without dedicated research staff.\n\nNeither platform is universally \"better.\" They reflect genuinely different research philosophies.\n\n## Pricing Comparison\n\n| | Koji | dscout |\n|--|------|--------|\n| Free tier | ✅ 10 credits, no card, one active study | ❌ |\n| Starting price | From €1 per qualified interview | Custom (contact sales) |\n| Pricing model | Transparent, public | Enterprise, opaque |\n| Free trial | ✅ | ❌ |\n| Target market | Startups to enterprise | Mid-market to enterprise |\n\nFor teams evaluating cost, Koji is significantly more accessible. dscout's opaque enterprise pricing is a consistent complaint in user reviews — you can't evaluate ROI without going through a sales process.\n\n## Our Take\n\nIf you're running **diary studies, mobile ethnography, or longitudinal research at an enterprise company with a dedicated research budget**, dscout is genuinely excellent at what it does. Its participant panel and mobile app are best-in-class for in-context research.\n\nIf you need **fast, scalable, AI-moderated interviews** — especially if you're a product team, startup, or agency that doesn't have a full-time researcher — Koji is the modern choice. You get the depth of qualitative interviews without the operational overhead, at a price point that doesn't require a multi-month procurement process.\n\nFor most teams doing product research, customer discovery, or feature validation, Koji delivers 10x faster insights than either traditional interviews or diary study workflows.\n\n**Ready to try Koji?** Start free — no credit card required. Run your first AI-moderated study in under 30 minutes.\n\n## Frequently Asked Questions\n\n**Can Koji and dscout be used together?**\nYes. Some teams use Koji for rapid interview-based discovery (getting fast answers to specific questions) and dscout for deeper longitudinal studies when they need to observe behavior over time. They serve complementary research methods.\n\n**Does dscout have AI features?**\ndscout has added some AI capabilities for planning and analysis, but its core workflow still requires significant human moderation and manual synthesis. Koji is AI-native — the AI conducts the interviews and generates the analysis automatically.\n\n**Is Koji a dscout alternative?**\nFor interview-based research, yes. For diary studies and longitudinal research specifically, they serve different needs. If your research primarily consists of one-on-one interviews or question-and-answer sessions, Koji is a strong alternative to dscout.\n\n**Which platform is better for startups?**\nKoji is significantly more accessible for startups. Transparent pricing as low as €1 per qualified interview, a free start with 10 credits, and no research expertise required make it the practical choice. dscout's enterprise pricing and setup requirements make it better suited for larger organizations.\n\n**Does Koji support voice interviews?**\nYes. Koji offers natural AI voice conversations where participants speak their responses. This captures tone, emotion, and nuance that text-based methods miss — something dscout's interview feature also supports via live video, but requires a human moderator.","category":"Research","lastModified":"2026-10-30T08:12:53.327646+00:00","metaTitle":"Koji vs dscout 2026 — Koji Blog","metaDescription":"Koji vs dscout: AI-moderated voice interviews vs diary study research. Honest feature comparison, pricing, and which platform is right for your team in 2026.","keywords":["koji vs dscout","dscout alternative","dscout review 2026","AI user research platform","diary study tool","qualitative research comparison","dscout pricing"],"aiSummary":"Koji and dscout are qualitative research platforms built for fundamentally different research paradigms. Koji automates one-on-one interviews with AI moderation and instant analysis, making it ideal for fast, scalable discovery research. dscout specializes in longitudinal diary studies and mobile ethnography with a 3M+ participant panel, better suited for enterprise teams tracking behavior over time. Koji offers transparent pricing as low as €1 per qualified interview, starting free with 10 credits; dscout uses custom enterprise pricing.","aiKeywords":["user research tools","AI interviews","diary studies","qualitative research","dscout","research platform comparison","remote interviews","participant panel"],"aiContentType":"comparison","faqItems":[{"answer":"For interview-based research, yes. Koji automates one-on-one AI interviews with instant analysis. dscout specializes in diary studies and longitudinal research — a different method. If you primarily run interview studies, Koji is a direct and more cost-effective alternative.","question":"Is Koji a dscout alternative?"},{"answer":"Koji is significantly better for startups. It has transparent pricing as low as €1 per qualified interview, a free start with 10 credits, and requires no research expertise. dscout targets enterprise teams and uses custom pricing that requires a sales conversation.","question":"Which is better for startups — Koji or dscout?"},{"answer":"dscout has added AI agents for planning and some analysis, but the core workflow still requires human moderators for live sessions and manual synthesis for diary entries. Koji is fully AI-native — the AI conducts interviews and generates analysis automatically.","question":"Does dscout have AI features?"},{"answer":"Yes. Some teams use Koji for fast, targeted interview research and dscout for deeper longitudinal studies. They serve complementary research methods rather than being direct substitutes in every context.","question":"Can Koji and dscout be used together?"},{"answer":"Yes. Koji offers natural AI voice conversations where participants speak their responses. Unlike dscout's live moderated video sessions, Koji voice interviews are fully automated — no human moderator required.","question":"Does Koji support voice interviews like dscout?"}],"relatedTopics":["user research tools","qualitative research platforms","AI in user research","research operations","product discovery"]},{"type":"blog","id":"35c7da10-1163-4427-825d-b640093303c9","slug":"how-to-conduct-remote-user-interviews-2026","title":"How to Conduct Remote User Interviews: The Complete Guide (2026)","url":"https://www.koji.so/blog/how-to-conduct-remote-user-interviews-2026","summary":"This comprehensive guide covers how to conduct remote user interviews in 2026, from defining research goals and writing discussion guides through recruiting participants, running sessions with professional moderation techniques, and analyzing findings. It covers both traditional manual analysis and AI-assisted methods, including how platforms like Koji automate the interview and analysis process. Based on current research showing 92% of teams now conduct virtual research sessions and AI reducing analysis time by up to 80%.","content":"Remote user interviews have become the standard method for qualitative research. With 92% of companies now conducting initial research sessions virtually, the skills to run great remote interviews are no longer optional — they're fundamental.\n\nThis guide walks you through every stage of a remote user interview: planning, recruiting, running the session, and turning raw conversation into actionable insights. We'll also show where AI tools like Koji are changing what's possible.\n\n## Why Remote User Interviews Matter\n\nRemote interviews remove geographic constraints entirely. You can talk to a user in Singapore in the morning and one in São Paulo in the afternoon, without booking a travel budget or a research lab. Remote hiring and research is 16% faster than traditional in-person approaches, with significantly lower coordination overhead.\n\nBut speed isn't the only benefit. Research from UserInterviews' State of User Research 2025 found that research demand increased from 55% in 2025 to 66% in 2026 — teams are doing *more* research, not less. Remote infrastructure is a core reason why.\n\nAccording to the Interaction Design Foundation, remote interviews are now the dominant format for UX and product research — chosen for their accessibility, scalability, and lower cost compared to in-person sessions.\n\n## Before You Start: Define Your Research Goal\n\nEvery strong interview starts with a clear objective. Before you schedule a single session, answer these questions:\n\n- **What decision will this research inform?** (e.g., \"whether to build Feature X\")\n- **Who is the right participant?** (role, behavior, experience level)\n- **What do you need to learn?** (beliefs, frustrations, workflows, mental models)\n- **How many interviews do you need?** (typically 5-8 per user segment to find patterns)\n\nIf you can't answer these clearly, you're not ready to recruit. Vague research questions produce vague answers.\n\n## Step 1: Write Your Discussion Guide\n\nA discussion guide is your roadmap — not a rigid script. Good discussion guides have:\n\n### Opening Questions (5 minutes)\nBuild rapport before diving into your topic. Ask about their role, how long they've been at their company, or a recent experience relevant to your research.\n\n*\"Can you walk me through a typical day in your role as a [job title]?\"*\n\n### Core Questions (25-35 minutes)\nThese are open-ended, behavior-focused questions. Avoid yes/no questions and leading questions.\n\n**Good:** *\"Tell me about the last time you [did the thing you're researching]. What happened?\"*\n**Bad:** *\"Do you find [our competitor] frustrating?\"*\n\nUse the Jobs-to-be-Done framework to probe motivation: *\"When you [did the task], what were you ultimately trying to accomplish?\"*\n\n### Probing Questions\nPrepare follow-up probes for each core question:\n- *\"Can you tell me more about that?\"*\n- *\"Why did you do it that way?\"*\n- *\"What did you do next?\"*\n- *\"How did that make you feel?\"*\n\n### Closing Questions (5 minutes)\n*\"Is there anything important about [topic] that I haven't asked about?\"* This often surfaces the most surprising insights.\n\n## Step 2: Recruit the Right Participants\n\nResearch quality is only as good as your participants. For product research, you want people who:\n\n- Represent your actual or target user\n- Have recently done the behavior you're studying\n- Can articulate their experience (avoid extreme power users who operate on autopilot)\n\n**Recruitment channels for remote interviews:**\n\n1. **Built-in panel recruitment**: in Koji, open Panel recruit, describe the audience, and see the price per completed interview in credits before you launch. Koji finds screened participants across 130+ countries and they land in the same study as everyone else. Quoting is free; launching is on paid plans.\n2. **Your own user base**: best signal, easiest coordination. Email or in-app recruiting, embed the interview in your product, or let Koji call them.\n3. **LinkedIn** — effective for B2B research and recruiting specific job titles\n4. **Existing customers**: high-quality participants for churn, feature, or satisfaction research\n5. **Panel services**, if you prefer to run recruitment yourself: UserInterviews, Respondent or Prolific can source a specialist audience. Share your Koji link with the people they find and the interviews land in the same study.\n\n**Screening:** Always use a screener survey before scheduling. Ask 3-4 qualification questions to ensure participants match your target profile. This prevents wasted sessions.\n\n**Incentives:** For 45-60 minute sessions, $50-100 USD (or equivalent gift cards) is standard. Under-compensating leads to no-shows and disengaged participants.\n\n## Step 3: Set Up Your Remote Interview Environment\n\nTechnical failures kill momentum. Set this up before your first session:\n\n**Video platform:** Zoom, Google Meet, or Microsoft Teams. Zoom is the most common for research — reliable, familiar to participants, and easy to record.\n\n**Recording:** Always record (with participant consent). You cannot take notes and actively listen at the same time. State clearly at the start: *\"I'd like to record this session so I can review it later. The recording won't be shared outside our research team. Is that okay?\"*\n\n**Note-taking setup:** Have a second window open with your discussion guide. Keep a running notes doc alongside it. Tag moments in the transcript as you go (e.g., \"PAIN POINT\", \"MOTIVATION\", \"SURPRISE\").\n\n**Your physical space:** Use headphones (not speakers — echo is distracting). Find a quiet room. Use a plain or blurred background. Good lighting matters — a front-facing window or ring light makes a significant difference to participant comfort.\n\n## Step 4: Run the Session\n\n### The First Two Minutes Matter Most\n\nParticipants arrive nervous. Your job in the first two minutes is to make them feel safe to be honest.\n\n- Greet them by name\n- Explain the purpose: *\"I'm here to learn from you — there are no right or wrong answers\"*\n- Set expectations: *\"We have about 45 minutes. I'll be taking notes and might pause to write things down\"*\n- Get recording consent\n- Ask if they have questions before you start\n\n### Active Listening Techniques\n\nThe most common mistake in user interviews is talking too much. Your goal is to get them talking and keep them talking.\n\n**The pause technique:** After they finish answering, wait 3-5 seconds before responding. Silence feels uncomfortable to them — they'll often fill it with the most honest thing they have to say.\n\n**Echo technique:** Repeat the last 3-4 words of what they said as a question: *\"You said you felt frustrated — frustrated?\"* Simple and remarkably effective at drawing out elaboration.\n\n**Avoid leading questions:** \"Was that frustrating?\" tells them what you want to hear. \"How did that feel?\" does not.\n\n**Follow the energy:** If they light up about something, pursue it even if it's off-script. The discussion guide is a map, not a cage.\n\n### Common Moderation Mistakes\n\n- **Asking multiple questions at once:** Pick one and ask it\n- **Explaining your product** during the session (you're researching their experience, not selling)\n- **Reacting visibly** to surprising answers (stay neutral so they feel safe continuing)\n- **Running over time:** Respect the agreed session length. End on time or ask for permission to continue.\n\n## Step 5: Analyze Your Data\n\nAnalysis is where most teams lose momentum. You have recordings and notes — now what?\n\n### Traditional Analysis (Manual)\n\n**Step 1 — Transcribe:** Most video tools (Zoom, Otter.ai) auto-transcribe. Review the auto-transcript for accuracy.\n\n**Step 2 — Tag and code:** Read through each transcript and tag relevant quotes with codes (themes). E.g., \"onboarding pain\", \"motivation: time saving\", \"competitor mention\".\n\n**Step 3 — Affinity mapping:** Move coded quotes onto a shared board (FigJam, Miro) and cluster them into themes. Themes that appear across 3+ interviews are your signal.\n\n**Step 4 — Write findings:** Translate clusters into insight statements: *\"Users struggle with X because Y, which leads them to Z.\"*\n\nManual analysis of 10 interviews typically takes 10-20 hours.\n\n### AI-Assisted Analysis (Modern)\n\nAI tools are dramatically compressing this timeline. According to Maze's Future of User Research 2026 report, **88% of researchers identified AI-assisted analysis and synthesis as the top trend for 2026** — and 69% already use AI in at least some of their research projects.\n\nWith tools like Koji, this entire analysis workflow is automated. Koji's AI interviewer conducts each session, and the platform automatically identifies themes, sentiment patterns, and key quotes across all interviews. Teams report going from raw interviews to publishable findings in hours rather than days or weeks.\n\nFor teams running manual interviews (recorded via Zoom), AI analysis tools can still cut synthesis time by up to 80% — a figure consistent across multiple 2025-2026 research benchmarks.\n\n## Step 6: Share Your Findings\n\nResearch that doesn't get read doesn't change decisions. Keep findings ruthlessly short:\n\n- **Executive summary:** 3-5 bullet points, decision-ready insights\n- **Key quotes:** 5-10 direct participant quotes that illustrate each theme\n- **Recommendations:** What should the team do next?\n\nAvoid 40-page research decks. Product teams and founders act on one-pagers.\n\n## How to Scale Remote Interviews\n\nThe biggest bottleneck in user research is moderator time. Each interview takes 45-60 minutes to run, plus analysis time. At 10 interviews, you're looking at 15-25 hours of work just to get to insights.\n\nThis is why AI-moderated interviews are the fastest-growing trend in research practice. Instead of replacing researchers, AI interviewers handle the execution layer — running sessions simultaneously, applying the same probing strategy consistently, and synthesizing results automatically.\n\nTeams using AI-moderated platforms like Koji can:\n- Run 50 interviews in the same time it would take to run 5 manually\n- Eliminate moderator scheduling as a bottleneck\n- Recruit participants inside the same study, with a per-respondent quote in credits approved before launch\n- Get analysis done automatically instead of spending a week in synthesis\n- Democratize research so non-researchers (PMs, founders, CS teams) can run studies independently\n\nAccording to the Maze Future of User Research 2026 report, the number of organizations where research is essential to all levels of business strategy nearly tripled — from 8% in 2025 to 22% in 2026. AI is the operational infrastructure that makes this scaling possible.\n\n## Remote Interview Tools Checklist\n\n| Tool | Purpose |\n|------|---------|\n| Zoom / Google Meet | Video sessions |\n| Calendly / Cal.com | Scheduling |\n| Otter.ai / Fireflies | Auto-transcription |\n| FigJam / Miro | Affinity mapping |\n| Notion / Confluence | Research repository |\n| [Koji](https://www.koji.so) | AI-moderated interviews + built-in panel recruitment + automated analysis |\n\n## Key Takeaways\n\n- Define a clear research objective before recruiting or writing questions\n- 5-8 interviews per user segment is enough to find patterns\n- Open-ended, behavior-focused questions get better data than yes/no questions\n- Record every session — never try to take notes and actively listen simultaneously\n- The pause technique (3-5 second silence after answers) consistently surfaces the best insights\n- AI tools can cut analysis time by up to 80%, and AI-moderated platforms like Koji eliminate the moderator bottleneck entirely\n\n## Frequently Asked Questions\n\n**How many remote user interviews do I need?**\nThe standard guidance is 5-8 interviews per user segment. Research by Nielsen Norman Group and others consistently shows that 5 interviews with the right participants surfaces 85%+ of significant usability and experience patterns.\n\n**How long should a remote user interview be?**\n45-60 minutes is the sweet spot. Long enough to build rapport and go deep; short enough that participants stay engaged and don't fatigue.\n\n**Do I need a dedicated UX researcher to run remote interviews?**\nNo. Modern AI tools like Koji are designed so that product managers, founders, and customer success teams can run rigorous research without specialized training. The AI handles interview moderation and analysis.\n\n**What's the best tool for remote user interviews?**\nFor manual moderated interviews, Zoom + a discussion guide + Otter.ai for transcription is the standard stack. For automated, scalable research, Koji recruits the participants if you need it, then conducts and analyzes AI-moderated voice or text interviews without a human moderator.\n\n**How do I prevent no-shows for remote interviews?**\nSend a confirmation email immediately after scheduling, a reminder 24 hours before, and a final reminder 1 hour before. Offer meaningful incentives ($50-100 USD for a 45-60 min session). Build a 10-15 minute buffer into your schedule for technical issues.","category":"Tutorial","lastModified":"2026-10-30T08:12:52.186202+00:00","metaTitle":"Remote User Interviews: Complete Guide 2026 — Koji","metaDescription":"Learn how to conduct remote user interviews in 2026 — recruiting, session setup, moderation techniques, analysis, and how AI tools cut research time by 80%.","keywords":["how to conduct remote user interviews","remote user interview guide","user interview best practices","remote UX research","user interview techniques","qualitative research guide 2026","online user interviews"],"aiSummary":"This comprehensive guide covers how to conduct remote user interviews in 2026, from defining research goals and writing discussion guides through recruiting participants, running sessions with professional moderation techniques, and analyzing findings. It covers both traditional manual analysis and AI-assisted methods, including how platforms like Koji automate the interview and analysis process. Based on current research showing 92% of teams now conduct virtual research sessions and AI reducing analysis time by up to 80%.","aiKeywords":["remote user interviews","user interview techniques","qualitative research methods","discussion guide","user research","AI analysis","moderation techniques","research synthesis"],"aiContentType":"tutorial","faqItems":[{"answer":"5-8 interviews per user segment is the standard. This typically surfaces 85%+ of significant patterns. Run more if you have multiple distinct user segments — 5-8 per segment, not 5-8 total.","question":"How many remote user interviews do I need?"},{"answer":"45-60 minutes is the sweet spot. Long enough to go deep with proper rapport-building, short enough that participants stay engaged. Stick to the agreed time — running over damages participant trust.","question":"How long should a remote user interview be?"},{"answer":"No. Modern AI tools are designed so product managers, founders, and customer success teams can run rigorous research without specialist training. Platforms like Koji have the AI conduct the interviews and generate the analysis automatically.","question":"Do I need a UX researcher to run remote user interviews?"},{"answer":"At minimum: a video platform (Zoom or Google Meet), a transcription tool (Otter.ai), and a place to store findings. For automated AI-moderated research, Koji handles recruiting, interviewing, transcription, and analysis in one platform.","question":"What tools do I need for remote user interviews?"},{"answer":"Send confirmation immediately after scheduling, a reminder 24 hours before, and a final nudge 1 hour before the session. Offer meaningful incentives ($50-100 USD for 45-60 minutes) and make it easy to reschedule if needed.","question":"How do I prevent no-shows in remote interviews?"}],"relatedTopics":["user research methods","UX research","qualitative research","product discovery","customer interviews","research operations"]},{"type":"blog","id":"d00cbc5a-df0a-4b7c-894d-2284ce496f00","slug":"ux-researcher-guide-scaling-with-ai-2026","title":"The UX Researcher's Guide to Scaling Research with AI (2026)","url":"https://www.koji.so/blog/ux-researcher-guide-scaling-with-ai-2026","summary":"This guide shows how UX researchers can use AI to scale research output in 2026, when demand has grown 20% year-over-year but team sizes have stayed flat. It covers AI-augmented workflows across study design, interview moderation, analysis, and reporting — including which research types benefit from AI moderation and which still require human judgment. Platforms like Koji enable researchers to shift from executor to strategist by automating the operational layer of research while preserving researcher value in methodology, ethics, and interpretation.","content":"Research demand is at an all-time high. In 2026, 66% of research teams report increased demand — up from 55% the year before. Meanwhile, most research teams are still the same size they were two years ago.\n\nThe math doesn't work. More product decisions need research input. More teams want access to user insights. But there are only so many hours in a researcher's week.\n\nAI is changing that equation — and the researchers who understand how to use it strategically are becoming dramatically more impactful. This guide covers exactly how.\n\n## The UX Research Scaling Problem\n\nIf you're a UX researcher, you know this feeling: a product team needs research insights, but the next available slot in your calendar is three weeks away. Meanwhile, the sprint starts Monday.\n\nThe State of User Research 2025 reported that **researcher burnout** and **backlog pressure** are the top two challenges for research teams — not skill gaps, not budget. The problem is *time*.\n\nTraditional qualitative research is inherently labor-intensive:\n- Recruiting takes 1-2 weeks\n- Each interview takes 45-60 minutes to run\n- Analysis of 10 interviews takes 10-20 hours\n- Report writing takes another 3-5 hours\n\nFor a team of one or two researchers supporting 5+ product squads, this math never closes.\n\n## How AI Is Changing the Research Workflow\n\nAccording to Maze's Future of User Research 2026 report, **88% of researchers identified AI-assisted analysis and synthesis as the #1 trend for 2026**, with 69% already using AI in at least some of their research projects.\n\nThe researchers reporting the biggest impact are those who have restructured their workflow — not just bolted AI onto existing processes.\n\nHere's what the modern AI-augmented research workflow looks like:\n\n### Stage 1: Study Design (AI as Research Consultant)\n\nTraditional: Researcher drafts brief, iterates with stakeholders, finalizes questions over 3-5 days.\n\nAI-augmented: AI consultant helps structure the research objective, suggests methodology, and drafts initial questions based on the business problem. Researcher reviews and refines. Time: hours, not days.\n\nTools like Koji include an AI consultant that walks you through research design — from defining the hypothesis to structuring the discussion guide. For researchers, this is particularly useful when supporting non-researcher teams who need guidance structuring their own studies.\n\n### Stage 2: Participant Interviews (AI as Interviewer)\n\nTraditional: Researcher schedules 10 interviews across 2 weeks, conducts each 45-60 minute session, takes notes alongside moderating.\n\nAI-augmented: AI interviewer conducts sessions simultaneously — 10, 50, or 100 interviews in the same time window. No scheduling bottleneck. No moderator fatigue. Consistent probing strategy applied to every participant.\n\nResearchers who adopt AI-moderated interviews for appropriate use cases (structured discovery, feature validation, satisfaction research) report freeing 60-80% of their interview time for higher-value synthesis and stakeholder work.\n\n*Important note: AI-moderated interviews are not appropriate for all research types. Sensitive research, complex usability testing with prototypes, or studies where the researcher needs to observe real-time behavior still benefit from human moderation. Strategic researchers know which method to apply when.*\n\n### Stage 3: Analysis and Synthesis (AI as Analyst)\n\nThis is where AI delivers its most significant ROI for research teams. According to multiple 2025-2026 benchmarks, **AI cuts qualitative analysis time by up to 80%**.\n\nWith traditional analysis, a researcher manually reviews 10 transcripts, codes observations, runs affinity mapping, and writes up insights. That's a 10-20 hour process.\n\nWith AI-assisted analysis:\n- Auto-transcription is instant\n- AI identifies recurring themes across all interviews\n- Sentiment patterns are extracted automatically\n- Key quotes are surfaced and categorized\n- Researcher reviews, validates, and adds interpretive judgment\n\nThe researcher's role shifts from *data processing* to *interpretation and judgment* — which is where their expertise actually adds irreplaceable value.\n\n### Stage 4: Reporting (AI as Writer)\n\nResearch reports are essential for organizational learning — but writing them is often the task researchers resent most. It's time-consuming, it's not where researchers feel most skilled, and it delays getting insights into the hands of teams who need them.\n\nAI can draft initial reports from synthesized findings, structure key takeaways in executive summary format, and format recommendations for different audiences (engineering vs. design vs. leadership).\n\nResearchers then edit, add nuance, and ensure the recommendations align with broader strategic context that only they understand.\n\n## What This Means for Your Research Practice\n\nThe researchers who are thriving in 2026 have reframed their role: from *executor* (the person who runs all the interviews) to *strategist* (the person who decides what research needs to happen, how to design it, and what the findings mean for the business).\n\nThis is a significant upgrade in leverage. Instead of being a bottleneck, you become an enabler — setting up AI-assisted research programs that other teams can run with your guidance and methodology.\n\n### The Research Democratization Opportunity\n\nThe number of organizations where research is essential to all levels of business strategy nearly tripled — from 8% in 2025 to 22% in 2026. Teams that achieve this don't do it by hiring more researchers. They do it by giving product managers, founders, and CS teams the tools and frameworks to run research themselves.\n\nFor UX researchers, this is a strategic opportunity. Building research systems — templates, tooling, quality gates, and lightweight training — multiplies your impact across the organization without multiplying your workload.\n\nKoji is particularly effective in this context. Researchers can set up study templates with rigorous discussion guides and probing strategies, then hand them to product teams to run independently. The AI ensures methodology quality. The researcher reviews outputs and adds interpretive expertise.\n\n## Building Your AI-Augmented Research Stack\n\n| Research Stage | Traditional Tool | AI-Augmented Tool |\n|----------------|-----------------|-------------------|\n| Study design | Notion/Confluence | Koji AI consultant |\n| Recruitment | Email outreach, panel marketplaces | Koji Panel recruit, personalised links |\n| Participant interviews | Zoom + moderator | Koji AI interviewer |\n| Transcription | Otter.ai, Rev | Auto (Koji, Fireflies) |\n| Analysis & synthesis | FigJam, Miro, manual | Koji auto-analysis, Notably |\n| Repository | Dovetail, EnjoyHQ | Dovetail, Koji |\n| Reporting | Slides, Notion | AI-drafted + researcher review |\n\n## Common Mistakes When Adopting AI Research Tools\n\n**Adopting AI without updating your workflow.** Adding an AI transcription tool to the same manual process doesn't unlock real efficiency. The gains come from restructuring the workflow, not just adding tools.\n\n**Using AI for everything.** AI-moderated interviews excel at structured, question-driven research. They're not the right tool for exploratory research where the researcher needs to pivot the interview based on unexpected findings, or for sensitive topics requiring human judgment and empathy.\n\n**Skipping quality validation.** AI analysis is excellent at pattern detection but can miss nuance or misclassify edge cases. Researcher review of AI outputs is not optional — it's the layer that makes findings trustworthy.\n\n**Not communicating the \"how\" to stakeholders.** Some stakeholders are skeptical of AI-generated research findings. Be transparent about your methodology — explain how the AI moderation works and what quality controls are in place. Researchers who do this build trust; those who don't risk credibility issues.\n\n## The Skills UX Researchers Need in 2026\n\nAI hasn't made research skills irrelevant — it's shifted which skills matter most.\n\n**Still essential:**\n- Research methodology expertise (knowing *when* to use which method)\n- Research ethics and participant safety\n- Stakeholder communication and insight storytelling\n- Strategic research planning (connecting research to business decisions)\n- Critical evaluation of AI outputs\n\n**Increasingly valuable:**\n- AI tool evaluation and workflow design\n- Research program management (enabling teams across the org)\n- Quantitative fluency (mixed-methods research is growing)\n- Data literacy for AI-generated analysis outputs\n\n## How Koji Helps UX Researchers Scale\n\nKoji is designed with the research team's scaling problem in mind. Here's what the workflow looks like in practice:\n\n1. **Design your study** with Koji's AI consultant — it helps structure your research objective, suggests probing strategies, and builds the discussion guide\n2. **Reach participants** with personalised links imported from a list when you want to track who answered, the generic anonymous link when candour matters, or Panel recruit when you need an audience you do not have (Koji finds them across 130+ countries at a price per completed interview in credits you approve first, on paid plans). They complete the AI-moderated interview on their own time (voice or text)\n3. **Review auto-synthesized findings** — themes, sentiment, and key quotes are ready when the last interview completes\n4. **Generate a research report** with one click — publishable findings formatted for your team\n5. **Delegate repeat studies** to product teams using your templates — you maintain methodology quality without running every session yourself\n\nFor UX researchers supporting multiple product squads, this model converts you from a bottleneck into a force multiplier.\n\n**Ready to reclaim your research calendar?** Koji's free plan lets you run your first AI-moderated study today — no credit card required.\n\n## Key Takeaways\n\n- Research demand grew 20% year-over-year in 2026; AI is the only practical way to scale supply\n- AI cuts qualitative analysis time by up to 80% — the biggest time win in the research workflow\n- The strategic shift: from researcher-as-executor to researcher-as-enabler\n- AI-moderated interviews are not appropriate for all research types — knowing when to apply them is a core skill\n- Building self-serve research systems (with AI tooling) multiplies researcher impact across the organization\n\n## Frequently Asked Questions\n\n**Will AI replace UX researchers?**\nNo. AI handles the operational layer — running interviews, transcribing, and identifying surface-level patterns. Researchers provide irreplaceable value in methodology design, ethical oversight, stakeholder communication, and interpretive judgment. AI augments researchers; it doesn't replace them.\n\n**What types of research are best suited for AI moderation?**\nAI-moderated interviews work best for structured, question-driven research: feature validation, concept testing, customer discovery, satisfaction research, and churn analysis. They're less suited for exploratory research requiring real-time pivoting, prototype usability tests, or sensitive topics requiring deep human empathy.\n\n**How do I maintain research quality when using AI tools?**\nSet rigorous discussion guides (AI will only probe as well as your questions allow), validate AI-generated themes against raw transcripts, and review outlier or contradictory findings manually. AI quality is highest when researchers design studies carefully and review outputs critically.\n\n**How do I get my organization to trust AI-moderated research?**\nBe transparent about your methodology. Explain how AI moderation works, what quality controls are in place, and how findings are validated. Running a pilot study alongside traditional interviews and comparing outputs is an effective way to build organizational confidence.\n\n**Is Koji suitable for enterprise research teams?**\nYes. Koji supports both individual researchers and teams, with features for sharing studies, managing multiple projects, and generating reports that can be distributed across an organization. The free plan is a good starting point; paid plans scale for teams running research continuously.","category":"Research","lastModified":"2026-10-30T08:12:51.106499+00:00","metaTitle":"UX Researcher Guide: Scaling Research with AI 2026","metaDescription":"How UX researchers scale their impact with AI tools in 2026 — restructuring workflows, automating analysis, and enabling research across the org without burnout.","keywords":["UX researcher guide AI 2026","scaling user research","AI tools for UX researchers","user research automation","qualitative research AI","research team scaling","UX research workflow"],"aiSummary":"This guide shows how UX researchers can use AI to scale research output in 2026, when demand has grown 20% year-over-year but team sizes have stayed flat. It covers AI-augmented workflows across study design, interview moderation, analysis, and reporting — including which research types benefit from AI moderation and which still require human judgment. Platforms like Koji enable researchers to shift from executor to strategist by automating the operational layer of research while preserving researcher value in methodology, ethics, and interpretation.","aiKeywords":["UX research scaling","AI in user research","research workflow automation","qualitative analysis AI","research democratization","AI moderated interviews","research operations"],"aiContentType":"guide","faqItems":[{"answer":"No. AI handles operational tasks — running interviews, transcribing, identifying patterns. Researchers provide irreplaceable value in methodology design, ethics, stakeholder communication, and interpretive judgment. AI makes researchers more impactful, not redundant.","question":"Will AI replace UX researchers?"},{"answer":"AI-moderated interviews work best for structured research: feature validation, concept testing, customer discovery, and satisfaction studies. They're less suited for exploratory research requiring real-time pivoting, prototype usability tests, or sensitive topics requiring deep human empathy.","question":"What types of research are best for AI moderation?"},{"answer":"Build self-serve research systems: create rigorous study templates, adopt AI-moderated interview tools like Koji, and enable product teams to run their own studies with researcher oversight. This multiplies your impact across the organization without multiplying your workload.","question":"How do I scale research without hiring more researchers?"},{"answer":"Multiple 2025-2026 benchmarks show AI can cut qualitative analysis time by up to 80%. Manual synthesis of 10 interviews typically takes 10-20 hours; AI-assisted analysis reduces this to 2-4 hours of researcher review and validation.","question":"How much time can AI save in research analysis?"},{"answer":"Be transparent about methodology. Explain how AI moderation works and what quality controls are in place. Running a comparison pilot — AI-moderated alongside traditional interviews — and showing output quality is the most effective way to build organizational trust.","question":"How do I get stakeholders to trust AI-generated research findings?"}],"relatedTopics":["UX research","AI in research","research operations","qualitative research","research democratization","product discovery"]},{"type":"blog","id":"a42effde-c2cb-41f5-95e7-ceaa280b59aa","slug":"koji-vs-lookback-2026","title":"Koji vs Lookback: AI-Native Research vs Live Moderated Sessions (2026)","url":"https://www.koji.so/blog/koji-vs-lookback-2026","summary":"Koji and Lookback take fundamentally different approaches to user research. Lookback excels at live, human-moderated sessions with a stakeholder observer lobby. Koji automates the entire research workflow — AI conducts the interview, synthesizes results across all sessions, and generates insights without a moderator. The best choice depends on whether your team needs live observation or scalable async AI interviews.","content":"Choosing between Koji and Lookback? Here is an honest breakdown to help you decide.\n\nBoth tools serve user researchers who want to learn from real customers. But they take fundamentally different approaches: Lookback is built around live, human-moderated sessions with powerful observer tools; Koji is an AI-native platform where the AI conducts the interview, analyzes results, and generates insights — all without a moderator in the room.\n\nNeither is universally better. The right choice depends on your research goals, team size, and how much moderator bandwidth your team actually has.\n\n## Quick Comparison\n\n| Feature | Koji | Lookback |\n|---------|------|---------|\n| AI-moderated interviews | ✅ Voice + text, fully automated | ❌ Requires a human moderator |\n| Live observer lobby | ❌ Not applicable (async AI) | ✅ Core feature |\n| Automated cross-session analysis | ✅ Themes, sentiment, insights | ⚠️ Per-session summaries only (Eureka AI) |\n| Research repository | ❌ | ❌ No repository layer |\n| Voice interviews | ✅ AI voice conversations | ✅ Recorded video (human-led) |\n| Unmoderated testing | ❌ | ✅ Included on all paid plans |\n| Participant recruitment | ✅ Built-in panel recruitment, quoted in credits; or use your own list | ⚠️ User Interviews integration (paid add-on) |\n| Free plan | ✅ 10 free credits, one active study at a time | ✅ 5-session trial (60 days) |\n| Starting price | From €1 per qualified interview, €3 voice. Pay as you go, no subscription | ~$25/mo (Starter, 10 sessions/year) |\n| Chrome requirement for participants | ❌ | ✅ Chrome only |\n\n## Where Koji Shines\n\n### Scale Without a Research Team\n\nKoji's core strength is eliminating the bottleneck of human moderation. Once your study is configured, the AI interviewer conducts all conversations autonomously, whether that is 10 interviews or 1,000. There is no scheduling, no session facilitation, and no post-session transcription to manage. You do not need your own audience either. If you have a list, a product, or a community, Koji interviews them. If you do not, Koji recruits the people for you: describe the audience you want and Koji returns a live per-respondent quote in credits that you approve before launch.\n\nAccording to the Maze *Future of User Research 2026* report, 66% of teams experienced an increase in research demand last year — but headcount and tooling have not kept pace. Koji is built specifically for this gap: enabling teams to run more research without hiring more researchers.\n\n### Cross-Session Synthesis at Speed\n\nWhere Lookback's Eureka AI summarizes individual sessions, Koji analyzes across all interviews simultaneously — surfacing themes, sentiment patterns, and insight clusters from the full dataset. Manual thematic analysis of a 20-interview study typically consumes 40–60 hours of analyst time. Koji compresses this to minutes.\n\nAccording to a 2023 UserInterviews ROI report, traditional agencies charge $15,000–$27,000 for a moderate qualitative study and take 4–8 weeks to deliver. Koji's model is designed to deliver comparable depth in hours, not weeks.\n\n### Voice Conversations That Feel Natural\n\nKoji's AI interviewer conducts natural voice conversations in real time. Participants respond the way they would to a real person — not by filling out a form. This produces richer qualitative data than surveys while requiring zero moderator time.\n\n### Always-On Research Cadence\n\nTeresa Torres, author of *Continuous Discovery Habits*, argues that the single highest-leverage research habit is one customer interview per week, every week. Koji makes this achievable without dedicated researcher bandwidth — you set up your study once and keep gathering insights continuously.\n\n## Where Lookback Shines\n\n### Live Moderated Sessions\n\nLookback's Observer Lobby is genuinely excellent. Stakeholders and teammates join a virtual room to watch sessions in real time, leaving time-stamped comments without disrupting the participant. For research where seeing a participant's facial expressions, hesitations, or emotional reactions matters — new product concepts, sensitive topics, complex workflows — live moderated sessions with Lookback are hard to beat.\n\n### Stakeholder Alignment\n\nOne of the underappreciated challenges in user research is getting executives and product managers to actually engage with findings. Lookback's observer experience solves this problem directly: when stakeholders watch real users struggle with a product in real time, the insight lands differently than a slide deck ever could.\n\n### Cross-Platform Session Capture\n\nLookback supports moderated and unmoderated sessions across desktop, iOS, and Android, making it a practical choice for teams testing mobile apps in native environments. The platform was rebuilt in 2024 with significantly improved video reliability and support for sessions up to four hours.\n\n### Eureka AI Assistance\n\nLookback expanded its Eureka AI in 2025 with three new workflows — Stakeholder Interviews, Stakeholder Goals, and Automatic Surfacing of Findings. The Insight Button lets researchers capture key moments during live sessions with a single click, which Eureka then annotates automatically.\n\n## Where Lookback Falls Short\n\n**Analysis requires external tools.** Lookback's Eureka provides per-session summaries, but meaningful cross-study analysis still requires researchers to export data into Dovetail, Notion, or spreadsheets. \"Lookback plus Dovetail\" is a widely cited pairing in the market — which means two subscriptions, two workflows, and added complexity.\n\n**Manual moderation is a bottleneck.** Every moderated session requires a researcher to schedule, facilitate, and actively moderate. For teams without dedicated research staff, this constraint severely limits how much research they can run. According to the UserInterviews *State of User Research 2025*, 80% of researchers now use AI tools — primarily to reduce the operational overhead of tasks like transcription and synthesis.\n\n**Chrome-only participant experience.** Lookback requires participants to use the Chrome browser, creating friction and potential drop-off — particularly on mobile.\n\n**No quality assurance on session data.** Lookback does not review session quality, so low-effort or disengaged participants may go undetected in your dataset.\n\n**No research repository.** Unlike Dovetail or Condens, Lookback does not serve as a research repository. There is no cross-project search, hierarchical coding, or theme clustering across studies.\n\n## Best Choice For...\n\n- **Choose Koji if**: You need to run interviews at scale, lack a dedicated research team, or want AI to conduct and analyze conversations without moderator involvement. Koji is ideal for continuous discovery, product feedback loops, and any scenario where speed and volume matter more than live observation.\n\n- **Choose Lookback if**: You are running live moderated sessions where observation and stakeholder presence are critical — concept testing, sensitive user research, or exploratory studies where seeing participant reactions in real time is essential.\n\n- **Consider both if**: Your team runs live moderated sessions for deep discovery (Lookback) while also needing to gather feedback at scale between projects (Koji). Many mature research teams use session capture tools and AI interview tools as complementary capabilities.\n\n## Pricing Comparison\n\n| Plan | Koji | Lookback |\n|------|------|---------|\n| Free/Trial | 10 free credits, one active study | 5 sessions, 60-day trial |\n| Entry tier | Pay as you go, no subscription, from €1 per qualified interview | ~$25/mo (10 moderated sessions/year) |\n| Mid tier | Volume pricing and plans when you want them | ~$149/mo (100 sessions/year) |\n| Enterprise | Custom | Custom (unlimited) |\n\n**Important context:** Lookback's Starter plan is nominally cheaper, but it includes only 10 moderated sessions per year — roughly one per month. Teams that want to practice continuous discovery at even a weekly cadence would need the Professional tier (~$149/mo), with additional sessions at ~$17.80 each. Koji never caps interview counts; you pay per qualified interview, from €1, which makes it more economical for high-volume research programs.\n\n## The Research Demand Gap\n\nThe UX research software market was valued at $245 million in 2024 and is projected to reach $720 million by 2033 — a 12.7% CAGR (Fortune Business Insights). Driving that growth: 58% of product professionals now use AI in their research workflows, up from 44% in 2024, according to the UserInterviews *State of User Research 2025*.\n\nThe tools that will win in this environment are the ones that reduce friction between \"we need customer insight\" and \"we have customer insight.\" For live moderated research with stakeholder presence, Lookback reduces that friction. For async AI-moderated research at scale, Koji eliminates it entirely.\n\n## Ready to Try Koji?\n\nIf your team is looking for a way to run more research without adding moderator overhead, Koji is the fastest path from question to insight. Start with a free study — no credit card required.\n\n[Start your first AI-moderated study →](/docs/getting-started)\n\n*Last verified: April 2026. Lookback pricing and features sourced from lookback.com.*\n\n## Frequently Asked Questions\n\n**Can I migrate research data from Lookback to Koji?**\nLookback exports session recordings, transcripts, and notes. Since Koji analyzes conversational interview data rather than video files, direct migration is not applicable — but insights from past Lookback studies can inform your Koji research brief and study design.\n\n**Does Koji replace Lookback for all use cases?**\nNo. Koji does not support live observer sessions or video-based moderated research. For studies where watching participants react in real time is essential, Lookback remains the better tool for those specific sessions.\n\n**Is Koji's AI as good as a human moderator?**\nFor most discovery, feedback, and validation research, Koji's AI interviewer produces insights comparable to human-moderated interviews — and significantly faster at scale. For highly sensitive or deeply exploratory research requiring improvised follow-up, human moderation still has advantages.\n\n**What is the biggest reason teams switch from Lookback to Koji?**\nMost commonly: teams want to increase research velocity without increasing headcount. When the bottleneck is moderator availability and analysis time, Koji removes both constraints simultaneously.\n\n**Does Koji offer stakeholder observation like Lookback?**\nNot in a live format. Koji's research reports and insight dashboards give stakeholders full visibility into findings — but there is no live observer lobby for watching sessions in real time.","category":"Research","lastModified":"2026-10-30T08:12:50.012436+00:00","metaTitle":"Koji vs Lookback: AI Research vs Live Moderated Sessions — Koji","metaDescription":"Honest 2026 comparison of Koji and Lookback for user research. Features, pricing, and which tool fits your team best.","keywords":["koji vs lookback","lookback alternative","user research tools 2026","AI research platform","moderated user research","live user testing","qualitative research software","lookback io review"],"aiSummary":"Koji and Lookback take fundamentally different approaches to user research. Lookback excels at live, human-moderated sessions with a stakeholder observer lobby. Koji automates the entire research workflow — AI conducts the interview, synthesizes results across all sessions, and generates insights without a moderator. The best choice depends on whether your team needs live observation or scalable async AI interviews.","aiKeywords":["user research tools","Koji vs Lookback","AI moderated interviews","live user research","qualitative research platforms","research tool comparison","continuous discovery tools"],"aiContentType":"comparison","faqItems":[{"answer":"Lookback exports session recordings, transcripts, and notes. Since Koji analyzes conversational interview data rather than video files, direct migration is not applicable — but insights from past Lookback studies can inform your Koji research brief and study design.","question":"Can I migrate research data from Lookback to Koji?"},{"answer":"No. Koji does not support live observer sessions or video-based moderated research. For studies where watching participants react in real time is essential, Lookback remains the better tool for those specific sessions.","question":"Does Koji replace Lookback for all use cases?"},{"answer":"For most discovery, feedback, and validation research, Koji's AI interviewer produces insights comparable to human-moderated interviews — and significantly faster at scale. For highly sensitive or deeply exploratory research requiring improvised follow-up, human moderation still has advantages.","question":"Is Koji's AI as good as a human moderator?"},{"answer":"Most commonly: teams want to increase research velocity without increasing headcount. When the bottleneck is moderator availability and analysis time, Koji removes both constraints simultaneously.","question":"What is the biggest reason teams switch from Lookback to Koji?"},{"answer":"Not in a live format. Koji's research reports and insight dashboards give stakeholders full visibility into findings, but there is no live observer lobby for watching sessions in real time.","question":"Does Koji offer stakeholder observation like Lookback?"}],"relatedTopics":["user research tools","AI interviews","moderated research","qualitative research software","research platform comparison","continuous discovery"]},{"type":"blog","id":"1d6362ce-f174-4cc5-90c8-64b94c56da5a","slug":"ai-feature-pricing-research-product-managers","title":"Pricing an AI Feature: How Product Managers Research Willingness to Pay Before Launch","url":"https://www.koji.so/blog/ai-feature-pricing-research-product-managers","summary":"A product manager playbook for pricing an AI feature before launch. Covers 2026 evidence on AI pricing models (ICONIQ: consumption pricing 35% to 42%, outcome-based 18% to 23%, 37% plan to change; Futurum: 42.9% of buyers prefer consumption pricing for GenAI; Simon-Kucher: AI revenue impact typically below 10%), the four things to learn (value unit, current spend, tolerance for variable bills, approval threshold), example questions, and how to run the study with AI-moderated interviews in Koji.","content":"To price an AI feature with no benchmark, interview 20 to 30 of the people who would pay for it before you launch. Ask what they spend today on the job the feature does, which unit of value they would accept being billed on (seats, usage, outcomes or a flat add-on), and where the price starts to need sign-off. Do not ask \"what would you pay?\" Run those conversations as AI-moderated interviews and you can have them done in a week, with a report that shows which pricing model your buyers will accept and why, before the price goes live and gets hard to change.\n\n*Updated October 2026: added the Price Test question type, which finds the price that earns the most.*\n\n## Why AI feature pricing is harder than normal pricing\n\nPricing a normal feature is mostly a packaging question: which tier does it sit in? An AI feature adds two problems. Each use costs you money in inference, and buyers have no settled sense of what the thing is worth. Recent data shows how unsettled the market is.\n\n- **Nobody has picked a model yet.** ICONIQ's [2026 State of AI report](https://www.iconiq.com/growth/reports/state-of-ai-2026) found that subscription and platform fees are still the most common way to charge for AI (57% of companies). Consumption-based pricing rose from 35% to 42% in about six months, and outcome-based pricing from 18% to 23%. Companies now combine 1.7 pricing models on average, and 37% plan to change their AI pricing model in the next 12 months.\n- **Buyers want AI priced differently from the rest of your product.** Futurum's 1H 2026 software buyer research found [42.9% of buyers prefer consumption pricing for GenAI features](https://futurumgroup.com/press-release/42-9-of-buyers-prefer-consumption-pricing-for-genai-reject-it-for-core-saas/), while preference for consumption pricing on core software fell to 30.1%. The same customer may want your seat licence and your AI add-on billed in two different ways.\n- **Most AI launches are not moving revenue much.** Simon-Kucher reports that [96% of companies plan to launch new AI features in the next two years](https://www.simon-kucher.com/en/insights/how-win-ai-2026), yet for most the revenue impact stays modest, typically below 10%.\n\nPut those together. Everyone is shipping AI, few are getting paid well for it, and a third of companies expect to change how they charge within a year. A PM who launches at a guessed price is likely to reprice once, in public, and repricing an AI feature after launch costs more than a week of interviews up front.\n\n## What you are really trying to learn\n\nA willingness-to-pay study for an AI feature has to answer four questions. None of them is \"what is the right number?\"\n\n1. **What is the value unit?** Does the buyer think of the feature as more seats' worth of work, as a number of tasks done, or as an outcome such as a resolved ticket or a qualified lead? The unit they already use in their head is the one they will accept on an invoice.\n2. **What does the job cost today?** Staff hours, an agency, another tool. Money already being spent tells you a budget exists, and roughly how big it is.\n3. **How much unpredictability can they live with?** Usage pricing sounds fair until a finance team cannot forecast it. Ask how they would explain a variable bill internally, and whether they would want a cap.\n4. **Who signs off, and at what amount?** A PM who loves the feature and a procurement team that will not approve it add up to no sale. Find the amount where the purchase stops being something one person can decide.\n\n### Questions to ask (and the ones to avoid)\n\nAnchor every question to what people did, not what they imagine.\n\n- \"Walk me through the last time you did [the job the AI feature does]. Who did it, and how long did it take?\"\n- \"What do you spend on that today, in tools, services or staff time?\"\n- \"If this were billed per [task / report / resolved ticket], how would you estimate what you'd use in a month?\"\n- \"Would a bill that changes month to month be a problem for you? What would make it acceptable?\" (probe: caps, prepaid credits, an annual commitment)\n- \"Which would be easier to get approved: a fixed add-on per seat or a usage-based charge? Why?\" (single choice, then probe)\n- \"At what monthly amount would this need someone else to sign off?\"\n- Close with the four Van Westendorp questions (too cheap to trust, a bargain, getting expensive, too expensive), asked about the feature as you described it.\n\nAvoid \"How much would you pay for this?\" and \"Would you pay $X?\" Both produce numbers people do not stand behind. Also avoid naming your planned price. It anchors every answer that follows.\n\n### How many interviews\n\nFor a single buyer segment, 15 to 20 interviews usually show the same value units and objections repeating. If you sell to two segments that budget differently (for example, small teams paying by card and larger companies going through procurement), plan 12 to 15 per segment. That is 25 to 30 conversations, which is a lot of calendar time for one PM and very little for an AI moderator running them in parallel.\n\n### Mistakes that ruin pricing research\n\n- **Interviewing only fans.** Your power users will tolerate almost any price. Include people on cheaper plans and people who evaluated you and left.\n- **Treating \"it depends on value\" as an answer.** Ask what value they would measure, and how.\n- **Ignoring cost to serve.** Interviews tell you what people will accept. Your inference cost tells you what you can afford. A usage price below your cost per task is a pricing error, however popular it is.\n- **Pricing once.** Re-run a short version after three months of real usage, when buyers can talk about actual bills.\n\n## How to run it in Koji\n\nYou can do this on any plan with text interviews. Voice interviews need the Interviews or Enterprise plan.\n\n1. **Start from the Willingness-to-Pay Research template.** It is built on the rules above. It anchors to current spend, never asks directly what someone would pay, maps who approves the purchase, and closes with a verbal Van Westendorp exercise. Rename it for your AI feature and change the problem statement to describe the job the feature does.\n2. **Fill in the brief.** Set the decision (\"launch as a per-seat add-on, usage-based credits or a hybrid\") and your current hypothesis. If you have a one-page description of the feature or a pricing draft for internal use, attach it in the Studio chat so the interviewer understands the product before it asks anything. Keep the price itself out of the interview guide.\n3. **Add structured questions where you need numbers.** Koji supports open-ended, scale, single choice, multiple choice, ranking and yes/no questions, each with its own follow-up depth. Use a single-choice question for \"seat add-on or usage-based\", a ranking for the value units you are considering, and a scale for how much bill variability they can tolerate. For the price itself, add a Price Test (Gabor-Granger) question: enter 5 to 7 prices, evenly spaced, and Koji starts each participant at a random price in the middle of your list. They say how likely they are to buy at that price on a five-point scale, and Koji moves up or down until it finds the highest price they would still buy at, usually in about three steps. Then it asks why, in their own words. The report shows the share who would buy at each price, the revenue each price implies and the price that earns the most, with the quotes behind each point. Read it as directional: people overstate what they would buy, and the best price is always one of the prices you tested. It reads best from about 30 interviews, works in text and voice interviews, and prices lock once responses come in. Each one becomes a chart in the report, and the interviewer still asks why. See the [structured questions guide](/docs/structured-questions-guide).\n4. **Screen for the people who hold budget.** A screening question such as \"Who approves software purchases in your team?\" lets you turn away people who cannot speak to cost. Follow the [screening questions guide](/docs/adding-screening-questions-to-your-study).\n5. **Preview before you invite.** Take the interview yourself on the Preview tab, by text or voice. Previews are free and never appear in the report. Listen for anything that hints at your price.\n6. **Invite customers and prospects.** Send the link to customers, or import a list with [personal invite links](/docs/personalized-interview-links) on the Interviews and Enterprise plans. To hear from buyers who have never used your product, [panel recruitment](/docs/panel-recruitment) lets you build an audience and see the cost on any plan; launching a recruitment needs a paid plan.\n7. **Read the report.** It opens with what you learned, then the reasons behind it, then what it means for your decision. Click any finding to read the interviews behind it. The responses grid puts every structured answer in a table you can export as CSV, which is handy for plotting Van Westendorp answers in a spreadsheet.\n\n### What it costs\n\nA text interview is 1 credit and a voice interview is 3, and credits cost €1 on paid plans. Free includes a one-time grant of 10 credits. Insights is €29 a month with 29 credits, and Interviews is €79 a month with 79 credits, so a 30-interview text study fits inside the Interviews plan. Interviews that score under 3 out of 5 on quality do not use credits and are kept out of the report, so a respondent who clicks through without saying much costs you nothing. Plans change, so check the pricing page before you budget.\n\n## What you walk into the pricing review with\n\nA good AI pricing study gives a PM four things they can defend:\n\n- **The value unit buyers already use.** For example: \"Support leads think in resolved tickets, not seats. Eleven of 14 described the job that way.\"\n- **A tolerance range for variable bills.** For example: \"Usage pricing was acceptable to most, but only with a monthly cap or prepaid credits.\"\n- **An approval threshold.** The monthly amount where the purchase moves from the team lead's card to procurement, per segment.\n- **A price corridor.** The Van Westendorp answers give a range where the feature is neither suspiciously cheap nor too expensive. Combine it with your cost per task to set the floor.\n\nThat turns the pricing meeting from opinions about competitors into a decision about evidence. For the general method, the guides to [pricing research interviews](/docs/pricing-research-interviews), the [Van Westendorp price sensitivity meter](/docs/van-westendorp-price-sensitivity-meter) and the [Gabor-Granger method](/docs/gabor-granger-pricing-method) go deeper. If you plan to raise prices on an existing plan to fund the AI work, read the [price increase research guide](/docs/price-increase-research-guide) too.\n\n## Why Koji for AI feature pricing research\n\n| What a PM weighs | Pricing survey | Pricing consultant | Calls you run yourself | Koji |\n| --- | --- | --- | --- | --- |\n| Finds the value unit | Only if you guessed it in the options | Yes | Yes | Yes, with follow-up on every answer |\n| Time to results | Days | Weeks to months | Weeks of scheduling | Days |\n| Reaches 25 to 30 buyers | Yes | Usually fewer | Rarely | Yes, in parallel |\n| Avoids anchoring on your price | Depends on design | Yes | Hard when you built it | Guardrails keep your price out |\n| Numbers and reasons together | Numbers only | Both | Reasons only | Charts plus the conversation behind them |\n| Cost | Tool cost | Five figures or more | Your time | From €29 a month |\n\nKoji is not a conjoint tool. If you need a statistically modelled price for a large consumer market, run a conjoint survey as well. Interviews tell you which models and units to put in that survey, so it measures the right thing. Our roundup of the [best pricing research tools](/blog/best-pricing-research-tools-2026) compares the options.\n\n## Start before the price is public\n\nCreate a study from the Willingness-to-Pay Research template, swap in your AI feature, add the budget-holder screening question and preview it. Once your first five interviews are in, read them and adjust the brief, then let the rest run. You can [start free](/signup) with 10 credits, or see [Koji for product managers](/for/product-managers) for the other studies PMs run. For related reading, see the [AI feature adoption playbook](/blog/ai-feature-adoption-research-product-managers), [pricing research without a consultant](/blog/pricing-research-without-consultant) and [pricing survey questions](/blog/pricing-survey-questions).\n","category":"Use Cases","lastModified":"2026-10-30T07:12:02.809019+00:00","metaTitle":"AI Feature Pricing Research for PMs (2026)","metaDescription":"How to price an AI feature with no benchmark: interview 20 to 30 buyers about value units, usage versus seat pricing and sign-off limits before you launch.","keywords":["ai feature pricing research","pricing ai features","ai add-on pricing","usage based pricing research","willingness to pay interviews","product manager pricing research"],"aiSummary":"A product manager playbook for pricing an AI feature before launch. Covers 2026 evidence on AI pricing models (ICONIQ: consumption pricing 35% to 42%, outcome-based 18% to 23%, 37% plan to change; Futurum: 42.9% of buyers prefer consumption pricing for GenAI; Simon-Kucher: AI revenue impact typically below 10%), the four things to learn (value unit, current spend, tolerance for variable bills, approval threshold), example questions, and how to run the study with AI-moderated interviews in Koji.","aiKeywords":["ai feature pricing","willingness to pay","usage-based pricing","outcome-based pricing","product management","pricing research","ai-moderated interviews"],"aiContentType":"guide","faqItems":[{"answer":"Interview 20 to 30 people who would pay for it. Ask what the job costs them today, which unit they would accept being billed on, how much bill variability they can tolerate, and at what amount the purchase needs sign-off. Avoid asking what they would pay.","question":"How do you price an AI feature with no benchmark?"},{"answer":"It depends on how your buyers think about the value. ICONIQ's 2026 data shows subscription fees still lead at 57%, with consumption pricing rising to 42%. Futurum found 42.9% of buyers prefer consumption pricing for GenAI features. Interviews tell you which your buyers will accept and what safeguards (caps, prepaid credits) they need.","question":"Should an AI feature be usage-based or a per-seat add-on?"},{"answer":"Hypothetical price questions produce numbers people do not stand behind, and naming your planned price anchors every answer after it. Anchor to current spend and past purchases, then use the four Van Westendorp questions at the end.","question":"Why not just ask customers what they would pay?"},{"answer":"About 15 to 20 for one buyer segment, or 12 to 15 per segment if you sell to groups that budget differently.","question":"How many interviews do I need for AI pricing research?"},{"answer":"Text interviews work on every plan, including Free with a one-time grant of 10 credits. Voice interviews need Interviews or Enterprise. A text interview costs 1 credit and a voice interview 3; Interviews is EUR 79 a month with 79 credits.","question":"Which Koji plan do I need?"},{"answer":"Yes. Add a Price Test (Gabor-Granger) question with 5 to 7 prices. Each participant is walked up or down your price list until Koji finds the highest price they would still buy at, then asked why. The report shows the share who would buy at each price and the price that earns the most. It is directional and reads best from about 30 interviews.","question":"Can I run a Gabor-Granger price test in an AI interview?"}],"relatedTopics":["ai feature pricing research","willingness to pay","product managers","usage-based pricing"]},{"type":"blog","id":"cbc4079b-d8b6-437f-921e-22278a1e5efa","slug":"dovetail-alternatives-2026","title":"8 Best Dovetail Alternatives in 2026: Why Teams Are Moving to AI-Native Research","url":"https://www.koji.so/blog/dovetail-alternatives-2026","summary":"Comprehensive 2026 comparison of the 8 best Dovetail alternatives. Lead recommendation: Koji, the AI-native platform that conducts AI-moderated voice and text interviews and synthesizes insights automatically — replacing not just the repository but the entire upstream research workflow Dovetail assumes you handle elsewhere. Other alternatives covered: Condens, HeyMarvin, Looppanel, Notably, Aurelius, EnjoyHQ, BuildBetter.","content":"# 8 Best Dovetail Alternatives in 2026: Why Teams Are Moving to AI-Native Research\n\n**Quick answer:** The best Dovetail alternatives in 2026 are Koji (AI-moderated interviews + automatic analysis), Condens (lighter repository), HeyMarvin (AI-native repo), Looppanel (better transcription), Notably (analysis-focused), Aurelius (lightweight), EnjoyHQ (free tier strong), and BuildBetter (multi-source ingestion). Koji is the only tool on this list that **runs the interviews for you** — the rest still require you to recruit, schedule, and moderate before you ever open the repository.\n\nDovetail built the modern research repository category. It's a brilliant tool for organizing transcripts, tagging quotes, and rolling insights up into a searchable knowledge base. But in 2026, teams are rethinking whether a repository is the right starting point at all.\n\nIf you spend ten weeks per study scheduling interviews, paying moderators, transcribing audio, and tagging themes — and *then* dump the result into Dovetail — your bottleneck isn't storage. It's the upstream work. AI-native platforms now collapse that pipeline into a single tool that recruits, interviews, transcribes, analyzes, and reports. Dovetail is the destination. The new generation owns the entire journey.\n\nThis guide breaks down the 8 best Dovetail alternatives for 2026 — what each does well, where each falls short, and why most product teams should evaluate AI-moderated platforms like Koji before paying per-seat fees on a repository they only fill once a quarter.\n\n## Why teams are leaving Dovetail in 2026\n\nThree structural problems push teams toward alternatives:\n\n![A research team talking over laptops in a meeting room, the kind of group asking whether its research repository still fits how it works](https://images.unsplash.com/photo-1702047149248-a6049168d2a8?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHw3fHxyZXNlYXJjaCUyMHRlYW0lMjBsYXB0b3BzJTIwbWVldGluZyUyMHJvb218ZW58MXwwfHx8MTc5MzM0MzAxMXww&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Creatopy](https://unsplash.com/@creatopy?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\n1. **Per-seat pricing punishes democratization.** Dovetail starts at $29 per user per month and scales by seat. The moment you give product managers, designers, engineers, and execs read access — exactly what good research democratization looks like — costs explode. Teams report pricing structures where features that lived on legacy \"Team\" plans got pushed up to \"Business\" tiers, creating near-4x cost jumps to maintain feature parity ([Usercall, 2026](https://www.usercall.co/post/dovetail-pricing-explained-what-you-really-pay-who-it-s-for-and-smarter-alternatives)).\n\n2. **A repository doesn't conduct research.** Dovetail starts working *after* you've already done the hard part. You still need to recruit participants, run interviews, transcribe audio, and clean transcripts before any of Dovetail's tagging or theming features kick in. The total cost of one user study isn't the Dovetail seat — it's the weeks of upstream labor.\n\n3. **Quantitative data lives somewhere else.** Bringing structured survey data, NPS scores, and behavioral signals into a Dovetail study is awkward at best. Teams end up running parallel stacks: Dovetail for qualitative, Qualtrics or SurveyMonkey for quantitative, and a fourth tool to glue the insights together.\n\nThe 2026 question isn't \"which repository should I buy?\" It's \"do I actually need a repository if my research tool already produces synthesized insights, themes, and one-click reports?\"\n\n## The 8 best Dovetail alternatives in 2026\n\n### 1. Koji — best for teams that want AI to run the entire research lifecycle\n\n**Best for:** Founders, PMs, and researchers who want insights in hours, not weeks — without recruiting, scheduling, or moderating manually.\n\nKoji is fundamentally a different category from Dovetail. Where Dovetail organizes research after the fact, Koji **conducts the interviews itself** using AI-moderated voice and text conversations. You write a research brief, bring your own participants or let Koji recruit them for you, and Koji's AI consultant runs every interview in parallel — probing, building rapport, asking follow-ups, and adapting to what the participant says.\n\n**Why it beats Dovetail for most teams:**\n\n- **AI-moderated voice and text interviews.** A human moderator does 4–6 interviews per day at peak; Koji runs hundreds simultaneously ([HBR, 2026](https://hbr.org/2026/04/how-ai-helps-scale-qualitative-customer-research)). No scheduling, no Zoom, no calendar Tetris.\n- **Automatic thematic analysis built in.** Themes, quotes, and insight statements are generated as conversations complete. No tagging marathon. No coding sessions. No \"who has bandwidth to synthesize this batch?\" Slack thread.\n- **6 structured question types** — open-ended, scale, single-choice, multiple-choice, ranking, and yes/no — let you mix qualitative depth with quantitative rigor in the *same* interview, eliminating the parallel-stack problem. See the [structured questions guide](/docs/structured-questions-guide) for how this works.\n- **One-click reports** with an AI consultant that answers follow-up questions about the data, like a researcher you can interrogate at 2am.\n- **Quality gate:** only conversations that score 3+ on rigor consume credits, so you don't pay for junk transcripts. See [how the quality gate works](/docs/how-the-quality-gate-works).\n- **Pricing that doesn't scale by seat.** Interviews start as low as €1 per qualified interview and €3 per qualified voice interview. Start with pay as you go. No subscription needed. You pay only for the interviews your study actually uses. Your study pauses until you top up. Nothing is ever charged automatically. Team workspaces come with the Pro and Scale plans.\n\n**The bottom line:** if you're picking between Dovetail (organize after the fact) and Koji (run the research and get insights instantly), the question is whether your team's bottleneck is *storage* or *getting the research done at all*. For 90%+ of product teams, it's the latter.\n\n[Start free with 10 credits →](https://www.koji.so)\n\n### 2. Condens — best for \"Dovetail Lite\" at lower cost\n\nCondens markets itself as a more affordable, focused repository. Pricing comes in around $14,400/year for 10 researchers ([BuildBetter, 2026](https://blog.buildbetter.ai/best-dovetail-alternatives-2026/)) — significantly cheaper than Dovetail's per-seat structure once you scale past a few licenses.\n\n**Strengths:** Cleaner UI, better video clipping, transparent pricing, good for medium-sized research teams.\n**Weaknesses:** Same structural limitation as Dovetail — it organizes research, doesn't conduct it. You still own the entire upstream pipeline. Read our [Koji vs Condens comparison](https://www.koji.so/blog/koji-vs-condens-2026) for a deeper breakdown.\n\n### 3. HeyMarvin — best for AI-assisted analysis on existing recordings\n\nHeyMarvin pivoted hard into AI in 2024–2025 with auto-transcription, theme detection, and quote extraction. It's a strong tool if you already have a pipeline of recorded calls (sales, support, customer success) and want to mine them automatically.\n\n**Pricing:** Standard plan starts around $6,000+ ([BuildBetter, 2026](https://blog.buildbetter.ai/best-dovetail-alternatives-2026/)).\n**Limitation:** It still doesn't *run* interviews. You bring the recordings; HeyMarvin organizes and analyzes them.\n\n### 4. Looppanel — best for transcription accuracy\n\nLooppanel's claim to fame is transcription quality. In head-to-head testing on the same audio, it produced a single error vs Dovetail's four ([BuildBetter, 2026](https://blog.buildbetter.ai/best-dovetail-alternatives-2026/)). That matters more than it sounds — every transcription error is a quote that can't be cited and an insight that gets diluted.\n\n**Strengths:** Unlimited collaborators on Pro plan, accurate transcripts, decent AI tagging.\n**Weaknesses:** File-import limits, still requires you to run interviews elsewhere.\n\n### 5. Notably — best for visual synthesis\n\nNotably leans into visual workflows: digital sticky notes, affinity mapping, and AI-generated themes you can spatially organize. Excellent for researchers who think in canvases.\n\n**Best for:** Design researchers and UX teams that prefer Miro-style synthesis over linear repositories. See our [Koji vs Notably comparison](https://www.koji.so/blog/koji-vs-notably-2026).\n\n### 6. Aurelius — best for solo and lightweight teams\n\nAurelius is the no-frills, fast-loading alternative for solo researchers and small teams who don't need enterprise governance. Lower learning curve than Dovetail. Lower price ceiling, too.\n\n### 7. EnjoyHQ — best free tier for early-stage teams\n\nEnjoyHQ's free plan is unusually generous. The paid Enterprise tier starts around $1,000/month ([BuildBetter, 2026](https://blog.buildbetter.ai/best-dovetail-alternatives-2026/)), which is a steep step up — but the free plan can carry an early-stage team a long way.\n\n### 8. BuildBetter — best for B2B SaaS multi-source ingestion\n\nBuildBetter pulls from 100+ sources — sales calls, support tickets, Slack, meeting recordings — and surfaces customer signals across all of them. If your insights live in Gong, Intercom, and Slack rather than dedicated research interviews, BuildBetter consolidates them.\n\n## Dovetail alternatives compared at a glance\n\n| Tool | Conducts interviews? | Auto-analysis? | Per-seat pricing? | Best for |\n|---|---|---|---|---|\n| **Koji** | ✅ AI voice + text | ✅ Themes, quotes, reports | ❌ Flat per workspace | Teams that want end-to-end research automation |\n| Dovetail | ❌ | Partial (AI tagging) | ✅ $29/user/mo+ | Mature research orgs with dedicated teams |\n| Condens | ❌ | Partial | Mixed | Mid-size teams wanting Dovetail Lite |\n| HeyMarvin | ❌ | ✅ AI tagging | ✅ Seat-based | Teams with existing call recordings |\n| Looppanel | ❌ | ✅ AI tagging | Mixed (unlimited viewers) | Teams that prioritize transcript accuracy |\n| Notably | ❌ | ✅ Visual themes | ✅ Seat-based | Visual/design researchers |\n| Aurelius | ❌ | Partial | ✅ Seat-based | Solo researchers, small teams |\n| EnjoyHQ | ❌ | Partial | ✅ Seat-based | Early-stage teams (free plan) |\n| BuildBetter | Aggregates from sources | ✅ AI synthesis | ✅ Seat-based | B2B SaaS multi-source teams |\n\n## How to choose the right Dovetail alternative\n\nAsk three questions before you commit:\n\n**1. Where does your research bottleneck actually sit?** If it's *storage and findability*, a repository (Dovetail, Condens, Aurelius) is the right category. If it's *getting interviews done at all*, you need an AI-native platform like Koji that owns the upstream pipeline.\n\n**2. How many people need access?** Per-seat pricing turns democratization into a budget conversation. If you want product managers, engineers, and execs to read insights without filing a procurement ticket, prefer flat-rate platforms.\n\n**3. Do you mix qualitative and quantitative?** Repositories struggle with structured data. Koji's [structured question types](/docs/structured-questions-guide) — scale, ranking, yes/no, multiple-choice — let you collect both in one interview.\n\n## When Dovetail is still the right answer\n\nTo be fair: Dovetail is the right tool if you have a 10+ person research team with dedicated researchers, you already have a steady pipeline of interviews from external recruiters, and your primary workflow is collaborative coding and tagging. For mature in-house research orgs running hundreds of studies a year, Dovetail's tagging and search remain best-in-class.\n\nFor everyone else — founders, product teams, researchers running 1–5 studies per quarter — paying per seat for a repository while still doing all the interview work manually is the wrong shape. The bottleneck has moved upstream.\n\n## Why Koji is the modern default\n\nKoji isn't a Dovetail clone. It's a different category answer: instead of organizing research after it happens, run the research with AI and get synthesized insights immediately. AI-moderated voice interviews probe and follow up like a senior researcher. Automatic thematic analysis surfaces patterns across hundreds of conversations. The AI consultant lets you ask questions of your data — \"what did SMB users say about pricing?\" — and get instant, citation-linked answers.\n\nThe result: from research question to validated insight in **hours, not weeks**. No moderators, no transcription bottleneck, no tagging marathon, no per-seat negotiations.\n\n[Run your first AI-moderated study free →](https://www.koji.so)\n","category":"Research","lastModified":"2026-10-30T06:50:11.941123+00:00","metaTitle":"8 Best Dovetail Alternatives in 2026 (Compared & Ranked)","metaDescription":"Compare the 8 best Dovetail alternatives in 2026 — Koji, Condens, HeyMarvin, Looppanel, Notably, Aurelius, EnjoyHQ, and BuildBetter. Why per-seat repositories lose to AI-native platforms that run the interviews for you.","keywords":["dovetail alternatives","dovetail competitors","user research repository","dovetail vs","best research repository tools","AI customer research"],"aiSummary":"Comprehensive 2026 comparison of the 8 best Dovetail alternatives. Lead recommendation: Koji, the AI-native platform that conducts AI-moderated voice and text interviews and synthesizes insights automatically — replacing not just the repository but the entire upstream research workflow Dovetail assumes you handle elsewhere. Other alternatives covered: Condens, HeyMarvin, Looppanel, Notably, Aurelius, EnjoyHQ, BuildBetter.","aiKeywords":["dovetail alternatives 2026","dovetail vs koji","AI research repository","AI-moderated interviews","research repository pricing","condens vs dovetail","heymarvin alternatives"],"aiContentType":"comparison","faqItems":[{"answer":"For most product teams, Koji is the best Dovetail alternative because it conducts AI-moderated interviews end-to-end and synthesizes insights automatically — eliminating the upstream recruiting, scheduling, and moderating work that Dovetail assumes you handle elsewhere. For teams that specifically need a research repository (not interview automation), Condens, HeyMarvin, and Looppanel are the strongest like-for-like alternatives.","question":"What is the best Dovetail alternative in 2026?"},{"answer":"Three structural reasons: per-seat pricing scales painfully (Dovetail starts at $29/user/month), the platform organizes research after the fact rather than running it, and integrating quantitative data alongside qualitative studies remains awkward. AI-native platforms collapse all four steps (recruit, interview, transcribe, analyze) into one tool.","question":"Why do teams leave Dovetail in 2026?"},{"answer":"Dovetail is a repository that organizes research after you have conducted it elsewhere. Koji is an AI-native platform that conducts AI-moderated voice and text interviews itself, transcribes in real time, applies automatic thematic analysis, and generates one-click reports. With Koji, you skip the entire upstream pipeline of recruiting, scheduling, and moderating that Dovetail assumes you handle separately.","question":"How is Koji different from Dovetail?"},{"answer":"Dovetail starts at $29/user/month and scales by seat. Condens lists around $14,400/year for 10 researchers. HeyMarvin Standard plans start around $6,000. EnjoyHQ Enterprise begins at $1,000/month. Koji is not priced per seat. Interviews start as low as €1 per qualified interview and €3 per qualified voice interview. Start with pay as you go. No subscription needed. You pay only for the interviews your study actually uses, and the whole workspace shares one balance with no per-seat fees.","question":"How much does Dovetail cost vs alternatives?"},{"answer":"Yes for most teams running 1–10 studies per quarter, where Koji eliminates the need for a separate repository because it produces synthesized insights and citation-linked reports automatically. Mature research organizations with 10+ dedicated researchers and hundreds of legacy studies may still benefit from a dedicated repository — many run Koji for new research alongside Dovetail for archived insights.","question":"Can Koji replace Dovetail completely?"},{"answer":"Dovetail handles B2B research well as a repository, but you still need to recruit B2B participants elsewhere (Respondent.io, your CRM, customer success outreach), pay for moderators or do interviews yourself, and transcribe — Dovetail kicks in after all that. Koji conducts B2B interviews directly with participants you import from your CRM or customer list, and when you need people you do not already have, Koji recruits them: describe the role, industry and screening questions, then approve a live per-respondent quote in credits before launch.","question":"What about Dovetail for B2B research?"}],"relatedTopics":["Dovetail alternatives","research repository tools","AI customer research","qualitative analysis tools","customer interview platforms"]},{"type":"documentation","id":"d418c979-771c-44eb-a4e9-488fc352c784","slug":"find-ideal-customer-profile-three-rounds-of-interviews","title":"Find Your Ideal Customer Profile in 3 Rounds of Interviews","url":"https://www.koji.so/docs/find-ideal-customer-profile-three-rounds-of-interviews","summary":"To find an ideal customer profile, interview in three rounds: about 5 interviews in each of 3 or 4 candidate segments to find the sharpest pain, 10 to 15 deep interviews in the strongest segment to learn the trigger, alternatives and buyer, then 10 to 15 interviews with new people from that segment found through a real sales channel, ending with a commitment test. Write the ICP as one sentence covering role, company, situation, current tool and buying trigger.","content":"# Find Your Ideal Customer Profile in 3 Rounds of Interviews\n\n**Short answer:** you find your ideal customer profile (ICP) by interviewing in three rounds, each narrower than the last. **Round 1 goes wide:** about 5 interviews in each of 3 or 4 candidate segments, to see where the problem is most painful. **Round 2 goes deep:** 10 to 15 interviews inside the one or two segments that stood out, to learn what triggers the search, what they use today and who pays. **Round 3 confirms:** 10 to 15 interviews with new people found through the channel you'd actually sell through, to check that the pattern holds and that you can reach them. Most early-stage teams can run all three in four to six weeks.\n\nThis is for the founder with a product and some early users who can't yet say who the product is really for. Some users love it, some drift away, and every sales call feels like a different pitch. That isn't a product problem yet. It's usually an ICP problem, and interviews are the quickest way to fix it.\n\n**Try it this week:** list every type of customer you've sold to or talked to, then circle the 3 or 4 you'd most like to be right about. Those are your Round 1 segments.\n\n## Why a vague ICP costs so much\n\nWhen CB Insights analysed 111 startup post-mortems, 35% named \"no market need\" as a reason the company failed, second only to running out of cash (CB Insights, Why Startups Fail, 2021). Many of those companies had users. What they lacked was a group of people with the problem badly enough, often enough, and with the money to pay for a fix.\n\nA vague ICP hides that. If you describe your customer as \"SMBs\" or \"product teams\", your interviews will contradict each other, your messaging will be generic, and you won't be able to tell a weak market from a mixed one.\n\nSuperhuman shows the other side. When the team first ran the Sean Ellis survey (\"How would you feel if you could no longer use Superhuman?\"), 22% of users said \"very disappointed\", well under the 40% benchmark Ellis uses for product-market fit. Instead of changing the product for everyone, the team studied the users who would be very disappointed, built the product around what they loved and what held back users like them, and reached 58% within about a year ([First Round Review, 2018](https://review.firstround.com/how-superhuman-built-an-engine-to-find-product-market-fit/)). Narrowing who it was for came first.\n\nSteve Blank put the principle in the first line of his Customer Development Manifesto: \"There are no facts inside your building, so get outside.\" Your ICP is a guess until customers confirm it.\n\n## The three rounds at a glance\n\n![Process of three rounds of ICP interviews: go wide across segments, go deep in the strongest one, confirm with new people, then write the ICP.](/visuals/icp-three-rounds-of-interviews/icp-three-rounds-of-interviews.png \"Each round narrows who you talk to, so the last one tests a single, specific customer\")\n\n| Round      | Who you talk to                                      | Interviews          | What you're deciding                                  |\n| ---------- | ---------------------------------------------------- | ------------------- | ----------------------------------------------------- |\n| 1. Wide    | 3 to 4 candidate segments                            | About 5 per segment | Which segment has the sharpest pain                   |\n| 2. Deep    | The 1 or 2 strongest segments                        | 10 to 15            | What triggers the search, what they use now, who pays |\n| 3. Confirm | New people from that segment, found where you'd sell | 10 to 15            | Whether the pattern holds and you can reach them      |\n\nWhy these numbers? In a well-known study, Greg Guest, Arwen Bunce and Laura Johnson found that most of the themes in their 60 interviews had already appeared within the first 12, for a fairly similar group of people (Field Methods, 2006). That only holds when the group is similar. A mixed group of \"everyone who might buy\" needs far more interviews to reach the same point, which is exactly why you narrow before you go deep.\n\n## Before Round 1: write down your candidate segments\n\nDescribe each segment by things you could filter on, not by personality:\n\n- **Role:** who has the problem day to day (an operations lead, a clinic owner, a first PM)\n- **Company:** size, industry, stage (a 20 to 50 person agency, a seed-stage B2B SaaS company)\n- **Situation:** the circumstance that makes the problem urgent (just hired a second salesperson, just lost a big customer)\n\nSituation is the one founders skip, and it's often the one that matters most. \"Agencies\" is a market. \"Agencies that just lost their biggest client\" is a segment with a reason to act this month.\n\n## Round 1: go wide\n\n**Goal:** find the segment where the problem is most frequent and most painful.\n\n**Who:** about 5 people from each of 3 or 4 segments. Mix current users, people who tried you and left, and people who've never heard of you.\n\n**What to ask:** stick to past behaviour, not opinions about your idea.\n\n- Tell me about the last time this problem came up. What happened?\n- How often does that happen?\n- What did you do about it? What does that cost you, in time or money?\n- What have you tried that didn't work?\n- If this disappeared tomorrow, what would change for you?\n\n**How to score each interview:** right after each one, rate four things from 1 to 3: how often the problem happens, how much it costs them today, whether they're already paying or working around it, and how easy they'd be to find again. A segment where most people score high on all four is your lead candidate. A segment where people describe the problem but have never tried to solve it usually isn't.\n\n## Round 2: go deep\n\n**Goal:** understand the strongest segment well enough to describe the moment someone starts looking for a solution.\n\n**Who:** 10 to 15 people from the one or two segments that scored best. If two segments are close, keep both and compare them.\n\n**What to ask:**\n\n- What happened right before you started looking for a better way?\n- What else did you look at? Why did you pick what you use now?\n- Who else was involved in the decision? Who signed off on the money?\n- What would a solution have to do for you to switch?\n- What would make you not trust a new tool for this?\n\n**What you're listening for:** the same trigger, the same alternatives and the same buyer showing up across interviews. If Round 2 answers still contradict each other, the segment is still too broad. Split it by the place where the answers diverge (company size, role, or situation) and keep going.\n\n## Round 3: confirm\n\n**Goal:** check that the pattern holds with people you didn't already know, and that you can reach them.\n\n**Who:** 10 to 15 new people from the segment, found through the channel you'd actually use to sell: a community they belong to, a job title you can search for, a list you can buy, or a panel filtered to the segment. If you can't find 10 of them without favours, that's a finding in itself.\n\n**What to ask:** repeat the core questions, then add a commitment test:\n\n- Walk me through the last time this came up.\n- What do you use for this today, and what does it cost you?\n- Would you be up for a pilot, or a call with whoever signs off on this?\n\nThe last question is the commitment test. Ask for something that costs them a little: a follow-up call with their manager, a pilot, a pre-order or a letter of intent. Polite interest isn't a signal. A next step is.\n\nIf you already have active users from this segment, run the Sean Ellis question with them too. Compare the share who'd be \"very disappointed\" inside the segment with the share across all your users. If the segment scores clearly higher, you've found your focus.\n\n![Question bank for ICP interviews in three sections: going wide across segments, going deep in one segment and confirming with new people.](/visuals/icp-interview-question-bank-three-rounds/icp-interview-question-bank-three-rounds.png \"Reuse these as your interview guide, one section per round\")\n\n## Running the interviews with real people\n\nA few habits make each round faster and the answers more honest:\n\n![A founder talking with a customer over coffee and taking notes, the kind of short conversation each interview round is built from](https://images.unsplash.com/photo-1758519290277-97c47f189860?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHw5fHxmb3VuZGVyJTIwaW50ZXJ2aWV3aW5nJTIwY3VzdG9tZXIlMjBjb2ZmZWV8ZW58MXwwfHx8MTc5MzM0MzAxMXww&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Vitaly Gariev](https://unsplash.com/@silverkblack?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\n- **Talk to the person with the problem, not the person you hope will buy.** In B2B they're often different, and you need both.\n- **Keep it short.** Twenty to thirty minutes is enough for Round 1. People say yes to short asks.\n- **Write notes the same day.** Score the four factors right after each conversation, while you still remember the details.\n- **Don't pitch until the end.** Once you describe your product, every answer bends towards it.\n- **Ask for introductions.** \"Who else do you know who deals with this?\" is the cheapest recruiting channel there is, and the answer tells you whether the segment talks to itself.\n\n## Writing your ICP after Round 3\n\nWrite it as one sentence you could hand to a salesperson or put into a search filter:\n\n> **[Role] at [company type and size] who [situation or trigger], currently [what they use now], and who [what makes them buy].**\n\nFor example: \"Operations leads at 20 to 80 person B2B service firms who just took on their first enterprise client, currently tracking work in spreadsheets, and who buy when a missed deadline costs them a renewal.\"\n\nThen list who you are **not** for. Saying no to the segments that scored low in Round 1 is what turns a profile into a strategy. The [ICP guide](/docs/ideal-customer-profile-icp) has a full template with the fields to fill in.\n\n## Mistakes that waste a round\n\n- **Segments defined by demographics alone.** Age and company size rarely explain who buys. The situation usually does.\n- **Only talking to fans.** Your happiest users tell you why the product works, not who else it would work for. Include people who left.\n- **Asking \"would you use this?\"** Future-tense questions get polite yeses. Ask what they did last time.\n- **Stopping at Round 2.** A pattern among people you already knew may not hold among strangers. Round 3 is the one that tells you whether you can sell to them.\n- **Never revisiting it.** Your ICP will shift as the product grows. Rerun Round 3 each time you move into a new market or a new price point.\n\n## How Koji helps\n\nKoji runs AI-moderated interviews in text or voice, so you can run all three rounds without booking calls and see the patterns across them in one place.\n\n- **Write the guide by asking.** Describe what you want to learn in the research assistant (\"find out which of these four customer types feels this problem most\") and it drafts the study, questions and screening included, for you to edit. The template library includes interview guides such as Jobs to Be Done.\n- **Screen for each segment.** Add screening questions with qualifying answers, so only people in the segment you're testing get through. Add a single-choice question for role or company size to tag each interview by segment.\n- **Interview your own users.** Import your customer list into Participants, group people into audiences by segment and send each audience an invitation and reminder from Koji, with each person's own link and sent, opened and started tracking (paid plans).\n- **Find people you don't know yet.** For Round 3, Find people searches professional profiles by job title, seniority, location, company size and industry, and panel recruitment fields interviews from a research panel filtered to your audience (both on paid plans).\n- **Follow-ups on every answer.** The AI interviewer asks \"what happened next?\" and \"what did that cost you?\" the way you would, so every interview has the past-behaviour detail Round 1 scoring needs.\n- **See the segments in the report.** Each report shows themes and findings with the share of participants behind them and the quotes that support them, plus a Personas view that groups participants by needs, goals and behaviours, with the share each persona represents.\n\n![Koji personas table comparing three customer personas by need, goal, behaviour and quote](/email/shots/shot-personas.jpg \"The Personas view groups participants by need and behaviour, a quick check on whether your segments really differ\")\n\n## Related Resources\n\n- [Ideal Customer Profile (ICP)](/docs/ideal-customer-profile-icp): definition, template and examples\n- [Sean Ellis Test](/docs/sean-ellis-test-product-market-fit): the 40% question for product-market fit\n- [How Many Interviews Are Enough?](/docs/how-many-interviews-enough): sample sizes for qualitative research\n- [The Mom Test](/docs/mom-test-methodology): how to ask questions people can't lie to you about\n- [Screener Questions Guide](/docs/screener-questions-guide): recruit the right people for each round\n- [Structured Questions Guide](/docs/structured-questions-guide): tag interviews by role, size or segment\n\n## Sources\n\n- CB Insights, \"The Top 12 Reasons Startups Fail\", 2021 (analysis of 111 post-mortems)\n- Rahul Vohra, \"How Superhuman Built an Engine to Find Product/Market Fit\", First Round Review, 2018\n- Steve Blank, The Customer Development Manifesto\n- Greg Guest, Arwen Bunce and Laura Johnson, \"How Many Interviews Are Enough? An Experiment with Data Saturation and Variability\", Field Methods 18(1), 2006\n","category":"Research Methods","lastModified":"2026-10-30T06:50:11.374195+00:00","metaTitle":"Find Your Ideal Customer Profile in 3 Rounds of Interviews","metaDescription":"Go wide across 3 or 4 segments, go deep in the strongest, then confirm with new people. A founder playbook with questions and scoring for each round.","keywords":["ideal customer profile","how to find your icp","icp interviews","customer segmentation interviews","startup icp","customer discovery","product-market fit"],"aiSummary":"To find an ideal customer profile, interview in three rounds: about 5 interviews in each of 3 or 4 candidate segments to find the sharpest pain, 10 to 15 deep interviews in the strongest segment to learn the trigger, alternatives and buyer, then 10 to 15 interviews with new people from that segment found through a real sales channel, ending with a commitment test. Write the ICP as one sentence covering role, company, situation, current tool and buying trigger.","aiPrerequisites":["A product idea or early product","Access to a few current or potential customers"],"aiLearningOutcomes":["Turn a vague customer idea into 3 or 4 testable segments","Run and score wide, deep and confirming interview rounds","Spot when a segment is still too broad","Test commitment and reachability before you commit","Write a one-sentence ICP and a list of who you are not for"],"aiDifficulty":"beginner","aiEstimatedTime":"14 minutes"},{"type":"blog","id":"54932c10-6821-4f7d-9d6b-f19ef95636f6","slug":"how-many-customer-interviews-do-you-really-need","title":"How Many Customer Interviews Do You Really Need?","url":"https://www.koji.so/blog/how-many-customer-interviews-do-you-really-need","summary":"This guide explains how many customer interviews are needed for meaningful research insights. It debunks the '5 users is enough' myth, explains data saturation, provides sample size recommendations by research type (5-8 for usability, 12-15 for discovery, 20-30 for diverse segments), and shows how AI interviewers enable larger sample sizes affordably.","content":"# How Many Customer Interviews Do You Really Need?\r\n\r\nYou've finally gotten buy-in for customer research. You've crafted your interview guide, identified your target customers, and you're ready to start learning. But then comes the question that trips up even experienced researchers: **How many interviews do I actually need?**\r\n\r\nAsk five different experts and you'll get five different answers. \"Five users is enough!\" \"No, you need at least 20!\" \"It depends on saturation!\"\r\n\r\nThe confusion is understandable. Sample size guidance has historically been shaped by the constraints of traditional research—where every interview costs significant time and money. But those constraints are changing, and so should our thinking about sample sizes.\r\n\r\nLet's cut through the noise and give you a practical framework for determining the right number of customer interviews for your research goals.\r\n\r\n---\r\n\r\n## The \"5 Users\" Myth: Where It Came From and Why It Doesn't Always Apply\r\n\r\nYou've probably heard the famous claim: \"Testing with 5 users uncovers 85% of usability problems.\" This insight, popularized by Jakob Nielsen of the Nielsen Norman Group, has become one of the most cited (and misapplied) guidelines in product research.\r\n\r\nHere's the thing: **Nielsen's research was specifically about usability testing**—watching users interact with an interface to find problems. It was never meant to apply to all types of customer research.\r\n\r\n### What Nielsen Actually Said\r\n\r\nNielsen's original recommendation came with important caveats that often get overlooked:\r\n\r\n1. **It applies to finding interface issues**, not understanding customer needs, motivations, or behaviors\r\n2. **It assumes a homogeneous user group**—if you have distinct user segments, you need 5 per segment\r\n3. **It's meant for iterative testing**—test 5 users, fix issues, test another 5, repeat\r\n\r\nNielsen himself clarified: \"You will likely need to test with 15 users to uncover all the usability problems in a design—but it is better to run 3 tests with 5 users each, fixing the problems uncovered in each test before testing again.\"\r\n\r\n### Why Customer Interviews Are Different\r\n\r\nCustomer interviews aren't about finding interface bugs. They're about understanding:\r\n\r\n- What problems customers are trying to solve\r\n- How they currently approach those problems\r\n- What motivates their decisions\r\n- What barriers they face\r\n- What language they use to describe their needs\r\n\r\nThis type of discovery research requires enough conversations to identify patterns and validate that those patterns are real—not just individual opinions. And that typically requires more than 5 interviews.\r\n\r\n---\r\n\r\n## Understanding Data Saturation: When Have You Heard Enough?\r\n\r\nRather than fixating on a magic number, experienced researchers focus on **data saturation**—the point where new interviews stop revealing new insights.\r\n\r\nThink of it like panning for gold. The first few pans bring up plenty of nuggets. But eventually, each new pan yields less and less. When you're consistently coming up empty, you've likely found what there is to find.\r\n\r\n### How to Recognize Saturation\r\n\r\nYou're approaching saturation when:\r\n\r\n- **Themes repeat consistently** across interviews\r\n- **New interviews confirm** what you've already heard rather than adding new dimensions\r\n- **You can predict** how customers will respond to certain questions\r\n- **Your research questions feel answered** with confidence\r\n\r\n### What Research Shows About Saturation\r\n\r\nAcademic studies on data saturation have converged on some helpful guidelines:\r\n\r\n| Population Type | Typical Saturation Point |\r\n|-----------------|-------------------------|\r\n| Homogeneous, focused scope | 9-12 interviews |\r\n| Moderately diverse | 15-20 interviews |\r\n| Highly diverse or broad scope | 25-30+ interviews |\r\n\r\nResearch by Griffin and Hauser found that 20-30 interviews typically uncover 90-95% of customer needs. Meanwhile, a systematic review of saturation studies found that 9-17 interviews typically reach saturation for most qualitative research.\r\n\r\nThe key insight? **Saturation depends heavily on how diverse your customers are and how broad your research questions are.**\r\n\r\n---\r\n\r\n## Sample Size Recommendations by Research Type\r\n\r\nDifferent research goals require different sample sizes. Here's a practical framework:\r\n\r\n![Table of customer interview sample sizes by research type, from 5 to 8 for usability testing to 30 to 50 or more for market research.](/visuals/customer-interview-sample-size-by-research-type/customer-interview-sample-size-by-research-type.png \"Start from your research goal, then add interviews for each segment you need to compare.\")\r\n\r\n### Usability Testing: 5-8 Participants\r\n\r\nIf you're testing whether customers can complete specific tasks in your product, 5-8 participants per user segment will reveal most usability issues. This is where Nielsen's original guidance applies.\r\n\r\n**When this works:**\r\n- Testing specific features or flows\r\n- Identifying interface problems\r\n- Validating design solutions\r\n\r\n### Discovery Interviews: 12-20 Participants\r\n\r\nWhen you're trying to understand customer problems, needs, and behaviors, plan for more conversations. Discovery research requires identifying patterns that hold true across multiple customers.\r\n\r\n**When this applies:**\r\n- Understanding customer problems and pain points\r\n- Exploring how customers currently solve problems\r\n- Identifying unmet needs\r\n- Validating problem hypotheses\r\n\r\n### Segmented Research: 8-12 Per Segment\r\n\r\nIf you're researching across distinct customer segments (e.g., enterprise vs. SMB, new users vs. power users), you need enough interviews within each segment to identify segment-specific patterns.\r\n\r\n**The math:**\r\n- 3 segments x 10 interviews each = 30 total interviews\r\n- This ensures you can distinguish real segment differences from individual variation\r\n\r\n### Concept Validation: 15-25 Participants\r\n\r\nWhen testing new product concepts or value propositions, you need enough feedback to feel confident about patterns. Positive feedback from 3 people isn't validation—it could be coincidence.\r\n\r\n**Target metrics:**\r\n- Strong positive signals from 60%+ of participants suggests real potential\r\n- Consistent objections from 30%+ signals a problem worth addressing\r\n- Mixed results require more investigation\r\n\r\n### Market Research: 30-50+ Participants\r\n\r\nFor broader market understanding—like sizing opportunities or understanding competitive dynamics—you need larger samples to draw meaningful conclusions.\r\n\r\n**Consider these benchmarks:**\r\n- 30 minimum for directional insights\r\n- 50+ for more confident conclusions\r\n- 100+ to support quantitative analysis\r\n\r\n---\r\n\r\n## The Traditional Constraint: Why Teams Interview Too Few\r\n\r\nHere's the uncomfortable truth: most teams conduct fewer interviews than they should. Not because they don't understand the value, but because of practical constraints:\r\n\r\n### Time Investment\r\n\r\nA single 45-minute interview actually requires:\r\n- 15-30 min scheduling and coordination\r\n- 45-60 min conducting the interview\r\n- 30-60 min taking notes and initial analysis\r\n- Time for synthesis and pattern identification\r\n\r\n**Total: 2-3 hours per interview**\r\n\r\nAt 20 interviews, that's 40-60 hours of researcher time—more than a full work week.\r\n\r\n### Access Challenges\r\n\r\nFinding and recruiting the right customers takes effort:\r\n- Identifying qualified participants\r\n- Getting responses to outreach\r\n- Scheduling around availability\r\n- Managing no-shows (typically 10-20%)\r\n\r\n### Cost Considerations\r\n\r\nBetween researcher time, incentives ($50-200 per participant), and opportunity cost, traditional interviews often run **$500-1,500 per interview** at fully-loaded cost.\r\n\r\nAt those economics, teams make trade-offs. They interview fewer customers than ideal, prioritize only the highest-stakes research, or skip customer interviews altogether.\r\n\r\n---\r\n\r\n## How AI Changes the Sample Size Equation\r\n\r\nThis is where things get interesting. AI-powered interview tools fundamentally change the economics of customer research.\r\n\r\n### What AI Interviewers Enable\r\n\r\nWith AI handling the conversation:\r\n\r\n1. **Interviews run 24/7** — Customers participate on their schedule, not yours\r\n2. **No scheduling coordination** — Send a link, get responses\r\n3. **Automated transcription and analysis** — Insights emerge faster\r\n4. **Consistent interview quality** — Every conversation follows your guide\r\n5. **Lower marginal cost** — Each additional interview costs little more than the incentive\r\n\r\n### The New Math\r\n\r\nTraditional approach: 10 interviews x $1,000/interview = $10,000\r\nAI-powered approach: 50 interviews x $100/interview = $5,000\r\n\r\n**Result: 5x the insights for half the cost.**\r\n\r\nThis doesn't just mean more interviews—it means:\r\n\r\n- **Confidence in patterns**: When 35 of 50 customers say the same thing, you know it's real\r\n- **Segment-level insights**: Interview enough customers to understand each segment\r\n- **Continuous learning**: Run ongoing research, not just periodic projects\r\n- **Statistical validity**: Large enough samples to support quantitative analysis alongside qualitative insights\r\n\r\n### When Human Interviews Still Matter\r\n\r\nAI interviewers are powerful, but they're not always the right choice:\r\n\r\n- **Sensitive topics** requiring careful rapport-building\r\n- **Complex exploratory research** with unpredictable directions\r\n- **High-stakes stakeholder interviews** where personal connection matters\r\n- **Deeply technical discovery** requiring expert follow-up questions\r\n\r\nThe best research programs use both: AI interviews for scale and breadth, human interviews for depth and nuance.\r\n\r\n---\r\n\r\n## A Practical Framework for Determining Your Sample Size\r\n\r\nRather than memorizing numbers, use this decision framework:\r\n\r\n### Step 1: Define Your Research Type\r\n\r\n| Research Type | Minimum Sample | Recommended Sample |\r\n|---------------|---------------|-------------------|\r\n| Usability testing | 5 | 8 per segment |\r\n| Discovery interviews | 12 | 15-20 |\r\n| Concept validation | 15 | 20-25 |\r\n| Segment research | 8/segment | 12/segment |\r\n| Market research | 30 | 50+ |\r\n\r\n### Step 2: Assess Audience Diversity\r\n\r\n- **Homogeneous audience** (similar contexts, needs, behaviors): Use minimum samples\r\n- **Moderately diverse** (some variation in use cases): Add 25-50%\r\n- **Highly diverse** (distinct segments, varied needs): Double the minimum or segment your research\r\n\r\n### Step 3: Consider Your Confidence Needs\r\n\r\n- **Exploratory/directional** insights: Minimum samples acceptable\r\n- **Informing significant decisions**: Recommended samples\r\n- **High-stakes or contentious topics**: Add 20-30% buffer\r\n\r\n### Step 4: Plan for Attrition\r\n\r\nRecruit 15-20% more than your target to account for:\r\n- No-shows\r\n- Technical issues\r\n- Poor-quality responses\r\n- Participants who don't match criteria\r\n\r\nWith Koji's panel the buffer is largely built in: study-level screening questions run before the interview, only interviews that pass the quality check are billed, and unused credit holds are refunded.\n\n### Step 5: Monitor Saturation as You Go\r\n\r\nDon't commit to a rigid number upfront. Instead:\r\n1. Start with your minimum viable sample\r\n2. Analyze in batches (every 5-10 interviews)\r\n3. Track whether new themes are emerging\r\n4. Stop when you're consistently hearing familiar patterns\r\n\r\n---\r\n\r\n## Common Sample Size Mistakes (And How to Avoid Them)\r\n\r\n### Mistake 1: Stopping at 5 Because \"Nielsen Said So\"\r\n\r\n**The fix:** Recognize that 5 users is for usability testing, not discovery research. For understanding customer needs, start with 12 minimum.\r\n\r\n### Mistake 2: One-Size-Fits-All Sample Sizes\r\n\r\n**The fix:** Match your sample size to your research type and audience diversity. A B2B startup with one customer segment needs fewer interviews than an enterprise product serving five distinct personas.\r\n\r\n### Mistake 3: Not Accounting for Segments\r\n\r\n**The fix:** If you're researching across segments, you need critical mass within each segment—not just total interviews. 20 interviews across 4 segments means only 5 per segment, which isn't enough.\r\n\r\n### Mistake 4: Treating Each Study in Isolation\r\n\r\n**The fix:** Build cumulative knowledge. If you interviewed 15 enterprise customers last quarter, you might only need 8 more for a focused follow-up study.\r\n\r\n### Mistake 5: Prioritizing Quantity Over Quality\r\n\r\n**The fix:** More interviews only helps if they're well-conducted. Ensure your interview guide is solid, your participants match your criteria, and your analysis is rigorous.\r\n\r\n---\r\n\r\n## Making Customer Research Sustainable\r\n\r\nThe real goal isn't to nail the \"right\" sample size for a single study. It's to build a sustainable research practice where customer insights continuously inform decisions.\r\n\r\n### Shift from Projects to Programs\r\n\r\nInstead of occasional deep-dive studies, consider:\r\n\r\n- **Continuous discovery**: Regular small-batch interviews (5-10/week)\r\n- **Always-on feedback channels**: Embedded interview opportunities in your product\r\n- **Rapid research sprints**: Quick-turn studies for specific questions\r\n\r\n### Use the Right Tool for Each Job\r\n\r\n| Need | Best Approach |\r\n|------|---------------|\r\n| Broad pattern identification | AI interviews (30-50+) |\r\n| Deep exploration | Human interviews (8-12) |\r\n| Quick validation | AI interviews (15-20) |\r\n| Sensitive topics | Human interviews (10-15) |\r\n| Ongoing pulse | AI interviews (continuous) |\r\n\r\n### Build Institutional Knowledge\r\n\r\nDon't let insights die in slide decks. Create:\r\n- A searchable repository of interview insights\r\n- Cross-study synthesis and pattern tracking\r\n- Shared understanding of core customer segments\r\n\r\n---\r\n\r\n## Key Takeaways\r\n\r\n1. **The \"5 users\" rule is for usability testing, not customer interviews.** Discovery research typically requires 12-20 participants to reach saturation.\r\n\r\n2. **Saturation matters more than arbitrary numbers.** Monitor when new interviews stop revealing new insights, and stop there.\r\n\r\n3. **Match sample size to research type.** Usability testing (5-8), discovery (12-20), concept validation (15-25), and market research (30-50+) all have different needs.\r\n\r\n4. **Account for audience diversity.** Diverse populations require larger samples. If researching segments, ensure adequate coverage within each.\r\n\r\n5. **AI changes the economics.** When interviews cost less, you can afford the sample sizes that give you real confidence in your findings.\r\n\r\n6. **Think programs, not projects.** Sustainable research practices beat occasional heroic efforts.\r\n\r\nThe question isn't really \"how many interviews do I need?\" It's \"how can I make customer conversations a regular part of how we build products?\" When research is easy and affordable, the sample size question becomes much less fraught.\r\n\r\n---\r\n\r\n## Start Talking to More Customers Today\r\n\r\nWith Koji, you can run 50 customer interviews for the same effort traditional methods require for 5. Our AI interviewer conducts thoughtful, conversational interviews at any scale—giving you the sample sizes you need for confident decision-making.\r\n\r\nKoji recruits the participants too: describe the audience in Panel recruit and approve a per-respondent quote in credits, or import your own customer list so each person gets a personalized link, and the interviews land in the same study. Stop guessing about what customers want. Start knowing.\r\n\r\n[See how Koji works](/demo) | [Start your free trial](/signup)\r\n\r\n---\r\n\r\n*This article was adapted for Koji from industry research on sample sizes in qualitative user research. For the original research, see the Nielsen Norman Group's work on interview sample sizes and academic studies on data saturation in qualitative research.*","category":"Research","lastModified":"2026-10-30T06:29:46.172627+00:00","metaTitle":"How Many Customer Interviews Do You Need? Sample Size Guide","metaDescription":"Learn exactly how many customer interviews you need for meaningful insights. From 5-user myths to AI-powered scaling, get the sample size right.","keywords":["customer interview sample size","how many user interviews","qualitative research sample size","customer research participants","interview saturation","user research sample size","product research interviews"],"aiSummary":"This guide explains how many customer interviews are needed for meaningful research insights. It debunks the '5 users is enough' myth, explains data saturation, provides sample size recommendations by research type (5-8 for usability, 12-15 for discovery, 20-30 for diverse segments), and shows how AI interviewers enable larger sample sizes affordably.","aiKeywords":["customer interview methodology","qualitative research saturation","sample size determination","research participant recruitment","AI-powered interviews","continuous discovery"],"aiContentType":"guide","faqItems":[{"answer":"\\\"For most customer research, 12-15 interviews with a homogeneous audience will uncover the majority of insights. For diverse customer segments or broader research questions, plan for 20-30 interviews. The key is monitoring for data saturation—when you stop hearing new information.\\\"","question":"\\\"How many customer interviews do I need for meaningful insights?\\\""},{"answer":"\\\"The '5 users' rule was designed for usability testing to find interface issues, not customer interviews. For discovery research and understanding customer needs, you typically need 12-20 participants to reach saturation and uncover meaningful patterns.\\\"","question":"\\\"Is 5 users really enough for customer interviews?\\\""},{"answer":"\\\"Data saturation is the point where conducting more interviews stops revealing new insights or themes. Research shows this typically occurs around 12-15 interviews for focused studies, though diverse populations or broad research scopes may require more.\\\"","question":"\\\"What is data saturation in customer research?\\\""},{"answer":"\\\"AI interviewers make larger sample sizes economically viable by automating moderation and analysis. This means teams can interview 50-100+ customers for the same cost as 5-10 traditional interviews, enabling both qualitative depth and quantitative confidence.\\\"","question":"\\\"How does AI change the sample size equation for customer interviews?\\\""}],"relatedTopics":["customer-research-methods","interview-best-practices","qualitative-research","research-planning","AI-interviews"]},{"type":"blog","id":"445d3897-a27e-43b8-8c73-e31a4e1de92e","slug":"koji-vs-listen-labs-2026","title":"Koji vs Listen Labs (2026): AI Interviews After the Salesforce Deal","url":"https://www.koji.so/blog/koji-vs-listen-labs-2026","summary":"Koji vs Listen Labs comparison, updated October 2026. Salesforce signed a definitive agreement to acquire Listen Labs on September 29, 2026, with closing expected in its fiscal Q4 2027 and no price disclosed. Listen Labs runs AI-moderated video and voice interviews, has a recruiting network Salesforce puts at 50M+ people and 120+ languages, raised a $69M Series B led by Ribbit Capital, and sells through demos and guided pilots without published prices (its own guide cites $85 to $150 per interview for most enterprise customers). Koji is self-serve: free with 10 credits, Insights at 29 euro a month and Interviews at 79 euro a month, text interviews at 1 credit and voice at 3, only interviews scoring 3 or more of 5 use credits. Koji runs voice, text and phone interviews in 30+ languages with six structured question types, and since October 2026 has a Participants directory with recruitment email, built-in incentives on paid plans and automatic removal of personal details. Choose Listen Labs for global managed consumer research and video, Koji for continuous research with your own customers.","content":"**Short answer:** Listen Labs and Koji both run AI-moderated customer interviews, but they are bought and used very differently. Listen Labs is an enterprise platform with a large recruiting network, sold through a demo and a guided pilot, and in September 2026 Salesforce signed a deal to acquire it. Koji is self-serve: you can start free, write one research question, and have an AI interviewer talking to your customers by voice, text or phone the same day, with participants, incentives and reports in the same place. Pick Listen Labs if you are an enterprise insights team that wants a managed panel at global scale and is comfortable with its future inside Salesforce. Pick Koji if you want to run research continuously with your own customers, on published prices.\n\n## What changed in 2026\n\nListen Labs had a big year. It raised a [$69M Series B led by Ribbit Capital](https://customerthink.com/?p=1075298), with Sequoia Capital among its existing investors. Then on September 29, 2026, [Salesforce signed a definitive agreement to acquire Listen Labs](https://www.salesforce.com/news/stories/salesforce-signs-definitive-agreement-to-acquire-listen-labs/). Salesforce did not disclose a price (press reports put it around $2 billion, unconfirmed), said it plans to bring Listen Labs into Marketing Cloud and Service Cloud, and expects the deal to close in the fourth quarter of its fiscal 2027, subject to regulatory approval.\n\nFor a buyer, an acquisition is not a reason to panic, but it is a reason to ask questions. Roadmaps, packaging and pricing often change when a startup joins a large suite, and an independent research tool can become one feature of a bigger contract. If you are evaluating Listen Labs now, ask how it will be sold after closing, whether it will need other Salesforce products, and what happens to your data and pricing at renewal.\n\nKoji has also changed a lot since this comparison was first written. In October 2026 it added a workspace-wide Participants directory with recruitment email, built-in participant incentives, and a setting that removes personal details from transcripts before analysis. Those launches close most of the gap between a self-serve interview tool and a full research platform.\n\n## Koji vs Listen Labs at a glance\n\n|  | Listen Labs | Koji |\n|---|---|---|\n| Status in October 2026 | Salesforce deal signed, closing expected in its fiscal Q4 2027 | Independent, self-serve |\n| Interview formats | Video and voice | Voice, text and phone calls |\n| Participants | Recruiting network Salesforce puts at 50M+ people | Your own list, a Participants directory, or panel recruitment |\n| Languages | 120+, according to Salesforce | 30+ interview languages |\n| Question style | Open-ended interviews with adaptive follow-ups | Open conversation plus six structured question types |\n| How you buy | Demo and guided pilot | Self-serve, free to start |\n\n![Table comparing Listen Labs and Koji on status, interview formats, participants, languages, question style and how you buy in 2026](/visuals/koji-vs-listen-labs-comparison-table/koji-vs-listen-labs-comparison-table.png \"The short version: managed enterprise scale on one side, self-serve research with your own customers on the other · Source: Salesforce announcement, September 29, 2026\")\n\n## Where Listen Labs is the stronger choice\n\n**Recruiting at global scale.** Salesforce describes the Listen Labs network as reaching more than 50 million potential participants, with interviews possible in more than 120 languages. If your program is a large consumer study across many markets at once, that breadth matters, and Koji supports 30+ interview languages today, and each interview opens in the participant’s own language, picked from their browser settings and location among the languages you turned on.\n\n**Video.** Listen Labs runs video interviews, which brand and creative teams use to see how people react. Koji interviews are voice, text or phone, with no video.\n\n**A managed, enterprise buying process.** Listen Labs is built for procurement-led buying. [Its own pricing guide](https://listenlabs.ai/articles/ai-customer-research-pricing-2026) says companies with 100 or more employees are directed to a guided pilot, and it describes a subscription with bundled credits spent per recruited participant at rates that vary by audience. If your company wants a pilot, a contract and an account team, that model fits.\n\n**Fit with Salesforce, after closing.** If your company already runs on Marketing Cloud and Service Cloud, Listen Labs may end up wired into the systems your teams use every day. That is a plan, not a shipped product, so judge it on what exists when you buy.\n\n## Where Koji is the stronger choice\n\n**You can start today, on published prices.** Koji starts free with 10 credits and no card. Paid plans are Insights at €29 a month with 29 credits and Interviews at €79 a month with 79 credits, with extra credits from €1 each and Enterprise on request. A text interview uses 1 credit and a voice interview 3. Listen Labs does not publish a price list. Its pricing guide puts per-interview pricing for most enterprise customers at $85 to $150, on top of platform access.\n\n**You only pay for interviews worth reading.** Each Koji interview gets a quality score from 1 to 5, and only interviews scoring 3 or more use credits or enter the report. Low-effort responses cost nothing. See [understanding quality scores](/docs/understanding-quality-scores).\n\n**Numbers and reasons in one conversation.** Koji blends open conversation with six structured question types: open-ended, scale, single choice, multiple choice, ranking and yes or no. One study returns the share of people who picked an option and the reasons behind it, and choice options can be shuffled per participant to avoid order bias. Read the [structured questions guide](/docs/structured-questions-guide).\n\n**Your own customers, without a separate research ops tool.** Koji keeps a workspace-wide Participants directory on paid plans. Import a CSV or add people one by one, group them into audiences, and email an audience a branded invitation or reminder from a study, with delivered, opened and started tracked per person. When you need people you do not have, [panel recruitment](/docs/panel-recruitment) lets you build an audience and see the cost on any plan, and launch on a paid plan.\n\n**Incentives built in.** On every paid plan, set a reward from €10 to €500 per participant, approve or decline each one with the interview quality score beside it, and Koji emails a link to claim a gift card, a prepaid Visa card or a charity donation. See [research participant incentives](/docs/research-participant-incentives).\n\n**Privacy at the source.** Turn on Remove personal details in a study, and Koji replaces names, email addresses, phone numbers, addresses and employers in each transcript with labels such as [NAME] when the interview finishes, before analysis. Read more in our guide to [automatic PII redaction in user research](/blog/pii-redaction-user-research).\n\n**Research from your AI assistant.** Koji has an [MCP connector](/docs/mcp-overview) for Claude and other assistants, so you can create studies and read results from a chat, plus a [headless API](/docs/headless-api-overview) on the Interviews and Enterprise plans to start interviews from your own product.\n\n![Koji voice interview in progress with an AI orb asking a follow-up question and a microphone button](/email/shots/shot-voice-interview.jpg \"A Koji voice interview: the AI asks a follow-up based on what the participant just said\")\n\n## Why this decision matters more now\n\nDemand for research keeps rising faster than research teams grow. In the [Maze Future of User Research Report 2025](https://maze.co/resources/user-research-report-2025/), 55% of respondents said demand for user research had increased over the past year, and Maze reports that 58% of product teams already use AI to speed up research. Most of the new demand comes from product managers, founders and marketers who need answers between sprints, not from programs big enough to justify an enterprise pilot.\n\n![A product team around a table with laptops, the kind of group that now runs its own customer research between sprints](https://images.unsplash.com/photo-1781246212288-7fa538344718?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHw1fHxwcm9kdWN0JTIwdGVhbSUyMG1lZXRpbmclMjBsYXB0b3BzJTIwb2ZmaWNlfGVufDF8MHx8fDE3OTI5MTEwMTF8MA&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Ngital](https://unsplash.com/@ngital?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\nThat is the real split between these two tools. Listen Labs is strongest when research is a large, planned program with a budget and a procurement process. Koji is strongest when research is a habit: a churn study this week, a pricing check next week, and onboarding interviews running all quarter.\n\n## Which should you pick?\n\n**Pick Listen Labs if you:**\n\n- Run large consumer studies across many countries and languages at once\n- Need video interviews\n- Want a managed pilot and an enterprise contract\n- Already run on Salesforce and want research inside that stack after the deal closes\n\n**Pick Koji if you:**\n\n- Want to start this week without a sales process\n- Mostly interview your own customers, users or prospects\n- Need numbers and reasons from the same study\n- Want participants, invitations, incentives and reports in one tool\n- Need published prices and want to pay only for interviews that pass a quality bar\n- Run research from Claude or another AI assistant\n\n## If you are already a Listen Labs customer\n\nYou do not need to decide anything before the deal closes. But your next renewal is a good moment to compare. Before you sign again, get these answers in writing:\n\n![Checklist of questions to ask before signing an AI interview platform contract, from what the platform fee includes to transcript export](/visuals/ai-interview-platform-contract-questions-checklist/ai-interview-platform-contract-questions-checklist.png \"Ask these before any renewal, and especially after an acquisition\")\n\nThen run a side-by-side test. Take one discussion guide you already use, paste it into Koji, and let the brief assistant turn it into a study. Take the interview yourself in a free [preview](/docs/previewing-your-interview), then send it to a small group of your own customers and compare the depth of answers and the report.\n\n## The bottom line\n\nListen Labs is a strong enterprise platform with a huge recruiting network, video interviews and, soon, a home inside Salesforce. If that is what your program needs, it belongs on your shortlist, with careful questions about what changes after closing.\n\nFor most teams running customer research in 2026, Koji is the better fit: self-serve, published pricing, voice, text and phone interviews in 30+ languages, structured questions inside the conversation, and participants, incentives and privacy controls in the same place.\n\n## Try Koji free\n\n[Start a free Koji study](https://www.koji.so/signup) with 10 credits and no card. Write one research question, let the AI draft the brief, and take the interview yourself in preview before you invite anyone. For more options, see our list of [Listen Labs alternatives](/docs/listen-labs-alternatives-2026) and the full [Listen Labs pricing breakdown](/blog/listen-labs-pricing-2026).\n","category":"Research","lastModified":"2026-10-30T06:28:10.295554+00:00","metaTitle":"Koji vs Listen Labs (2026): AI Interviews After the Salesforce Deal","metaDescription":"Salesforce is acquiring Listen Labs. Compare its enterprise AI interviews with Koji: self-serve, published prices, voice, text and phone, in 30+ languages.","keywords":["koji vs listen labs","listen labs alternative","listen labs salesforce","listen labs pricing","ai interview platform","ai moderated interviews","user research tools 2026"],"aiSummary":"Koji vs Listen Labs comparison, updated October 2026. Salesforce signed a definitive agreement to acquire Listen Labs on September 29, 2026, with closing expected in its fiscal Q4 2027 and no price disclosed. Listen Labs runs AI-moderated video and voice interviews, has a recruiting network Salesforce puts at 50M+ people and 120+ languages, raised a $69M Series B led by Ribbit Capital, and sells through demos and guided pilots without published prices (its own guide cites $85 to $150 per interview for most enterprise customers). Koji is self-serve: free with 10 credits, Insights at 29 euro a month and Interviews at 79 euro a month, text interviews at 1 credit and voice at 3, only interviews scoring 3 or more of 5 use credits. Koji runs voice, text and phone interviews in 30+ languages with six structured question types, and since October 2026 has a Participants directory with recruitment email, built-in incentives on paid plans and automatic removal of personal details. Choose Listen Labs for global managed consumer research and video, Koji for continuous research with your own customers.","aiKeywords":["koji vs listen labs","listen labs alternative","listen labs salesforce acquisition","ai interview platform","ai moderated interviews","self-serve user research"],"aiContentType":"comparison","faqItems":[{"answer":"Yes. Salesforce signed a definitive agreement to acquire Listen Labs on September 29, 2026, and expects the deal to close in the fourth quarter of its fiscal 2027, subject to regulatory approval. Salesforce did not disclose the price and plans to bring Listen Labs into Marketing Cloud and Service Cloud.","question":"Is Salesforce buying Listen Labs?"},{"answer":"Yes. Both run AI-moderated interviews and analyze them automatically. Koji is self-serve with published prices and runs voice, text and phone interviews in 30+ languages, while Listen Labs is sold through demos and guided pilots, runs video and voice interviews, and has a much larger recruiting network.","question":"Is Koji a Listen Labs alternative?"},{"answer":"Listen Labs does not publish a price list. Its own pricing guide says per-interview pricing for most enterprise customers is 85 to 150 dollars, on top of platform access. Koji starts free with 10 credits, then Insights is 29 euro a month with 29 credits and Interviews 79 euro a month with 79 credits, with extra credits from 1 euro. A text interview uses 1 credit and a voice interview 3, and only interviews scoring 3 or more on quality use credits.","question":"How much does Listen Labs cost compared to Koji?"},{"answer":"Yes, though at a smaller scale. Koji can recruit from its panel: build an audience and see the cost on any plan, and launch on a paid plan. Most Koji teams also interview their own customers, using the Participants directory, branded invitation emails and built-in incentives.","question":"Does Koji recruit participants like Listen Labs?"},{"answer":"No. Koji interviews are voice, text or phone. If you need video clips of participant reactions, Listen Labs is the better fit.","question":"Does Koji do video interviews?"},{"answer":"Nothing urgent. At your next renewal, get in writing what the platform fee includes, how credits are priced and when they expire, the renewal increase, and how to export transcripts and recordings. Then test one of your discussion guides in Koji with a free preview.","question":"What should Listen Labs customers do before the deal closes?"}],"relatedTopics":["Listen Labs Alternative","AI Customer Interviews","User Research Tools","AI Moderated Interviews","Enterprise Research","Self-Serve Research Platform"]},{"type":"blog","id":"cd6ef840-af1e-443b-8f06-da45c741e410","slug":"strella-alternatives-2026","title":"7 Best Strella Alternatives in 2026 for AI Customer Interviews","url":"https://www.koji.so/blog/strella-alternatives-2026","summary":"The best Strella alternatives in 2026 are Koji, Outset, Listen Labs, Conveo, Wondering, User Interviews and Lookback. Strella runs AI-moderated voice interviews with webcam and screen recording and recruits through User Interviews, which UserTesting acquired in January 2026; its pricing is usage-based but not public. Koji is the most direct self-serve alternative, with published plans from free, a built-in panel in 130+ countries, a quality gate that only charges for interviews scoring 3 or more, and 30+ interview languages; it does not record video. Outset, Conveo and Listen Labs suit teams that need video.","content":"The best Strella alternatives in 2026 are **Koji, Outset, Listen Labs, Conveo, Wondering, User Interviews paired with a moderator, and Lookback**. If you want what Strella does, AI-moderated customer interviews with automatic synthesis, but with prices you can see and a panel inside the product, Koji is the most direct swap. If recorded webcam and screen sessions are essential, Outset and Conveo are the closest matches. If you only need recruiting, User Interviews is the source most of these tools draw on anyway.\n\nStrella is a well-built product with real traction. It raised a [$14M Series A led by Bessemer Venture Partners](https://www.research-live.com/article/news/strella-secures-14m-investment/id/5143863) in October 2025, and lists Amazon, Duolingo, Apollo GraphQL and Chobani as customers. But it is sales-led, and it depends on a partner for participants. That sends a lot of teams looking.\n\n## What Strella does\n\nStrella runs AI-moderated live voice interviews. The AI asks the discussion guide's questions, probes with follow-ups, and synthesises results as interviews come in. It also offers human-moderated and hybrid studies, webcam and screen recording, embedded stimuli, AI-generated discussion guides, automatic highlight reels and a research repository ([product page](https://www.strella.io/product)). Recruitment runs through a [User Interviews integration](https://www.userinterviews.com/product-announcements/strella-recruitment-integration).\n\n## Why teams look for Strella alternatives\n\n**No public pricing.** Strella describes its pricing as usage-based with no per-seat charges, but there is no price list and no free trial. You need a sales call to find out what one study costs.\n\n**Recruiting depends on a partner.** Participants come through User Interviews, which UserTesting [acquired in January 2026](https://www.usertesting.com/blog/usertesting-acquires-user-interviews). User Interviews still runs as a standalone product, but if you rely on Strella for recruiting, your participant supply now sits inside a competitor's group.\n\n**A small review base.** Strella has very few public reviews on G2 and none on Capterra, so it is hard to check claims independently before you buy.\n\n**Speed is the whole point.** Research teams are stretched. In Maze's [Future of User Research 2025](https://maze.co/resources/user-research-report-2025/) report, **63% of respondents named time and bandwidth as their top challenge**, and **48% named recruiting the right participants**. A tool that takes weeks to buy and set up undercuts the reason you wanted AI moderation in the first place.\n\n## Strella alternatives compared\n\n| Tool | AI moderated | Recruiting | Public pricing | Video recording |\n| --- | --- | --- | --- | --- |\n| **Koji** | Yes, voice and text | Built-in panel or your own users | Yes, from free | No |\n| **Outset** | Yes, video, voice, text | 25+ panel integrations | No | Yes |\n| **Listen Labs** | Yes, voice, video, text | Built-in network | No | Yes |\n| **Conveo** | Yes, video, audio, text | Panel incentives on paid plan | Partly | Yes |\n| **Wondering** | Yes, plus prototype tests | Panel | Partly | Check with vendor |\n| **User Interviews** | No (recruiting only) | 6M+ panel | Yes, pay as you go | n/a |\n| **Lookback** | No (human-moderated) | Bring your own | Yes | Yes |\n\n## 1. Koji\n\n[Koji](https://www.koji.so) is an AI-native research platform built for teams that want to run customer interviews themselves, today. You describe your research goal, the brief assistant drafts the interview (including [screening questions](/docs/adding-screening-questions-to-your-study)), and an AI interviewer runs every conversation by voice or text, probing for the reason behind each answer. Koji then builds an aggregated report with themes, sentiment, segment filters and quotes.\n\nWhy it works as a Strella replacement:\n\n- **Prices you can read.** Free (10 one-time credits), Insights at €29/month with 29 credits, Interviews at €79/month with 79 credits, Enterprise from €500/month. Text interviews cost 1 credit and voice interviews 3. Compare plans in the [plan comparison guide](/docs/plan-comparison-guide).\n- **Recruiting built in.** Koji's [panel recruitment](/docs/panel-recruitment) reaches participants in 130+ countries. You see the price per completed interview before you launch, choose text, voice or both, and soft-launch part of the goal first to check quality. Credits are held at launch and unused credits come back.\n- **A quality gate on every interview.** Only interviews that score 3 or more use credits or appear in the report, and the score is capped by how much the participant actually said. Low-effort completes cost you nothing. See [understanding quality scores](/docs/understanding-quality-scores).\n- **Free rehearsal.** [Preview your interview](/docs/previewing-your-interview) by text or voice before it goes live, at no cost.\n- **Works where your team works.** 30+ interview languages, shared team workspaces, CSV export on every paid plan, plus JSON export, API access and an [MCP server](/docs/mcp-overview) on Interviews and Enterprise.\n\nWhat Strella does that Koji does not: Koji does not record webcam or screen video, and it has no human-moderated mode. If you need highlight reels of faces, Koji is not the right fit. If you need to understand why customers churn, buy or switch, and you need the answer this week, it is. For a direct comparison, read [Koji vs Strella](https://www.koji.so/blog/koji-vs-strella-2026).\n\n## 2. Outset\n\nOutset is the closest feature match. It runs AI-moderated interviews by video, voice and text in more than 40 languages, plus usability tests, and connects to more than 25 recruiting panels. It raised a [$30M Series B](https://outset.ai/series-b) in December 2025 led by Radical Ventures, bringing total funding to $51M, and works with Microsoft and Nestlé. Pricing is custom and not published. Choose Outset if you are an enterprise insights team that needs video.\n\n## 3. Listen Labs\n\nListen Labs runs AI interviews by voice, video and text, and is being bought. On September 29, 2026, [Salesforce signed a definitive agreement to acquire Listen Labs](https://www.salesforce.com/news/stories/salesforce-signs-definitive-agreement-to-acquire-listen-labs/), citing a network of more than 50 million potential participants and support for more than 120 languages. The deal is expected to close in Salesforce's fourth fiscal quarter of 2027. Pricing is not public. A strong option for Salesforce customers; others should ask what the acquisition means for their contract.\n\n## 4. Conveo\n\nConveo is a YC-backed AI interview platform focused on consumer research, used by brands such as Unilever and AB InBev. It raised a [$50M Series A](https://conveo.ai/mediakit/series-a) in September 2026. Participants answer by video, audio or text. Conveo publishes a Free plan and a Starter plan at $1,250/month billed annually, with Enterprise on request. Good for CPG and consumer brands.\n\n## 5. Wondering\n\nWondering combines AI-moderated interviews with prototype and website tests, which makes it a good fit for design teams. Pricing is partly public through software directories. Check its current plan list before you buy.\n\n## 6. User Interviews (plus a moderator)\n\nIf what you really valued in Strella was the participant pool, go to the source. User Interviews has a panel of more than 6 million people and a [pay-as-you-go option](https://www.userinterviews.com/support/paygo-and-the-student-experience) at $49 per B2C participant and $98 per B2B participant, before incentives. You still need something to run the interview. Pair it with Koji by sharing your Koji interview link, or use Koji's own panel and keep everything in one bill.\n\n## 7. Lookback\n\nLookback is for human-moderated sessions: live interviews and recorded tests with you or your researcher in the room. It publishes annual plans. It is not an AI moderator, so it trades scale for control.\n\n## How to pick the right Strella alternative\n\nAsk three questions.\n\n1. **Do you need video?** If yes, Outset, Conveo or Listen Labs. If the transcript and the insight are enough, Koji is faster and cheaper.\n2. **Who runs the studies?** If product managers, designers or founders will run their own research, pick a self-serve tool with a free plan. Maze's [2026 report](https://maze.co/blog/future-user-research-2026/) found that **39% of product managers already produce research insights themselves**, and research demand rose 20% year over year.\n3. **How much procurement can you stand?** Most tools on this list need a sales call. Koji lets you sign up, preview and launch the same afternoon.\n\nThe direction of travel is clear. Qualtrics' [2026 Market Research Trends report](https://www.qualtrics.com/articles/strategy-research/market-research-trends) found that **78% of researchers expect AI agents to run more than half of research projects by 2028**. The tools that win will be the ones teams can actually adopt without a six-week buying cycle.\n\n## Switch to Koji\n\nKoji gives you the core of Strella, AI-moderated interviews, automatic synthesis and recruiting, in a product you can try before you talk to anyone. Write your brief, preview it for free, recruit from the panel or share your link, and only pay for interviews that pass the quality gate.\n\n[Start free on Koji](https://www.koji.so/signup) and run your first AI-moderated study today.","category":"Comparisons","lastModified":"2026-10-30T06:27:55.857757+00:00","metaTitle":"7 Best Strella Alternatives in 2026 | AI Customer Interviews","metaDescription":"Looking for a Strella alternative? Compare Koji, Outset, Listen Labs, Conveo, Wondering, User Interviews and Lookback on pricing, recruiting and video.","keywords":["strella alternatives","strella competitors","strella.io alternatives","strella pricing","ai customer interview tools","ai moderated interviews 2026","strella vs koji"],"aiSummary":"The best Strella alternatives in 2026 are Koji, Outset, Listen Labs, Conveo, Wondering, User Interviews and Lookback. Strella runs AI-moderated voice interviews with webcam and screen recording and recruits through User Interviews, which UserTesting acquired in January 2026; its pricing is usage-based but not public. Koji is the most direct self-serve alternative, with published plans from free, a built-in panel in 130+ countries, a quality gate that only charges for interviews scoring 3 or more, and 30+ interview languages; it does not record video. Outset, Conveo and Listen Labs suit teams that need video.","aiKeywords":["strella alternatives","ai customer interviews","strella pricing","outset","listen labs","conveo","user interviews panel","koji"],"aiContentType":"comparison","faqItems":[{"answer":"The strongest Strella alternatives in 2026 are Koji, Outset, Listen Labs, Conveo, Wondering, User Interviews and Lookback. Koji suits teams that want self-serve AI interviews with public pricing; Outset, Conveo and Listen Labs suit teams that need recorded video.","question":"What are the best Strella alternatives?"},{"answer":"Strella does not publish a price list. It describes its pricing as usage-based with no per-seat charges, and there is no free trial, so you need to speak to sales for a quote.","question":"How much does Strella cost?"},{"answer":"Strella recruits through an integration with User Interviews, a panel of more than 6 million people. UserTesting acquired User Interviews in January 2026, and User Interviews continues to run as a standalone product.","question":"Where does Strella get its participants?"},{"answer":"No. Koji runs AI-moderated voice and text interviews and does not record webcam or screen video, and it has no human-moderated mode. It focuses on getting deep answers and an automatic report quickly.","question":"Does Koji record webcam or screen video like Strella?"},{"answer":"Yes. Koji has a free plan with 10 one-time credits and free interview previews, so you can test a full AI-moderated interview before paying. Conveo also lists a free plan.","question":"Can I try a Strella alternative for free?"}],"relatedTopics":["ai customer interviews","strella pricing","participant recruitment","ai moderated research","video vs voice interviews"]},{"type":"blog","id":"adebdef9-4e5e-41f5-8cce-bff1e1417dae","slug":"discuss-io-alternatives-2026","title":"Discuss.io Alternatives (2026): 7 Options for Live and AI-Moderated Qual","url":"https://www.koji.so/blog/discuss-io-alternatives-2026","summary":"The best Discuss.io alternative depends on what you want to replace. For live video interviews and focus groups with observers, Recollective and Lookback are closest; Lookback publishes plans from $299 a year and Recollective offers live-only video at $2 a minute or less. For large live group conversations, Remesh. To stop scheduling moderators, AI-moderated interview platforms: Koji, Outset, Conveo and Listen Labs. Koji is self-serve with published prices (free start with 10 credits, Insights at 29 euros a month, Interviews at 79 euros a month) and runs voice, text and phone interviews in 30+ languages, but has no live video rooms.","content":"**Short answer:** The best Discuss.io alternative depends on which part of Discuss.io you want to replace. If you want the same live video interviews and focus groups with observers, look at Recollective or Lookback. If you want live group conversations at a scale no video room can hold, look at Remesh. If you want to stop scheduling moderators altogether, look at AI-moderated interview platforms: Koji, Outset, Conveo or Listen Labs. Koji is the one you can start today on published prices, by voice, text or phone, with your own customers or a research panel.\n\n## Why teams look past Discuss.io\n\n[Discuss.io](https://discuss.io/platform) is a mature platform for live video qualitative research. Researchers run one-to-one interviews and focus groups in video rooms, stakeholders watch from a hidden backroom, and sessions are recorded and clipped. Its AI assistant, Genie, summarises each question with supporting quotes and clips and pulls themes across transcripts, and the company has added an AI Interview Agent for self-paced interviews.\n\n![A researcher on a video call at a laptop, the kind of live session that has to be booked against a moderator's calendar](https://images.unsplash.com/photo-1606770347238-77fcfd29906c?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHw2fHx2aWRlbyUyMGNhbGwlMjBtZWV0aW5nJTIwbGFwdG9wfGVufDF8MHx8fDE3OTI5OTc0MTF8MA&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Compare Fibre](https://unsplash.com/@comparefibre?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\nSo why switch? The reasons we hear most:\n\n- **Live sessions need a moderator, every time.** Each interview is booked against someone's calendar. That caps how many conversations a team can run in a week.\n- **The price is not published.** Discuss.io sells through sales, and reviewers on [Capterra](https://www.capterra.co.uk/software/204049/discuss-io) describe it as priced like a service.\n- **Many questions do not need video.** Why did a customer churn? What were they using before? Those answers come through just as well by voice or text.\n\nThe market is moving the same way. Qualitative work is a small slice of research spend: in ESOMAR's Global Market Research figures for 2023, [qualitative made up about 14% of global research turnover against about 70% for quantitative](https://researchworld.com/articles/how-are-research-methods-evolving). Teams that want more qual on the same budget are turning to AI. In the [2025 GRIT Business Outlook](https://greenbook.org/insights/artificial-intelligence-and-machine-learning/the-sea-change-insights-from-the-2025-grit-business-outlook), 67% of research suppliers said they now build generative AI directly into client deliverables. And in-house demand keeps rising: 66% of respondents to Maze's [Future of User Research Report 2026](https://maze.co/blog/future-user-research-2026/) said demand for research had grown in the past year.\n\n## Discuss.io alternatives at a glance\n\n| Tool | Format | Who moderates | Best for |\n|---|---|---|---|\n| Discuss.io | Live video interviews and focus groups | A human, with AI summaries | Brand teams who want to watch live |\n| Koji | Voice, text and phone interviews | An AI interviewer | Self-serve interviews with your customers |\n| Remesh | Live online group conversations | A human, with AI analysis | Hearing from large groups at once |\n| Recollective | Communities, diaries and live webcam sessions | Human or AI | Longer studies over days or weeks |\n| Lookback | Live and unmoderated sessions | A human, or none | UX teams with their own participants |\n| Outset | Text, voice and video interviews | An AI interviewer | Research teams running interviews at volume |\n| Conveo | Video, voice and text interviews | An AI interviewer | Consumer brands testing concepts |\n\n![Table comparing Discuss.io with six alternatives, Koji, Remesh, Recollective, Lookback, Outset and Conveo, by format, moderator and best use](/visuals/discuss-io-alternatives-comparison-table/discuss-io-alternatives-comparison-table.png \"Start with who you want running the conversation, then pick the tool\")\n\n## 1. Koji: AI interviews without scheduling\n\n[Koji](https://www.koji.so) replaces the moderator rather than the video room. You describe what you want to learn, an assistant drafts the brief and questions, and an AI interviewer runs every conversation, asking follow-ups based on each answer. Participants pick voice or text, or take a phone call, in any of 30+ interview languages, at a time that suits them.\n\nHere is how a study runs:\n\n1. **Ask your question.** Type what you want to learn in one message, and the assistant drafts the brief and questions.\n2. **Set the guide.** Mix open questions with structured ones and choose how deep follow-ups go.\n3. **Preview for free.** Take the interview yourself by text or voice before anyone else sees it.\n4. **Invite people.** Share a link, import a CSV for personal links, or recruit from Koji's panel.\n5. **AI runs the interviews.** Participants talk or type in their own language, any time of day.\n6. **Read the report.** Findings, personas and recommendations, each one linked to the quotes behind it.\n\n![Process showing how to run an AI-moderated study in Koji in six steps, from asking your question to reading the report](/visuals/ai-moderated-study-process-in-koji/ai-moderated-study-process-in-koji.png \"No moderator calendar to fill: participants answer when it suits them\")\n\n**What it costs:** You start free with 10 credits. Insights is €29 a month with 29 credits, Interviews is €79 a month with 79 credits, and Enterprise is priced on request. A text interview uses 1 credit and a voice interview 3. Only interviews with a [quality score](/docs/understanding-quality-scores) of 3 or more out of 5 use credits.\n\n**Good to know:** Koji has no live video rooms or backroom. If watching faces in real time matters to your stakeholders, keep a video tool for those sessions. For the rest, see [how voice interviews work](/docs/ai-voice-interviews) and [how panel recruitment works](/docs/panel-recruitment).\n\n## 2. Remesh: live conversations with large groups\n\n[Remesh](https://help.remesh.ai/getting-started-remesh-live) runs live online conversations where a moderator asks open questions, participants type answers, and then they vote on each other's responses. Agreement and differences between segments show up in real time, so you can hear from far more people than a focus group holds. In 2026 Remesh [expanded its AI assistant, Remy](https://www.remesh.ai/resources/from-question-to-defensible-decision-2026-at-remesh-so-far), to help plan, build, recruit for and analyse studies, and added video interview tools.\n\n**Pricing:** custom quote, not published.\n\n**Pick it if** you want the energy of a live group at survey scale, such as reacting to a message or policy with several hundred people at once.\n\n## 3. Recollective: communities, diaries and live sessions\n\n[Recollective](https://recollective.com/qualitative-research-recollective) covers the longer side of qual: online communities, diaries, mobile tasks and pop-up groups, plus live webcam interviews and focus groups. Its AI features include transcription, summaries, themes, automatic translation and AI-moderated interviews.\n\n**Pricing:** unusually open for this category. [Recollective's pricing page](https://recollective.com/pricing) lists a live-only option with no platform fee, where you pay for video minutes at $2 a minute or less, and larger plans priced by the number of participants and studies.\n\n**Pick it if** you run studies over days or weeks and want live sessions in the same place.\n\n## 4. Lookback: live and unmoderated sessions for UX teams\n\n[Lookback](https://help.lookback.io/en/articles/5160015-which-pricing-plan-should-i-choose) records moderated and unmoderated sessions on websites, apps and prototypes, with observers able to watch live. You bring your own participants; it has no panel.\n\n**Pricing:** published, billed annually. Lookback's help centre lists plans from $299 a year for 10 sessions, with larger team plans priced by session volume.\n\n**Pick it if** you mainly need moderated usability sessions with your own users and want a lighter tool than an enterprise qual platform.\n\n## 5. Outset: AI-moderated interviews at volume\n\n[Outset](https://outset.ai/pricing) runs AI-moderated interviews where participants answer by text, voice or video, in more than 40 languages. You can bring your own participants or recruit through Outset. It [raised a $30 million Series B in December 2025](https://www.mrweb.com/drno/news39184.htm), led by Radical Ventures.\n\n**Pricing:** custom subscription, not published.\n\n**Pick it if** you are a research team that wants AI moderation with video responses and is happy to buy through sales.\n\n## 6. Conveo: AI video interviews for consumer brands\n\n[Conveo](https://pitchdrive.com/stories/conveo-raises-usd50m-series-a-as-its-ai-led-interviews-reach-the-world-s-biggest-brands) is a Belgian, Y Combinator-backed platform for AI-moderated consumer interviews by video, voice or text, with automatic theming. It raised a $50 million Series A in September 2026, led by DST Global, mainly to grow in the US.\n\n**Pricing:** not published.\n\n**Pick it if** you are a consumer brand or agency testing concepts and want to see faces without booking moderators.\n\n## 7. Listen Labs: enterprise AI interviews\n\nListen Labs runs AI-moderated interviews by video, voice and text with a large recruiting network. On September 29, 2026, Salesforce [signed a deal to acquire it](https://www.salesforce.com/news/stories/salesforce-signs-definitive-agreement-to-acquire-listen-labs/), with closing expected in Salesforce's fiscal fourth quarter of 2027. Pricing is not published. Read our [Listen Labs vs UserTesting](/blog/listen-labs-vs-usertesting-2026) comparison for more.\n\n## How to choose\n\nAsk three questions, in this order.\n\n**Do stakeholders need to watch live?** If yes, stay with live video: Discuss.io, Recollective or Lookback. If they mainly read the findings, an AI interviewer saves the most time.\n\n**One-to-one or group?** For large groups reacting together, Remesh. For individual stories, an interview tool.\n\n**How do you want to buy?** Remesh, Outset, Conveo and Listen Labs sell through sales. Lookback, Recollective and Koji publish prices, and Koji lets you run a real study before you talk to anyone.\n\n## Moving a live video study to AI interviews\n\nMost teams do not switch everything at once. A practical first step is to take one recurring study, such as churn or onboarding interviews, and run it with an AI interviewer alongside your next live round. Use the same guide in both. Compare the themes, the depth of the quotes and the time it took. If the AI round finds the same story, move that study over and keep live video for the sessions where watching matters.\n\n## Try Koji on your next study\n\nPick the question you would normally book five live sessions for. [Start a free study on Koji](https://www.koji.so), [preview the interview yourself](/docs/previewing-your-interview), and send it to your customers. You will have voice and text interviews, and a report you can check against the quotes, without filling a single moderator slot.\n","category":"Comparisons","lastModified":"2026-10-30T06:27:54.431368+00:00","metaTitle":"Discuss.io Alternatives (2026): 7 Live and AI Qual Tools","metaDescription":"Compare 7 Discuss.io alternatives for live video and AI-moderated qualitative research: Koji, Remesh, Recollective, Lookback, Outset, Conveo and Listen Labs.","keywords":["discuss.io alternatives","discuss io competitors","live video qualitative research","online focus group platform","ai moderated interviews","remesh alternatives","qualitative research tools"],"aiSummary":"The best Discuss.io alternative depends on what you want to replace. For live video interviews and focus groups with observers, Recollective and Lookback are closest; Lookback publishes plans from $299 a year and Recollective offers live-only video at $2 a minute or less. For large live group conversations, Remesh. To stop scheduling moderators, AI-moderated interview platforms: Koji, Outset, Conveo and Listen Labs. Koji is self-serve with published prices (free start with 10 credits, Insights at 29 euros a month, Interviews at 79 euros a month) and runs voice, text and phone interviews in 30+ languages, but has no live video rooms.","aiKeywords":["Discuss.io","Koji","Remesh","Recollective","Lookback","Outset","Conveo","Listen Labs","live video qualitative research","AI-moderated interviews"],"aiContentType":"comparison","faqItems":[{"answer":"It depends on what you need. Recollective and Lookback are closest for live video sessions with observers, Remesh suits large live group conversations, and Koji, Outset, Conveo and Listen Labs replace the human moderator with an AI interviewer. Koji is the only one of the AI options with published self-serve pricing.","question":"What is the best alternative to Discuss.io?"},{"answer":"Discuss.io does not publish pricing and sells through its sales team. Reviewers on Capterra describe it as priced like a service rather than a self-serve tool.","question":"How much does Discuss.io cost?"},{"answer":"For many questions, such as why customers churn or what they used before, yes: an AI interviewer can ask follow-ups by voice or text without scheduling. When stakeholders need to watch reactions live, or when group dynamics are the point, keep a live video or live group tool.","question":"Can AI interviews replace live video focus groups?"},{"answer":"Lookback lists annual plans from $299 a year for 10 sessions, Recollective lists a live-only option at $2 a minute or less for video, and Koji starts free with 10 credits, with paid plans at 29 and 79 euros a month. Remesh, Outset, Conveo and Listen Labs quote through sales.","question":"Which Discuss.io alternatives publish their prices?"},{"answer":"No. Koji runs AI-moderated interviews by voice, text and phone in 30+ languages. Teams that need live video often keep a video tool for those sessions and use Koji for interviews where watching is not needed.","question":"Does Koji support video interviews?"}],"relatedTopics":["Discuss.io alternatives","Live video qualitative research","Online focus groups","AI-moderated interviews","Qualitative research tools","Remesh alternatives"]},{"type":"blog","id":"09545e54-7f54-4c4e-9666-b13104877c15","slug":"listen-labs-vs-usertesting-2026","title":"Listen Labs vs UserTesting (2026): AI Interviews or Recorded Usability Tests?","url":"https://www.koji.so/blog/listen-labs-vs-usertesting-2026","summary":"Listen Labs and UserTesting answer different research questions. Listen Labs runs AI-moderated interviews by video, voice and text, with a recruiting network Salesforce puts above 50 million people; Salesforce signed a deal to acquire it on September 29, 2026. UserTesting records think-out-loud usability sessions and live moderated sessions with AI insight summaries; its parent company was renamed Auros the same day and has a 7.6 million person network. Neither publishes prices. Choose UserTesting to watch people use a product, Listen Labs for enterprise-scale interview studies, and Koji for self-serve AI interviews by voice, text or phone on published prices.","content":"**Short answer:** Listen Labs and UserTesting solve different problems. Listen Labs runs AI-moderated interviews: an AI interviewer talks to participants by video, voice or text, asks follow-up questions and writes up the findings. UserTesting records people using your product or prototype while they think out loud, and adds live moderated sessions and AI summaries on top. Pick Listen Labs when you need to understand why people choose, switch or churn, at volume. Pick UserTesting when you need to watch people struggle with a screen. Both are enterprise contracts with no public price list, and both changed shape on the same day in 2026. If you want AI interviews you can start this week on published prices, with your own customers, [Koji](https://www.koji.so) is the third option worth a look.\n\n## One sold, one renamed, on the same day\n\nOn September 29, 2026, both companies made news. Salesforce [signed a definitive agreement to acquire Listen Labs](https://www.salesforce.com/news/stories/salesforce-signs-definitive-agreement-to-acquire-listen-labs/), describing it as an AI customer research platform whose recruiting network reaches more than 50 million potential participants, with interviews possible in more than 120 languages. Salesforce plans to bring it into Marketing Cloud and Service Cloud and expects the deal to close in the fourth quarter of its fiscal 2027, subject to regulatory approval. Terms were not disclosed; [press reports put the price at around $2 billion](https://siliconangle.com/2026/10/01/salesforce-to-acquire-ai-customer-research-startup-listen-labs-in-reported-2b-deal/).\n\nThe same day, UserTesting [announced it was renaming the company Auros](https://www.constellationr.com/insights/news/usertesting-rebrands-auros-aims-bring-humans-ai-experience). Auros is now the parent of UserTesting and User Interviews, which it bought in January 2026, plus a growing business supplying people to train and evaluate AI models. The company puts its network at 7.6 million verified people and says existing contracts and day-to-day operations are unchanged. UserTesting stays as a product name.\n\nNeither change breaks anything today. But both mean the tool you buy in 2026 sits inside a bigger strategy than it did a year ago. Ask each vendor what your contract looks like at renewal.\n\n## Why this choice comes up more often now\n\nResearch teams are being asked for more with the same headcount. In Maze's [Future of User Research Report 2026](https://maze.co/blog/future-user-research-2026/), 66% of respondents said demand for research had grown, up from 55% the year before, and 69% said they use AI in at least some of their research projects. A year earlier, [Maze's 2025 report](https://maze.co/resources/user-research-report-2025/) found time and bandwidth were the top challenge for 63% of teams.\n\nThat pressure pushes teams toward tools that run sessions for them. Listen Labs and UserTesting both promise that, in very different ways.\n\n## Listen Labs vs UserTesting at a glance\n\n|  | Listen Labs | UserTesting | Koji |\n|---|---|---|---|\n| Core method | AI-moderated interviews by video, voice and text | Recorded think-out-loud tests and live moderated sessions | AI-moderated interviews by voice, text and phone |\n| Who asks the questions | An AI interviewer that adapts its follow-ups | Your written tasks, or you, live | An AI interviewer, plus six structured question types |\n| Participant network | 50M+ potential participants, per Salesforce | 7.6M verified people across Auros brands | Your own customers, plus a research panel |\n| AI analysis | Agents for study design and synthesis | AI insight summaries grouped into themes | Reports with each finding traced to quotes |\n| 2026 ownership news | Salesforce deal signed, not yet closed | Parent company renamed Auros | Independent |\n| Public price list | No | No | Yes, free to start |\n\n![Table comparing Listen Labs, UserTesting and Koji on research method, who asks the questions, participant network, AI analysis, 2026 news and pricing](/visuals/listen-labs-vs-usertesting-vs-koji-comparison-table/listen-labs-vs-usertesting-vs-koji-comparison-table.png \"Three different answers to the question of who runs the session · Source: Salesforce and Auros announcements, September 29, 2026\")\n\n## What Listen Labs does well\n\n**Interviews that adapt.** Listen Labs' AI interviewer asks an opening question and then follows the participant's answer, the way a good moderator would. That makes it a tool for \"why\" questions: why people picked a competitor, why a feature never caught on, how a buyer describes the problem in their own words.\n\n**Scale and reach.** A recruiting network that Salesforce puts above 50 million people, and more than 120 languages, makes multi-country consumer studies practical in days rather than months.\n\n**Money and momentum.** Listen Labs [raised a $69 million Series B in January 2026](https://pulse2.com/listen-labs-69-million-series-b/), led by Ribbit Capital with Sequoia, Conviction and Pear VC, and said it had run more than one million interviews.\n\n**Where it is weaker.** It is sold through a demo and pilot, not self-serve. Listen Labs' [own pricing guide](https://listenlabs.ai/articles/ai-customer-research-pricing-2026) puts per-interview pricing for most enterprise customers at $85 to $150, before platform access. And with the Salesforce deal pending, its packaging after closing is an open question.\n\n## What UserTesting does well\n\n**Watching behaviour.** UserTesting's core is the recorded session: a participant works through tasks on your site, app or prototype while saying what they think, and you get the video, a synced transcript, clips and highlight reels. Nothing in an interview replaces seeing someone miss the button you thought was obvious.\n\n**Live sessions when you need them.** [Live Conversation](https://help.usertesting.com/hc/en-us/articles/38017055971997) lets a researcher moderate in real time, and newer AI summaries are built from the transcript at session and study level.\n\n**AI summaries tied to evidence.** On completed tests, [AI insight summaries](https://help.usertesting.com/hc/en-us/articles/35103184242205) group responses into themes and link each theme back to the sessions behind it.\n\n**Design workflow.** In January 2026 it launched [UserTesting for Figma](https://www.businesswire.com/news/home/20260127038899/en/UserTesting-Unveils-UserTesting-for-Figma-Embedding-AI-Powered-Customer-Insights-Directly-into-Design-Workflows), which generates test plans from prototypes.\n\n**Where it is weaker.** Sessions are scripted tasks, so you get what people do and say, not a probing conversation about their motives. Pricing is not published: [Vendr's purchase data](https://www.vendr.com/marketplace/usertesting) puts typical contracts between about $12,000 and $113,642 a year, with a median around $40,000. And its [press release on the User Interviews deal](https://www.usertesting.com/company/newsroom/press-releases/usertesting-acquires-user-interviews) shows who it is built for: more than 3,000 customers, including 75 of the Fortune 100.\n\n## When you need to watch people use the product\n\nSome questions can only be answered by watching. Can a new customer finish onboarding without help? Do people understand the pricing page? Where does checkout lose them? For these, UserTesting's recorded sessions are the stronger tool, and an interview, AI or human, is a weak substitute. People are poor reporters of their own clicks.\n\n![A person working through a task on a laptop, the kind of moment a recorded usability session is built to capture](https://images.unsplash.com/photo-1758873271835-0ab4b611e659?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHw5fHxwZXJzb24lMjB1c2luZyUyMGxhcHRvcCUyMHRoaW5raW5nfGVufDF8MHx8fDE3OTI5OTc0MTB8MA&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Vitaly Gariev](https://unsplash.com/@silverkblack?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\nOther questions can only be answered by asking. Why did a customer cancel? What were they using before? What would make them pay more? Here a screen recording tells you very little. You need someone to ask, listen and ask again, which is what Listen Labs and Koji do.\n\nMost product teams need both, at different moments. The mistake is buying one platform and forcing every question through it.\n\n## Questions to ask either vendor before you sign\n\nBoth companies sell annual contracts, and both are in the middle of a corporate change. Get answers in writing.\n\n![Checklist of questions to ask before signing an AI interview platform contract, covering billing, quality, recruitment, data and exit terms.](/visuals/ai-interview-platform-contract-questions-checklist/ai-interview-platform-contract-questions-checklist.png \"Run both quotes through the same list, then add one question: what changes at renewal after the September news?\")\n\nFor Listen Labs, also ask whether you will need other Salesforce products after the deal closes. For UserTesting, ask which participant network your plan draws on now that User Interviews sits under the same roof.\n\n## Where Koji fits\n\nKoji is an AI interview platform like Listen Labs, sold the way most software is: you sign up and start. It suits teams who want the \"why\" without a five-figure contract.\n\n![Koji voice interview in progress with an AI orb asking a follow-up question and a microphone button](/email/shots/shot-voice-interview.jpg \"A Koji voice interview: the AI interviewer asks a follow-up based on the last answer\")\n\n- **Published prices.** You start free with 10 credits. Insights is €29 a month with 29 credits, Interviews is €79 a month with 79 credits, and Enterprise is priced on request. A text interview uses 1 credit, a voice interview 3, and a phone interview 0.6 credits a minute.\n- **You pay for usable interviews only.** Every interview gets a quality score from 1 to 5, and only those scoring 3 or more use credits or enter the report. See [understanding quality scores](/docs/understanding-quality-scores).\n- **Voice, text and phone.** Participants choose how to answer, in [30+ interview languages](/docs/multilingual-research-guide). Every question, including answer options and scale labels, is translated when you publish.\n- **Your customers or a panel.** Interview your own users with a link or personal invites, or [recruit from Koji's research panel](/docs/panel-recruitment) on paid plans.\n- **Reports you can check.** Findings link to the quotes behind them, and you can [preview your interview](/docs/previewing-your-interview) for free before launch.\n\nWhat Koji does not do: video interviews or recorded screen sessions. If you need to watch someone use a prototype, pair Koji with a usability tool rather than expecting one platform to do both.\n\nFor a head-to-head on the AI interview side, read [Koji vs Listen Labs](/docs/koji-vs-listen-labs).\n\n## How to choose\n\n- **You need to see people use a product or prototype:** UserTesting.\n- **You are an enterprise insights team running large multi-country interview studies, and Salesforce is your stack:** Listen Labs.\n- **You want AI interviews with your own customers this week, on prices you can read:** Koji.\n\n## Try it on one question\n\nThe fastest way to decide is to run one real study. Write the question you most need answered, send it to 20 customers, and read the report. [Start a free study on Koji](https://www.koji.so) and you will have interviews back the same day.\n","category":"Comparisons","lastModified":"2026-10-30T06:27:53.111072+00:00","metaTitle":"Listen Labs vs UserTesting (2026): Which Should You Buy?","metaDescription":"Listen Labs runs AI interviews; UserTesting records usability sessions. Compare methods, panels, AI analysis, pricing and the September 2026 news.","keywords":["listen labs vs usertesting","usertesting vs listen labs","listen labs alternatives","usertesting alternatives","auros usertesting","ai moderated interviews","usability testing platform"],"aiSummary":"Listen Labs and UserTesting answer different research questions. Listen Labs runs AI-moderated interviews by video, voice and text, with a recruiting network Salesforce puts above 50 million people; Salesforce signed a deal to acquire it on September 29, 2026. UserTesting records think-out-loud usability sessions and live moderated sessions with AI insight summaries; its parent company was renamed Auros the same day and has a 7.6 million person network. Neither publishes prices. Choose UserTesting to watch people use a product, Listen Labs for enterprise-scale interview studies, and Koji for self-serve AI interviews by voice, text or phone on published prices.","aiKeywords":["Listen Labs","UserTesting","Auros","Salesforce acquisition","AI-moderated interviews","usability testing","Koji"],"aiContentType":"comparison","faqItems":[{"answer":"Listen Labs runs AI-moderated interviews, where an AI interviewer asks questions and follow-ups by video, voice or text. UserTesting records people completing tasks on your product while they think out loud, and also supports live moderated sessions. Listen Labs is built for understanding motives; UserTesting is built for watching behaviour.","question":"What is the difference between Listen Labs and UserTesting?"},{"answer":"Salesforce signed a definitive agreement to acquire Listen Labs on September 29, 2026. It expects the deal to close in the fourth quarter of its fiscal 2027, subject to regulatory approval. Terms were not disclosed.","question":"Did Salesforce buy Listen Labs?"},{"answer":"The company renamed itself Auros on September 29, 2026. Auros is the parent of the UserTesting and User Interviews products, which keep their names, and the company says existing contracts are unchanged.","question":"Is UserTesting now called Auros?"},{"answer":"Neither publishes a price list. Listen Labs' own pricing guide puts per-interview pricing for most enterprise customers at $85 to $150 before platform access. Vendr purchase data puts typical UserTesting contracts between about $12,000 and $113,642 a year, with a median around $40,000.","question":"How much do Listen Labs and UserTesting cost?"},{"answer":"Koji runs AI-moderated interviews by voice, text and phone on published prices. You start free with 10 credits; paid plans are Insights at 29 euros a month and Interviews at 79 euros a month, and only interviews with a quality score of 3 or more use credits. Koji does not run video interviews or recorded usability sessions.","question":"Is there a self-serve alternative to both?"}],"relatedTopics":["Listen Labs alternatives","UserTesting alternatives","AI-moderated interviews","Usability testing","Participant recruitment","Research platform pricing"]},{"type":"blog","id":"63c1235e-baf5-4c9a-a103-ef6aadba1e60","slug":"koji-vs-getwhy-2026","title":"Koji vs GetWhy (2026): Managed Video Studies or Self-Serve AI Interviews?","url":"https://www.koji.so/blog/koji-vs-getwhy-2026","summary":"Comparison of GetWhy and Koji, two AI qualitative research platforms. GetWhy (Copenhagen) runs AI-moderated video interviews with a multi-vendor panel it says reaches 300M people, with researchers reviewing each study; it sells through sales, with no public prices. Koji is self-serve: AI-moderated voice, text and phone interviews in 30+ languages, screening questions, a 1 to 5 quality score where only interviews scoring 3+ use credits, reports with findings traceable to quotes, and public plans (free with 10 credits, Insights EUR 29/month, Interviews EUR 79/month, Enterprise on request). Choose GetWhy for managed video studies at global brands; choose Koji to run your own interviews often at a known price.","content":"**Short answer:** GetWhy and Koji both use AI to run qualitative interviews, but they are built for different buyers. GetWhy is a sales-led platform for enterprise consumer brands: AI-moderated video interviews, a large multi-vendor panel and PhD-level researchers checking each study, sold per study through a sales conversation. Koji is self-serve: you write a research question, an AI interviewer runs voice, text or phone interviews in 30+ languages, and you get a report with every finding linked to the quotes behind it. Plans are public, from a free account to €79 a month. Pick GetWhy if you are a global brand team that wants video evidence and expert review on each project. Pick Koji if you want to run your own studies this week, as often as you like, at a price you can see before you start.\n\n## Koji vs GetWhy at a glance\n\n| | GetWhy | Koji |\n|---|---|---|\n| Interview format | AI-moderated video interviews | AI voice and text interviews, plus phone calls |\n| Who runs the study | GetWhy's platform, with its researchers reviewing each study | You, with an AI assistant that drafts the brief |\n| Participants | Multi-vendor panel GetWhy says reaches 300M people | Your own customers by link or CSV, or Koji's research panel |\n| Languages | 100+ languages and markets, per GetWhy | 30+ interview languages |\n| Turnaround | 48 hours from brief to insight, per GetWhy | Results update as each interview finishes |\n| Buying model | Contact sales, priced per study | Self-serve, starts free, public pricing |\n| Best fit | Enterprise consumer and CPG brands | Product, research and marketing teams of any size |\n\n![Table comparing GetWhy and Koji on interview format, who runs the study, participants, languages, turnaround, buying model and best fit](/visuals/koji-vs-getwhy-comparison-table/koji-vs-getwhy-comparison-table.png \"The short version: GetWhy sells managed video studies, Koji lets you run your own interviews\")\n\n## Why this choice is coming up more often\n\nAI moved from experiment to default in market research quickly. In Qualtrics' 2025 Market Research Trends report, based on a survey of about 3,200 researchers, **89% said they already use AI tools** regularly or experimentally, and **83% planned to significantly increase AI investment** in 2025 ([Qualtrics](https://www.qualtrics.com/articles/news/ai-to-drive-massive-changes-to-market-research-in-2025-qualtrics-report-says/)). The same report found **71% expect synthetic responses to make up more than half of data collection within three years**.\n\nThat last number is why platforms that still talk to real people matter. GetWhy and Koji both bet on real participants and use AI to make the conversations cheaper and faster, not to replace them. The industry they compete in is big: ESOMAR put global market research turnover at almost **$130 billion** after 12% growth in 2022 ([ESOMAR Global Market Research 2023](https://community.esomar.org/knowledge-center/library/Global-Market-Research-2023-pub2990)). A lot of that money still pays for weeks-long qualitative projects run by agencies, which is the spend both tools go after.\n\n## What GetWhy does\n\nGetWhy is a Copenhagen company. It describes its product as an AI research platform that \"designs studies, recruits participants, and runs AI-moderated video interviews\", with insights \"in days, not weeks\" ([getwhy.io](https://www.getwhy.io/)). The workflow runs from study design to recruitment, AI-moderated interviews, synthesis and what GetWhy calls activation. It says PhD-level researchers review and validate each study, so it is AI moderation with expert oversight rather than human moderators.\n\nWhat it says about itself:\n\n- **Video first.** Interviews are on video, and clips come with the findings. In February 2026 it launched **Stories**, which package findings as articles, videos and podcasts for internal audiences ([GetWhy blog](https://www.getwhy.io/blog/introducing-the-next-generation-of-ai-qualitative-research)).\n- **A big panel.** GetWhy says its multi-vendor recruitment reaches 300 million participants, with verification and fraud checks, and Respondent lists it as a partner ([Respondent](https://www.respondent.io/partners/getwhy)).\n- **Speed.** Its headline claim is 48 hours from brief to insight.\n- **Enterprise customers.** Press coverage of its funding names brands such as Nike, Heineken, Unilever, Carlsberg and The Coca-Cola Company.\n- **Funding.** A $34.5M Series A led by PeakSpan Capital in June 2024 ([Research Live](https://www.research-live.com/article/news/345m-funding-injection-for-getwhy/id/5126913)), DKK 60M of debt from CIBC Innovation Banking in February 2025, and a €17M extension in July 2025 ([PeakSpan](https://www.peakspancapital.com/resources/news/copenhagens-getwhy-raises-eu17m-to-scale-ai-powered-consumer-insights-eyes-us-expansion)).\n- **Enterprise plumbing.** It claims ISO/IEC 27001 compliance and storage in Europe ([GetWhy security](https://getwhy.io/security)), and in June 2026 announced a partnership with Market Logic to feed Stories into Market Logic's insights platform ([Market Logic](https://marketlogicsoftware.com/news/2026/getwhy-and-market-logic-partner-to-deliver-the-full-enterprise-insights-experience/)).\n\n**Pricing.** GetWhy does not publish prices. Its own comparison page against Kantar and Ipsos puts a study at $2,000 to $4,000, against $15,000 to $40,000 for a traditional agency project ([GetWhy](https://www.getwhy.io/articles/getwhy-vs-kantar-ipsos)). Treat that as the vendor's estimate; your quote will depend on markets, audience and volume.\n\n**Limits worth knowing.** Independent reviews are thin: G2 says there are not enough reviews to give buying insight, and the one recurring con is that setup and team training take time. The 48-hour, 300M-participant and 100+ language figures are all self-reported. And because every study goes through GetWhy's sales and research process, it suits planned projects better than quick questions from a product team.\n\n## What Koji does\n\nKoji is an AI-native research platform you run yourself. You type what you want to learn, the research assistant drafts a brief with questions, and an AI interviewer holds the conversations. What it does today, checked against the product this week:\n\n- **Voice, text and phone interviews.** Participants choose to talk or type on the interview page, and you can also send a call-back link or have Koji call a list of numbers. Interviews run in [30+ languages](/docs/multilingual-research-guide). There is no video: if you need faces on camera, Koji is not the tool.\n- **Open questions with structure.** Mix open questions, which the AI follows up on, with scales, single and multiple choice, ranking and yes/no questions in the same interview.\n- **Screening before the interview.** [Screening questions](/docs/adding-screening-questions-to-your-study) turn away people who don't qualify before the conversation starts, whatever link they came in on.\n- **Your participants or a panel.** Share a link, [import a CSV](/docs/importing-participants-csv) to give each person their own invite link (Interviews and Enterprise plans), or [recruit from Koji's research panel](/docs/panel-recruitment) on paid plans, choosing text, voice or both.\n- **A quality gate.** Every interview gets a [quality score from 1 to 5](/docs/understanding-quality-scores). Only interviews that score 3 or more use credits or enter the report, so a participant who clicks through with one-word answers costs you nothing.\n- **Reports you can check.** Findings, themes and personas, each finding [linked to the quotes behind it](/docs/tracing-report-findings-to-evidence), updating as interviews come in. Share a [public report](/docs/publishing-sharing-reports) with stakeholders.\n- **Free previews.** [Take your own interview](/docs/previewing-your-interview) by text or voice before launch. Previews never use credits.\n- **Works where you work.** An API, an embeddable interview and an [MCP connector](/docs/mcp-overview) for running studies from Claude, Cursor and other AI assistants.\n\n**Pricing.** A free account comes with 10 credits. Insights is €29 a month with 29 credits, Interviews is €79 a month with 79 credits, and Enterprise is priced on request. A text interview uses 1 credit and a voice interview 3; phone calls use 0.6 credits a minute. Extra credits are €1 each.\n\n![Koji voice interview in progress with an AI orb asking a follow-up question and a microphone button](/email/shots/shot-voice-interview.jpg \"A Koji voice interview: the AI asks a follow-up based on what the participant just said\")\n\n## Where they differ\n\n### Video versus voice and text\n\nThis is the biggest difference. GetWhy's whole proposition rests on video: clips of real consumers that a brand team can play in a meeting. If your stakeholders need to see faces, GetWhy does that and Koji doesn't.\n\n![A person on a commute talking into their phone with headphones, the kind of moment a voice interview fits into](https://images.unsplash.com/photo-1719922327752-46a8d31526bf?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHwxfHxwZXJzb24lMjB0YWxraW5nJTIwcGhvbmUlMjBoZWFkcGhvbmVzJTIwY29tbXV0ZXxlbnwxfDB8fHwxNzkxOTU3MDEwfDA&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Taiki Ishikawa](https://unsplash.com/@fl__q?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\nKoji bets on voice and text because they are easier for participants. People can answer on a phone during a commute, in a language they choose, without setting up a camera or worrying about how they look. That asks less of people, and it makes it easier to reach B2B users, who rarely want to be filmed. You still get verbatim quotes for every finding.\n\n### Managed studies versus self-serve\n\nGetWhy sells a study with researchers attached. That is useful when a brand team runs a few big projects a year and wants someone else to own the method.\n\nKoji gives the method to you. A product manager can go from a question to a live interview in an afternoon, read results the same day and run the next study the following week. Product teams that want to talk to customers every week, not every quarter, tend to need this.\n\n### Panel reach\n\nGetWhy's recruitment reach is its strength, especially for consumer studies across many countries. Koji's panel lets you describe your audience in plain language and target by criteria such as age and employment, and you see the cost per interview before you launch. Credits are held at launch and unused credits come back when the recruitment ends. For hard-to-reach B2B buyers, interviewing your own users through a link or CSV is often the better route.\n\n### How you buy\n\nGetWhy needs a sales conversation, and you get a price per study. Koji publishes its prices, starts free and charges only for interviews that pass the quality gate. You can work out what a 50-interview voice study costs before you sign up: 150 credits.\n\n## Before you sign with either\n\nWhatever you pick, put the same questions to both vendors. Ask for a free test interview, ask how low-effort answers are handled and whether you pay for them, and ask to trace one finding back to the raw quotes.\n\n![Checklist of eight questions to ask before choosing an AI interview platform, from free previews to data storage](/visuals/ai-interview-platform-vendor-checklist/ai-interview-platform-vendor-checklist.png \"Eight questions that separate a demo from a platform you can rely on\")\n\n## Who should choose which\n\n**Choose GetWhy if:**\n\n- You are an enterprise consumer brand running concept, packaging or ad studies across many markets.\n- Your stakeholders want video clips of consumers in the readout.\n- You prefer a managed study with researchers reviewing the work, and you buy research per project.\n\n**Choose Koji if:**\n\n- You want to run interviews yourself, quickly and often, without a sales cycle.\n- Your participants are your own customers or users, including B2B buyers who won't go on camera.\n- You need to mix open conversation with scales and choice questions, and want each finding traceable to quotes.\n- You want to see the price before you start and pay only for interviews that pass a quality check.\n\n## The bottom line\n\nGetWhy is a serious, well-funded option for global brands that want video-based qualitative studies with expert oversight. Koji is the faster and more open choice for teams that want to run their own AI interviews: voice, text or phone, 30+ languages, a quality gate on every interview and public prices from free.\n\n## Try Koji free\n\nThe best way to compare is to run a study. [Create a free Koji account](https://www.koji.so), type your research question, preview the interview yourself by voice or text, and send it to ten customers. Your 10 signup credits cover ten text interviews or three voice interviews, enough to see what the report looks like before you spend anything. For more on the method, read our [AI interview best practices](/docs/ai-interview-best-practices) or compare Koji with [Conveo](/blog/koji-vs-conveo-2026), another video-first platform.\n","category":"Comparisons","lastModified":"2026-10-30T06:27:38.347059+00:00","metaTitle":"Koji vs GetWhy 2026: AI Interview Platforms Compared","metaDescription":"GetWhy runs managed AI video studies for enterprise brands. Koji lets you run voice, text and phone AI interviews yourself, from free. Features and pricing.","keywords":["koji vs getwhy","getwhy alternative","getwhy pricing","getwhy review","ai video interviews","ai moderated interviews","ai consumer insights"],"aiSummary":"Comparison of GetWhy and Koji, two AI qualitative research platforms. GetWhy (Copenhagen) runs AI-moderated video interviews with a multi-vendor panel it says reaches 300M people, with researchers reviewing each study; it sells through sales, with no public prices. Koji is self-serve: AI-moderated voice, text and phone interviews in 30+ languages, screening questions, a 1 to 5 quality score where only interviews scoring 3+ use credits, reports with findings traceable to quotes, and public plans (free with 10 credits, Insights EUR 29/month, Interviews EUR 79/month, Enterprise on request). Choose GetWhy for managed video studies at global brands; choose Koji to run your own interviews often at a known price.","aiKeywords":["getwhy","getwhy alternative","ai video interviews","ai moderated interviews","consumer insights platform","qualitative research at scale","ai voice interviews","panel recruitment","research pricing","koji"],"aiContentType":"comparison","faqItems":[{"answer":"GetWhy is a Copenhagen-based AI research platform for consumer brands. It designs studies, recruits participants from a multi-vendor panel and runs AI-moderated video interviews, with researchers reviewing each study. It says it delivers insights within 48 hours.","question":"What is GetWhy?"},{"answer":"GetWhy does not publish prices; you get a quote from sales. Its own comparison with Kantar and Ipsos puts a study at $2,000 to $4,000. Koji publishes its prices: free with 10 credits, Insights at €29 a month, Interviews at €79 a month and Enterprise on request.","question":"How much does GetWhy cost?"},{"answer":"No. Koji runs AI-moderated voice and text interviews and phone call interviews, not video. If your stakeholders need video clips of consumers, GetWhy is built for that. Koji suits teams that want participants to answer easily from any device, in 30+ languages.","question":"Does Koji do video interviews like GetWhy?"},{"answer":"Koji, in most cases. A product manager can draft a study with the AI assistant, preview it for free, send it to their own users or Koji's panel and read a report the same day, without a sales process. GetWhy suits planned brand studies with expert oversight.","question":"Which is better for product teams?"},{"answer":"Yes. On paid plans you can recruit from Koji's research panel and see the cost per interview before launch. You can also share a link, or import a CSV so each person gets a personal invite link on Interviews and Enterprise plans.","question":"Can I recruit participants with Koji?"}],"relatedTopics":["ai moderated interviews","getwhy alternative","ai video research","consumer insights","research pricing"]},{"type":"blog","id":"2b6dff5e-46f9-45cf-b40f-77c975a75706","slug":"outset-alternatives-2026","title":"7 Best Outset Alternatives in 2026 (Pricing and Features Compared)","url":"https://www.koji.so/blog/outset-alternatives-2026","summary":"The best Outset alternatives in 2026 are Koji, Listen Labs, Strella, Conveo, Great Question, Maze and UserTesting. Outset runs AI-moderated video, voice and text interviews in 40+ languages but has custom-only pricing, no free plan and relies on 25+ external panels. Koji is the closest self-serve swap, with published pricing from free (Insights EUR 29/mo, Interviews EUR 79/mo), a built-in panel in 130+ countries and a quality gate that only charges for interviews scoring 3 or more; it does not record video. Strella and Conveo suit teams that need recorded video, and Listen Labs is being acquired by Salesforce.","content":"The best Outset alternatives in 2026 are **Koji, Listen Labs, Strella, Conveo, Great Question, Maze and UserTesting**. Which one fits depends on why you are leaving or skipping Outset. If the problem is price and the annual sales contract, Koji is the closest like-for-like swap: it runs AI-moderated voice and text interviews, recruits from a panel, writes the report, and publishes its prices, starting with a free plan. If you need recorded video of every participant for an enterprise insights team, Conveo and Strella are the nearest matches.\n\nOutset is a strong product. It has raised $51M, counts Microsoft and Nestlé as customers, and runs AI-moderated interviews in more than 40 languages. But it is built and priced for enterprise insights teams, and many product teams, founders and agencies find that they need the method without the procurement cycle.\n\n## Why teams look for Outset alternatives\n\nThree reasons come up again and again.\n\n**1. Pricing is not public.** Outset's [pricing page](https://outset.ai/pricing) lists custom plans (a standard subscription, support tiers and enterprise plans) and no numbers. There is no free plan to try it on your own. Third-party estimates exist, but we could not verify them, so expect a sales call and an annual contract before you run your first study.\n\n**2. It has no panel of its own.** Outset [connects to more than 25 panels](https://outset.ai/platform/recruit), including Prolific, User Interviews and Respondent, or you can bring your own customers. That is flexible, but it means a second vendor relationship and a second invoice for most studies.\n\n**3. It is heading towards enterprise CX.** When Outset announced its [$30M Series B](https://outset.ai/series-b) in December 2025, led by Radical Ventures, it also described a move into an \"AI-native Customer Experience Management platform.\" That is good news for large insights teams. It is less good news if you are a three-person product team that wants to talk to 40 customers this week.\n\nThere is also a market reason to look now. Interest in AI-moderated research is very high. Qualtrics' [2026 Market Research Trends report](https://www.qualtrics.com/articles/strategy-research/market-research-trends), based on more than 3,000 researchers in 17 countries, found that **95% of researchers already use AI tools regularly or are experimenting with them**, and **78% expect AI agents to run more than half of all research projects by 2028**. With that much demand, every vendor in this category is changing fast. Listen Labs, one of Outset's closest rivals, has just agreed to be bought by Salesforce. Re-checking your shortlist this quarter is sensible.\n\n## Outset alternatives at a glance\n\n| Tool | Interview modes | Recruiting | Public pricing | Best for |\n| --- | --- | --- | --- | --- |\n| **Koji** | AI voice and text | Built-in panel (130+ countries) or your own users | Yes: Free, €29, €79, Enterprise from €500/mo | Product teams, founders, agencies who want self-serve |\n| **Listen Labs** | AI voice, video, text | Large built-in network | No | Enterprise teams already on Salesforce |\n| **Strella** | AI voice with webcam and screen | Through User Interviews | No (usage-based) | Teams that want recorded sessions and highlight reels |\n| **Conveo** | AI video, audio, text | Panel incentives included on paid plan | Partly (Starter $1,250/mo billed annually) | Consumer brands and CPG |\n| **Great Question** | Human-moderated, AI moderation as add-on | User Interviews panel or your CRM | Yes ($129/seat/mo self-serve) | Research ops and panel management |\n| **Maze** | Unmoderated tests, AI moderator on Enterprise | Maze panel | Partly | Design teams doing prototype testing |\n| **UserTesting** | Recorded think-aloud, AI follow-ups in early access | Own network plus User Interviews | No | Large UX teams with existing contracts |\n\n## 1. Koji: the self-serve Outset alternative\n\n[Koji](https://www.koji.so) is an AI-native research platform. You describe what you want to learn, the research brief assistant drafts the interview, and an AI interviewer runs each conversation by voice or text, asking follow-up questions in real time. Koji then scores every interview, builds an aggregated report with themes, sentiment and quotes, and lets you share it as a public link.\n\nWhat makes it a strong Outset replacement:\n\n- **Published pricing.** Free to start (10 one-time credits), Insights at €29/month, Interviews at €79/month and Enterprise from €500/month. A text interview costs 1 credit, a voice interview costs 3, and extra credits are €1 on self-serve plans. See the [plan comparison guide](/docs/plan-comparison-guide).\n- **You only pay for useful interviews.** Every interview gets a quality score, and only interviews scoring 3 or more use credits or enter the report. Since September 2026 the score is capped by how much the participant actually said, so filler answers do not earn a top score. Read [how quality scores work](/docs/understanding-quality-scores).\n- **A panel inside the product.** From the Panel recruit tab you describe the audience, see the price before launch, and choose text, voice or both. Credits are held at launch and unused credits return at the end. A soft launch fills part of the goal first so you can check quality before releasing the rest, and accepted reports of low-effort or fraudulent completes are refunded. See [panel recruitment](/docs/panel-recruitment).\n- **Screening on every route.** [Screening questions](/docs/adding-screening-questions-to-your-study) run before the interview starts, whether the participant came from the panel, an email link or your website embed.\n- **Free previews.** You can take your own interview by text or voice before launch, and [previews](/docs/previewing-your-interview) never use credits.\n- **30+ interview languages**, team workspaces (3 seats on Interviews, 10 on Enterprise), an API and an [MCP server](/docs/mcp-overview) on Interviews and Enterprise, and brand themes on Enterprise.\n\nWhere Outset is ahead: Koji does not record participant video, and it offers fewer interview languages than Outset's 40+. If video clips for a highlight reel are the deliverable, look at Strella or Conveo. If the deliverable is the insight itself, Koji gets you there faster and for far less. For a feature-by-feature breakdown, see [Koji vs Outset](/docs/koji-vs-outset).\n\n## 2. Listen Labs\n\nListen Labs is the best-funded company in the category. It raised a $69M Series B in January 2026, and at the end of September 2026 [Salesforce signed a definitive agreement to acquire it](https://www.salesforce.com/news/stories/salesforce-signs-definitive-agreement-to-acquire-listen-labs/). Salesforce says Listen Labs draws on a network of more than 50 million potential participants and runs interviews in more than 120 languages. The team will join Salesforce AI Labs, and the product will complement Marketing Cloud and Service Cloud. The deal is expected to close in Salesforce's fourth fiscal quarter of 2027.\n\nPricing is not public. If your company runs on Salesforce, Listen Labs may become the default choice there. If you are not a Salesforce shop, the acquisition adds uncertainty about roadmap and contract terms that is worth asking about.\n\n## 3. Strella\n\nStrella runs AI-moderated voice interviews with webcam and screen recording, embedded stimuli, AI-generated discussion guides and automatic highlight reels. It recruits through a [User Interviews integration](https://www.userinterviews.com/product-announcements/strella-recruitment-integration). Strella raised a [$14M Series A led by Bessemer](https://www.research-live.com/article/news/strella-secures-14m-investment/id/5143863) in October 2025 and lists Amazon, Duolingo and Chobani as customers.\n\nStrella describes its pricing as usage-based with no per-seat charges, but publishes no numbers, so it is a sales conversation. It is the closest match to Outset if recorded sessions matter to you.\n\n## 4. Conveo\n\nConveo, a Y Combinator company, raised a [$50M Series A](https://conveo.ai/mediakit/series-a) in September 2026, led by DST Global. Its AI interviewer asks follow-up questions and participants answer by video, audio or text. It says more than 400 enterprises use it, including Google, Unilever and AB InBev.\n\nUnusually for this group, Conveo publishes some pricing: a Free plan and a Starter plan at $1,250 a month billed annually, which includes panel incentives, with Enterprise on request. It is a natural fit for consumer brands that want video.\n\n## 5. Great Question\n\nGreat Question is a research CRM first: panel management, scheduling, incentives and a repository. Its [self-serve plan](https://greatquestion.co/pricing) is $129 per seat per month, for up to five seats, and it lists AI-moderated interviews as an add-on. Choose it if your main problem is managing who you talk to, not running interviews at scale.\n\n## 6. Maze\n\nMaze is strongest at unmoderated prototype and usability testing. It has added an AI moderator, but Maze's help centre lists it as an Enterprise feature. Good if your team already lives in Figma prototypes and wants quick task-based tests.\n\n## 7. UserTesting\n\nUserTesting remains the big name in recorded think-aloud testing, and it bought User Interviews in January 2026. Its AI moderation feature adds AI follow-up questions to unmoderated think-aloud studies and is in limited early access. Pricing is not public and contracts are annual.\n\n## How to choose\n\nStart with the job, not the vendor.\n\n- **You want to talk to customers this week without a sales call:** Koji. Sign up, write the brief, preview it, recruit from the panel or share a link.\n- **You need recorded video clips for stakeholders:** Strella or Conveo.\n- **You are an enterprise insights team on Salesforce:** Listen Labs, once the deal closes.\n- **You need a participant CRM more than an interviewer:** Great Question.\n- **You mainly test prototypes:** Maze or UserTesting.\n\nDemand for research keeps growing faster than research teams. Maze's [Future of User Research 2026](https://maze.co/blog/future-user-research-2026/) report found that **research demand rose 20% year over year**, and that **39% of product managers now produce insights themselves**. That is the real reason AI-moderated interviews matter. The best tool is the one that people outside the research team can actually use.\n\n## Try Koji as your Outset alternative\n\nKoji gives you what most teams buy Outset for, AI-moderated interviews, a built-in panel and an automatic report, with prices you can read before you talk to anyone. Start on the free plan, [preview your interview](/docs/previewing-your-interview) for free, and only pay for interviews that pass the quality gate.\n\n[Create your free Koji account](https://www.koji.so/signup) and run your first study today.","category":"Comparisons","lastModified":"2026-10-30T06:27:25.087016+00:00","metaTitle":"7 Best Outset Alternatives in 2026 | AI Interview Platforms","metaDescription":"Compare the best Outset alternatives for 2026: Koji, Listen Labs, Strella, Conveo, Great Question, Maze and UserTesting. Pricing, panels, modes and fit.","keywords":["outset alternatives","outset.ai alternatives","outset competitors","outset pricing","ai moderated interview platforms","ai interview tools 2026","outset vs koji"],"aiSummary":"The best Outset alternatives in 2026 are Koji, Listen Labs, Strella, Conveo, Great Question, Maze and UserTesting. Outset runs AI-moderated video, voice and text interviews in 40+ languages but has custom-only pricing, no free plan and relies on 25+ external panels. Koji is the closest self-serve swap, with published pricing from free (Insights EUR 29/mo, Interviews EUR 79/mo), a built-in panel in 130+ countries and a quality gate that only charges for interviews scoring 3 or more; it does not record video. Strella and Conveo suit teams that need recorded video, and Listen Labs is being acquired by Salesforce.","aiKeywords":["outset alternatives","ai moderated interviews","outset pricing","listen labs salesforce","strella","conveo","koji ai interviews","ai interview platform comparison"],"aiContentType":"comparison","faqItems":[{"answer":"The main Outset alternatives are Koji, Listen Labs, Strella, Conveo, Great Question, Maze and UserTesting. Koji is the closest self-serve option with public pricing and a built-in panel. Strella and Conveo are closest if you need recorded video, and Listen Labs suits enterprise teams on Salesforce.","question":"What are the best Outset alternatives in 2026?"},{"answer":"Outset does not publish prices. Its pricing page lists custom plans (a standard subscription, support tiers and enterprise plans), there is no free plan, and contracts are typically annual, so you need to talk to sales for a quote.","question":"How much does Outset cost?"},{"answer":"No. Outset connects to more than 25 recruiting panels, including Prolific, User Interviews and Respondent, or you can bring your own customers. Koji, by contrast, has panel recruitment built into each study, with the price shown before launch.","question":"Does Outset have its own participant panel?"},{"answer":"Yes. Koji has a free plan with 10 one-time credits, enough for up to 10 text interviews, and free interview previews. Conveo also lists a free plan. Paid Koji plans start at EUR 29 a month.","question":"Is there a free alternative to Outset?"},{"answer":"No. Koji runs AI-moderated voice and text interviews and does not record participant video. If video highlight reels are essential, Strella, Conveo or Outset are better fits.","question":"Can Koji record video interviews like Outset?"},{"answer":"On September 29, 2026 Salesforce signed a definitive agreement to acquire Listen Labs. The team will join Salesforce AI Labs and the deal is expected to close in Salesforce's fourth fiscal quarter of 2027.","question":"What happened to Listen Labs?"}],"relatedTopics":["ai moderated interviews","outset pricing","ai interview platforms","participant recruitment","listen labs acquisition"]},{"type":"blog","id":"7c03794c-b0e3-4610-9fde-e114ca85aee1","slug":"koji-vs-bolt-insight-2026","title":"Koji vs Bolt Insight (2026): Managed CPG Qual or Self-Serve AI Interviews?","url":"https://www.koji.so/blog/koji-vs-bolt-insight-2026","summary":"A comparison of Bolt Insight's BoltChatAI and Koji, two platforms that use an AI moderator for one-to-one qualitative interviews. Bolt Insight is a managed, sales-led platform for global consumer brands that drafts the guide, recruits through panel partners and reports within about a day by its own account. Koji is self-serve with public pricing, running voice, text and phone interviews in 30+ languages with a quality gate and findings linked to quotes.","content":"**Short answer:** Bolt Insight and Koji both use an AI moderator to run one-to-one qualitative interviews at scale, but they are sold to different buyers. Bolt Insight's BoltChatAI is built for global consumer brands: it designs the discussion guide, recruits through panel partners in many markets and returns a report, usually within a day by its own account, priced through a sales conversation. Koji is self-serve: you type a research question, an AI interviewer runs voice, text or phone interviews in 30+ languages with your own customers or Koji's research panel, and every finding in the report links to the quotes behind it. Plans are public, from a free account to €79 a month. Pick Bolt Insight if you are a CPG insights team running multi-country consumer studies and want a managed partner. Pick Koji if you want to run your own studies this week, as often as you need, at a price you can see before you start.\n\n## Koji vs Bolt Insight at a glance\n\n| | Bolt Insight (BoltChatAI) | Koji |\n|---|---|---|\n| Interview format | AI-moderated chat; answers by text, audio, video or images | AI voice and text interviews, plus phone calls |\n| Who builds the study | BoltChatAI drafts the guide from your objective and audience | You, with a research assistant that drafts the brief |\n| Participants | Recruited through global panel partners | Your own customers by link or CSV, or Koji's research panel |\n| Languages and markets | 50+ languages and 120+ markets on its homepage; other Bolt pages give lower counts | 30+ interview languages |\n| Turnaround | Reports in as little as 24 hours, per Bolt | Results update as each interview finishes |\n| Buying model | Contact sales, no public price list | Self-serve with public pricing |\n| Best fit | Global CPG and consumer brands | Product, research and marketing teams of any size |\n\n![Table comparing Bolt Insight and Koji on interview format, study setup, participants, languages, turnaround, buying model and best fit](/visuals/koji-vs-bolt-insight-comparison-table/koji-vs-bolt-insight-comparison-table.png \"The short version: Bolt Insight runs managed consumer studies, Koji lets you run your own interviews\")\n\n## Why AI-moderated qual is now a buying decision\n\nTwo years ago, an AI moderator was an experiment. Now most research teams use some form of AI. In Qualtrics' 2025 Market Research Trends report, a survey of more than 3,000 researchers in 14 countries, **89% said they use AI tools regularly or in experimental phases**, and **83% of organisations planned to significantly increase AI investment** ([Qualtrics](https://www.qualtrics.com/articles/news/ai-to-drive-massive-changes-to-market-research-in-2025-qualtrics-report-says/)).\n\nThe evidence that AI moderation gets more out of people is also firmer than it was. In a randomised experiment on the Verasight panel with 3,160 respondents, AI-moderated interviews produced **4.8 times as many words** in free-response questions as standard written open-ends. The same study found completion **fell from 99.4% to 40.5%**, with drop-off concentrated among people less comfortable with AI tools ([Verasight, August 2026](https://verasight.io/reports/ai-moderated-interviews-vs-survey-open-ends)). Both numbers matter when you pick a platform: you want the depth, and you want to know how a vendor handles the people who leave.\n\nMoney is following. Bolt Insight raised **£7 million in December 2025**, led by Pembroke VCT, to scale BoltChatAI internationally ([Research Live](https://www.research-live.com/article/news/bolt-insight-secures-7m-investment/id/5145333)). So the real question for a buyer is no longer whether to use an AI moderator, but which kind of vendor fits how your team works.\n\n## What Bolt Insight does\n\nBolt Insight is a London company founded by Hakan Yurdakul, who spent 14 years at Unilever, and Kerem Turgay. Its product, BoltChatAI, launched in 2022 ([Bolt Insight](https://www.boltinsight.com/bolt-insight-receives-7m-in-funding/)). What it says about itself:\n\n- **One-to-one AI moderation.** BoltChatAI runs individual interviews that probe vague answers and pick up contradictions. Participants can respond by text, audio, video or images.\n- **A guide drafted for you.** You give it a research objective and a target audience, and it drafts the discussion guide.\n- **Many markets.** Its homepage claims 50+ languages and 120+ markets. Other Bolt pages give 25+ languages across 40+ countries, so ask which languages your markets get before you commit.\n- **Panel partners.** Participants come through global panel partners rather than your own customer list.\n- **Fast reports.** Bolt says reports can arrive within about 24 hours.\n- **Newer features.** It has added Dynamic Personas and a Meta-Analysis feature for comparing trends across past qual and quant studies ([CustomerThink](https://customerthink.com/boltchatai-the-ai-driven-qualitative-research-platform-enhanced-with-two-new-features/)).\n- **CPG clients.** The company says 11 of the world's top 15 consumer packaged goods companies use it, including Unilever, Danone and Reckitt, and that it has interviewed more than 5 million consumers.\n\n**Pricing.** Bolt Insight does not publish prices. GetApp lists a starting price of US$9,950 on a flat-rate model, but that listing rests on three reviews, so treat it as a rough signal and ask for a quote ([GetApp](https://www.getapp.co.uk/software/2077871/boltchatai)).\n\n**Limits worth knowing.** Most of the scale figures above come from Bolt's own marketing, and its pages disagree on language and market counts. Independent reviews are thin. And because studies are built around panel recruitment and sold through sales, it suits planned consumer projects better than a product manager's question on a Tuesday afternoon.\n\n## What Koji does\n\nKoji is an AI-native research platform you run yourself. You type what you want to learn, the research assistant drafts a brief, and an AI interviewer holds the conversations. What it does today:\n\n- **Voice, text and phone interviews.** Participants choose to talk or type, and you can also send a call-back link or have Koji call a list of numbers. Interviews run in [30+ languages](/docs/multilingual-research-guide). Voice interviews are on the Interviews and Enterprise plans; text works on every plan.\n- **Open questions with structure.** Mix open questions, which the AI follows up on, with scales, single choice, multiple choice, ranking and yes/no questions in one interview.\n- **Screening first.** [Screening questions](/docs/adding-screening-questions-to-your-study) turn away people who don't qualify before the conversation starts.\n- **Your customers or a panel.** Share a link, [import a CSV](/docs/importing-participants-csv) so each person gets a private invite link (Interviews and Enterprise), or [recruit from Koji's research panel](/docs/panel-recruitment) on paid plans, with the cost shown before you launch.\n- **A quality gate.** Every interview gets a [quality score from 1 to 5](/docs/understanding-quality-scores). Only interviews scoring 3 or more use credits or enter the report.\n- **Reports you can check.** Findings, themes and personas, each finding [linked to the quotes behind it](/docs/tracing-report-findings-to-evidence).\n- **Free previews.** [Take your own interview](/docs/previewing-your-interview) by text or voice before launch. Previews never use credits.\n\n**Pricing.** A free account comes with 10 credits. Insights is €29 a month with 29 credits, Interviews is €79 a month with 79 credits, and Enterprise is priced on request. A text interview uses 1 credit, a voice interview 3, and phone calls 0.6 credits a minute. Extra credits are €1 each.\n\n![Koji report summary with a side panel of supporting participant quotes behind one finding](/email/shots/shot-evidence-panel.jpg \"Click any finding in a Koji report to see the quotes behind it\")\n\n## Where they differ\n\n### A managed consumer study or a tool you drive\n\nBolt Insight is closer to an AI-powered research partner. You bring an objective, it brings the guide, the panel and the report. That is a good fit for a brand team that runs a set number of big studies a year across many countries.\n\nKoji hands you the controls. A product manager can go from a question to a live interview in an afternoon, read the first answers the same day and change the brief for the next wave. Teams that want to talk to customers every week tend to need that.\n\n### Who you interview\n\nBoltChatAI recruits consumers through panel partners. That is its strength for category and concept work. Koji does panel recruitment too, but most Koji studies start with people you already know: your trial users, churned accounts or B2B buyers, invited by link or a CSV of personal links. Those people are rarely on a consumer panel.\n\n![A researcher on a video call with a customer, the kind of person a product team invites from its own user list](https://images.unsplash.com/photo-1616587226157-48e49175ee20?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHw1fHxjdXN0b21lciUyMGludGVydmlldyUyMHZpZGVvJTIwY2FsbCUyMGxhcHRvcHxlbnwxfDB8fHwxNzkyNTYxODEzfDA&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [LinkedIn Sales Solutions](https://unsplash.com/@linkedinsalesnavigator?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\n### Media\n\nBolt accepts audio, video and image answers, which helps for packaging or in-home tasks. Koji is voice and text, with phone calls. There is no video or photo upload, so if you need participants to show you a product on a shelf, Bolt fits better.\n\n### How you buy\n\nBolt needs a sales conversation. Koji publishes its prices and charges only for interviews that pass the quality gate. You can price a 40-interview voice study before you sign up: 120 credits.\n\n## Before you sign with either\n\nPut the same questions to both vendors. Take a test interview yourself, ask how low-effort answers are handled and whether you pay for them, and trace one finding back to the raw quotes.\n\n![Checklist of eight questions to ask before choosing an AI interview platform, from free previews to data storage](/visuals/ai-interview-platform-vendor-checklist/ai-interview-platform-vendor-checklist.png \"Ask both vendors these eight questions before you sign\")\n\n## Who should choose which\n\n**Choose Bolt Insight if:**\n\n- You are a consumer or CPG brand running concept, packaging or usage studies in many countries.\n- You want the vendor to draft the guide, recruit the sample and deliver a report.\n- You need participants to send photos or video as part of the task.\n\n**Choose Koji if:**\n\n- You want to run interviews yourself, quickly and often, without a sales cycle.\n- Your participants are your own customers or users, including B2B buyers.\n- You want scales and choice questions inside the same conversation, and every finding traceable to quotes.\n- You want to see the price first and pay only for interviews that pass a quality check.\n\n## The bottom line\n\nBolt Insight is a credible, funded choice for global consumer brands that want a managed AI-moderated study. Koji is the faster and more open choice for teams that want to run their own AI interviews: voice, text or phone, 30+ languages, a quality gate on every interview and public prices from free.\n\n## Try Koji free\n\nThe quickest way to compare is to run a study. [Create a free Koji account](https://www.koji.so), type your research question, preview the interview yourself and send it to ten customers. Your 10 signup credits cover ten text interviews, enough to see the report before you spend anything. For a wider view, see our guide to [AI-moderated interview platforms](/blog/ai-moderated-interview-platforms-2026), or compare Koji with [GetWhy](/blog/koji-vs-getwhy-2026), another platform built for consumer brands.\n","category":"Comparisons","lastModified":"2026-10-30T06:27:12.820334+00:00","metaTitle":"Koji vs Bolt Insight 2026: BoltChatAI Compared","metaDescription":"Bolt Insight's BoltChatAI vs Koji: AI-moderated interviews compared on setup, participants, languages, turnaround and pricing, with who should pick which.","keywords":["koji vs bolt insight","boltchatai","bolt insight alternative","boltchatai pricing","ai moderated interviews","ai consumer research","cpg qualitative research"],"aiSummary":"A comparison of Bolt Insight's BoltChatAI and Koji, two platforms that use an AI moderator for one-to-one qualitative interviews. Bolt Insight is a managed, sales-led platform for global consumer brands that drafts the guide, recruits through panel partners and reports within about a day by its own account. Koji is self-serve with public pricing, running voice, text and phone interviews in 30+ languages with a quality gate and findings linked to quotes.","aiKeywords":["bolt insight","boltchatai","ai moderated interviews","ai qualitative research","consumer insights platform","cpg research","panel recruitment","research pricing","koji"],"aiContentType":"comparison","faqItems":[{"answer":"BoltChatAI is Bolt Insight's AI-moderated qualitative research platform. It runs one-to-one interviews that probe vague answers, drafts the discussion guide from your objective and audience, recruits through panel partners and, by Bolt's account, delivers reports in as little as 24 hours.","question":"What is BoltChatAI?"},{"answer":"Bolt Insight does not publish prices, so you need a quote. GetApp lists a starting price of US$9,950 on a flat-rate model, but that is based on only three reviews. Koji publishes its prices: a free account with 10 credits, Insights at €29 a month and Interviews at €79 a month.","question":"How much does Bolt Insight cost?"},{"answer":"Yes, if you want to run interviews yourself. Koji is self-serve, runs voice, text and phone interviews in 30+ languages with your own customers or Koji's research panel, and links every finding to quotes. Bolt Insight fits better if you want a managed consumer study or need photo and video answers.","question":"Is Koji a good Bolt Insight alternative?"},{"answer":"Koji is usually the better fit, because B2B buyers are rarely on consumer panels. You can invite your own customers by link or give each person a private invite link from a CSV, and screen people out before the interview starts.","question":"Which is better for B2B research?"},{"answer":"Yes. A free account includes 10 credits, enough for ten text interviews, and you can preview your own interview by text or voice as often as you like without using credits.","question":"Can I test Koji before paying?"}],"relatedTopics":["bolt insight","ai moderated interviews","consumer insights","cpg research","research pricing"]},{"type":"documentation","id":"00d87683-ffcb-42cc-a3ab-6a889826edce","slug":"ideal-customer-profile-icp","title":"Ideal Customer Profile (ICP): Definition, Template, and How to Build One with AI Interviews (2026)","url":"https://www.koji.so/docs/ideal-customer-profile-icp","summary":"An Ideal Customer Profile (ICP) is a firmographic and behavioral description of the perfect customer for your product. It differs from a buyer persona (which describes a human) and a target market (which is broader). A useful ICP includes 12 fields covering firmographics, buying committee, trigger event, current alternative, switching cost, and disqualifiers. The most common failure mode is writing the ICP from internal opinion instead of customer interviews. Koji compresses the ICP validation interview cycle from six weeks to about five days using AI-moderated conversations with built-in Customer Discovery and JTBD methodology frameworks.","content":"# Ideal Customer Profile (ICP): The Complete 2026 Guide With Templates and Examples\n\n**TL;DR:** An Ideal Customer Profile (ICP) is a detailed description of the company or customer that gets the most value from your product, generates the highest revenue at the lowest cost to serve, and is most likely to renew, expand, and refer. It is the single most important strategic document in B2B go-to-market. This guide covers ICP definition, the difference between ICP and personas, B2B and B2C templates, the data you need, real examples, and how Koji's AI interviewer validates your ICP through customer conversations in days instead of months.\n\n## What is an Ideal Customer Profile (ICP)?\n\nAn **Ideal Customer Profile (ICP)** is a firmographic and behavioral description of the perfect customer for your product — the segment that produces the highest lifetime value, lowest churn, fastest sales cycle, and strongest word-of-mouth. Unlike a buyer persona (which describes an individual person), an ICP describes a *company* or a *household-level segment* in B2C.\n\nA good ICP answers questions like:\n\n- What industry and sub-vertical?\n- Company size (employees, ARR, funding stage)?\n- Geography and language?\n- Technology stack and maturity?\n- Team structure (who buys, who uses, who pays)?\n- What \"trigger\" makes them ready to buy?\n- What's their current pain, current alternative, and current cost?\n\nWithout an ICP, every channel looks reasonable, every prospect looks worth pursuing, and your sales and product teams build for the loudest customer instead of the most valuable one. According to research from First Round Capital, companies with a documented ICP grow 68% faster than those without one.\n\n## ICP vs Buyer Persona vs Target Market\n\nThese three terms get conflated constantly. Here's the clean distinction:\n\n| | Target Market | ICP | Buyer Persona |\n|---|---|---|---|\n| Scope | A broad segment | A specific account type | A specific human |\n| Example | \"Mid-market SaaS\" | \"Series B–D SaaS, 50–500 employees, EU-based, product-led, Stripe in stack\" | \"Maya, VP Product, 38, swamped, optimizes for time-to-insight\" |\n| Used by | Marketing, board, strategy | Sales, RevOps, ABM, GTM | Product, marketing, design |\n| Time horizon | Multi-year | 12–18 months | 6–12 months |\n\nYour ICP narrows your target market. Your buyer personas live inside your ICP. You need all three, but only the ICP determines who your sales team chases this quarter.\n\n## ICP template: the 12 fields that matter\n\nUse this template as a starting point. Most teams over-engineer ICPs in the first version, then realize they don't have data for half the fields. Start with what you can verify from real customer data.\n\n1. **Industry / sub-vertical** — Not \"SaaS\" but \"vertical SaaS for healthcare staffing.\"\n2. **Company size** — Employees, ARR, customer count, or another scale signal.\n3. **Funding stage / growth signal** — Stage of company, growth rate, recent funding.\n4. **Geography & language** — Country, region, primary working language.\n5. **Tech stack** — What they already use that signals fit (e.g., \"uses Segment + Snowflake\").\n6. **Buying committee** — Economic buyer, champion, end user, blockers.\n7. **Trigger event** — What changes that makes them ready to buy *now* (new exec, new funding, new compliance rule).\n8. **Primary pain** — The job they're trying to do that your product addresses.\n9. **Current alternative** — In-house, a competitor, a manual process, or \"doing nothing.\"\n10. **Switching cost / inertia** — What makes them sticky with the current solution.\n11. **Success criteria** — How they'll measure ROI on your product.\n12. **Disqualifiers** — Red flags that make a deal a bad fit (small TAM, wrong stack, wrong stage).\n\n## How to build your ICP in 5 steps\n\n### Step 1: Mine your best existing customers\nStart with the top 10–20% of your existing customers by lifetime value, NPS, retention, and expansion. Pull firmographic data from your CRM and product analytics from your data warehouse. Look for what these accounts share that the rest of your customer base doesn't.\n\n### Step 2: Interview them\nFirmographics tell you *who* they are. Interviews tell you *why* they're your best customers — what they were doing before, what triggered the search, what they almost did instead, and what makes them sticky. This is the step most teams skip, and it's where Koji shines.\n\nWith Koji, you can:\n\n- Drop interview links into Slack, email, or a Loom intro from your CSM\n- Use the prebuilt **Customer Discovery** or **Jobs to Be Done** methodology framework\n- Let the AI interviewer run a 15-minute voice or text conversation that probes for triggers, alternatives, and success metrics\n- Get an auto-generated report with themes, quotes, and quality scores by the next morning\n\nA traditional ICP-validation interview round takes 6 weeks. Koji compresses it to 5 days.\n\n### Step 3: Find anti-patterns\nLook at your worst-fit customers too: low usage, high support load, churn risk, slow payment. Their shared traits are your **disqualifiers** — the fields you put in your \"do not pursue\" filter.\n\n### Step 4: Triangulate with quantitative signals\nAdd structured questions to your AI interviews — Koji supports six types (open_ended, scale, single_choice, multiple_choice, ranking, yes_no) — to quantify how many \"best fit\" customers share specific firmographic or behavioral traits. Now your ICP isn't just a story; it's a story with percentages.\n\n### Step 5: Document and socialize\nWrite the ICP as a one-page document. Include the 12 fields above, three illustrative customer examples, and a \"do not pursue\" list. Get sales, marketing, product, and CS to sign off. Revisit every 6 months.\n\n## B2B ICP example: a sample one-pager\n\n> **Company:** Series B–D vertical SaaS for healthcare staffing in North America\n> **Size:** 80–400 employees, $5M–$40M ARR\n> **Stack:** Salesforce or HubSpot, Snowflake or BigQuery, Slack-first culture\n> **Buying committee:** VP Operations (economic buyer), Head of People (champion), Field recruiters (users)\n> **Trigger event:** New compliance requirement (CMS staffing rule) or a single bad audit\n> **Primary pain:** Recruiter productivity is flat; manual data work eats 15+ hours per recruiter per week\n> **Current alternative:** A patchwork of spreadsheets + an outdated ATS\n> **Success criteria:** 30% reduction in time-to-fill within 6 months\n> **Disqualifiers:** Pre-Series A, agency model (not employer-of-record), Asia-Pacific (no compliance support)\n\nThat one-pager is 100x more useful to a sales team than \"we sell to mid-market SaaS.\"\n\n## B2C ICP example\n\nFor a B2C product, you'll replace firmographics with household-level data — household income, life stage, geography, channel, and behavioral triggers (just had a baby, just bought a house, just started a new job). The structure is the same.\n\n## How Koji helps you build and validate your ICP\n\nMost ICP projects die in the \"validate it with customers\" phase because the research takes too long. Koji removes that bottleneck:\n\n- **Built-in methodology frameworks** for Customer Discovery, Jobs to Be Done, and Open Exploration — the three most useful lenses for ICP work.\n- **AI-moderated interviews** that run 24/7 in voice or text, with intelligent follow-up probing on every answer.\n- **Personalized interview links** so you can target specific firmographic slices (e.g., \"Series B SaaS that signed up last month\") and prefill context for each respondent.\n- **Six structured question types** that quantify the qualitative — turn open-ended themes into ranked importance scores.\n- **Auto-generated reports** with quotes, themes, and quality scores ready to drop into your ICP one-pager.\n- **CRM sync** (HubSpot, Salesforce via Zapier) so ICP-fit signals flow straight back into your sales pipeline.\n\nA platform like Koji turns ICP validation from a once-a-year strategic project into a continuous always-on process.\n\n## Common ICP mistakes\n\n- **Confusing ICP with TAM.** TAM is \"how big is the market we *could* sell to.\" ICP is \"who do we *focus* on first.\"\n- **Writing the ICP from internal opinion.** If your ICP isn't grounded in real customer data and interviews, it's a wish list.\n- **Skipping disqualifiers.** Without an explicit \"do not pursue\" list, sales reps will chase anything that breathes.\n- **Confusing ICP with persona.** Persona = human. ICP = account.\n- **Setting it and forgetting it.** Your ICP should evolve. Re-validate it every 6 months — and Koji makes that practical.\n\n## When to refine your ICP\n\nRefresh your ICP whenever you:\n\n- Hit a new product release that shifts who benefits most\n- See win rate or expansion rate change significantly\n- Enter a new geography or vertical\n- Hire a new VP of Sales or Marketing\n- Get an investment or board change that resets growth targets\n\n## Related Resources\n\n- [Find Your ICP in 3 Rounds of Interviews](/docs/find-ideal-customer-profile-three-rounds-of-interviews): a founder playbook with questions and scoring for each round\n- [Customer Discovery Workshop: The Step-by-Step Playbook With Templates](/docs/customer-discovery-workshop-guide)\n- [Jobs to Be Done Framework: The Complete Guide](/docs/jobs-to-be-done-framework)\n- [User Persona Template: 7 Free Templates and Examples for Product Teams](/docs/user-persona-template)\n- [Customer Segmentation Research: How to Build Segments That Actually Drive Decisions](/docs/customer-segmentation-research-interviews)\n- [Win-Loss Analysis: How to Learn Why Deals Are Won and Lost](/docs/win-loss-analysis)\n- [Structured Questions Guide](/docs/structured-questions-guide)","category":"Research Methods","lastModified":"2026-10-30T05:58:27.390923+00:00","metaTitle":"Ideal Customer Profile (ICP): Template, Examples, and AI Validation | Koji","metaDescription":"Build a real Ideal Customer Profile with our 12-field template, B2B and B2C examples, and the 5-step process to validate your ICP with Koji's AI interviewer.","keywords":["ideal customer profile","icp","icp template","icp examples","b2b icp","customer profile","target customer profile","icp vs persona","icp definition","how to build an icp","icp framework","customer profiling"],"aiSummary":"An Ideal Customer Profile (ICP) is a firmographic and behavioral description of the perfect customer for your product. It differs from a buyer persona (which describes a human) and a target market (which is broader). A useful ICP includes 12 fields covering firmographics, buying committee, trigger event, current alternative, switching cost, and disqualifiers. The most common failure mode is writing the ICP from internal opinion instead of customer interviews. Koji compresses the ICP validation interview cycle from six weeks to about five days using AI-moderated conversations with built-in Customer Discovery and JTBD methodology frameworks.","aiPrerequisites":["Basic understanding of your customer base","Access to CRM or customer data","Time to interview 10–15 best-fit customers"],"aiLearningOutcomes":["Define the difference between ICP, target market, and buyer persona","Apply a 12-field ICP template to your business","Validate your ICP through customer interviews using AI moderation","Build a disqualifier list to keep sales focused","Refresh your ICP on a continuous cadence using Koji"],"aiDifficulty":"intermediate","aiEstimatedTime":"13 min read"},{"type":"documentation","id":"dfd2070d-0dc7-456e-b43a-7954a2c2979a","slug":"sean-ellis-test-product-market-fit","title":"Sean Ellis Test: The 40% Rule for Product-Market Fit (Complete 2026 Guide)","url":"https://www.koji.so/docs/sean-ellis-test-product-market-fit","summary":"The Sean Ellis Test is a 4-question survey that measures product-market fit by asking active users how they would feel without the product. If 40% or more say 'very disappointed', you have PMF and can scale growth. Below 40%, keep iterating. AI interview platforms like Koji automate the survey with conversational follow-up probing, voice or text modes, and auto-analysis — turning a 3-week manual process into 2 to 3 days.","content":"# Sean Ellis Test: The 40% Rule for Product-Market Fit (Complete 2026 Guide)\n\n**TL;DR:** The Sean Ellis Test (also called the PMF Survey or \"40% rule\") asks existing users a single question: *\"How would you feel if you could no longer use this product?\"* If 40% or more answer **\"very disappointed,\"** you have product-market fit. Below 40% means you need to keep iterating before scaling growth. The test is cheap, fast, and predictive — and AI interview platforms like Koji let you run it conversationally so you also capture the *why* behind each rating.\n\n## What is the Sean Ellis Test?\n\nThe Sean Ellis Test is a survey-based methodology for measuring product-market fit (PMF), created by growth expert Sean Ellis — the first marketer at Dropbox, LogMeIn, Eventbrite, and Lookout. Sean discovered a pattern: companies that struggled to scale always had fewer than 40% of users saying they'd be \"very disappointed\" without their product. Companies that scaled effortlessly always exceeded that threshold.\n\nInstead of measuring PMF through proxy metrics like retention curves or NPS, the Sean Ellis Test asks one direct question that captures *emotional dependence*. It is the closest thing to a single-number \"do we have PMF?\" gauge that exists in the product world.\n\nThe test has four core questions:\n\n1. **How would you feel if you could no longer use [product]?** — Options: *Very disappointed / Somewhat disappointed / Not disappointed / N/A I no longer use it*\n2. **What type of people do you think would most benefit from [product]?**\n3. **What is the main benefit you receive from [product]?**\n4. **How can we improve [product] for you?**\n\nThe first question is your PMF score. The next three tell you *who* loves the product, *why* they love it, and *what's blocking* even more love.\n\n## Why the 40% Threshold Matters\n\nSean Ellis benchmarked the \"very disappointed\" rate against ~100 startups. The pattern was unambiguous:\n\n- **Below 25%:** The product has not found a market. Major repositioning, audience pivot, or feature overhaul is needed.\n- **25% to 40%:** You are close but not there. Focus on the things your high-intent users tell you — usually a tightening of audience or a sharper value prop.\n- **40% or higher:** You have product-market fit. It is now safe to invest aggressively in growth.\n\nThis threshold is not arbitrary. Below 40%, paid acquisition typically leaks faster than referral can fill it. Above 40%, your product begins to compound — happy users tell other users, and growth becomes cheaper over time.\n\n## When to Run the Sean Ellis Test\n\nRun the test when:\n\n- You have **at least 30 to 40 active users** who have used the product more than twice in the last two weeks (this is the minimum sample for the question to be meaningful)\n- You're **deciding whether to invest in growth** (paid ads, SDR team, content engine) or keep iterating on product\n- You want to **track PMF over time** as a north-star metric on your product dashboard\n- You're **preparing for fundraising** and need a credible PMF signal investors trust\n\nDo *not* run it when users haven't experienced the product enough to know whether they'd miss it. Set a minimum usage threshold — for example, \"used at least 3 times in the past 14 days\" — before sending the survey.\n\n## How to Run the Sean Ellis Test in 6 Steps\n\n![Process showing six steps to run the Sean Ellis test, from defining active users and recruiting 40 respondents to tracking the score over time.](/visuals/sean-ellis-test-six-steps-process/sean-ellis-test-six-steps-process.png \"Survey only users who have felt the product, then read the open answers, not just the score. · Source: Sean Ellis, product-market fit survey\")\n\n### 1. Define Your \"Active User\" Segment\nPick the cohort that has experienced enough of the product to answer honestly. For most B2B SaaS, this is users who completed onboarding *and* returned at least twice. For consumer apps, it's users who hit a key activation event (sent first message, completed first transaction, etc.).\n\n### 2. Recruit at Least 40 Respondents\nStatistical reliability for the 40% threshold requires roughly 40 to 100 responses. Send the survey to a broader cohort — typical response rates are 10% to 20% for in-app prompts and 5% to 10% for email.\n\n### 3. Use a Conversational Format, Not a Static Survey\nTraditional survey tools collect single-word answers. The real insight is in the *open-ended* responses — what people would miss, why, who they'd recommend it to. AI-moderated interview platforms like Koji ask the four core questions and then automatically follow up: *\"You said you'd be 'very disappointed.' Can you tell me about the last time the product saved you time or solved a problem?\"* This unlocks 5x richer qualitative data than a static survey form.\n\n### 4. Segment the \"Very Disappointed\" Group\nThe most valuable data comes from segmenting *who* says \"very disappointed.\" Look at:\n\n- **What roles or job titles** are over-represented in this group?\n- **What use cases or features** do they describe in the open-ended questions?\n- **How did they find the product?**\n\nThese segments are your ICP. Marketing, sales, and product should be ruthlessly aligned around them.\n\n### 5. Mine the Open-Ended Responses\nThe \"what is the main benefit\" and \"how can we improve\" answers are gold — they're the *exact words* your future website headlines, ad copy, and roadmap should reflect. Koji's auto-analysis surfaces these themes and quotes automatically, so you don't have to read 100 transcripts manually.\n\n### 6. Track the Score Over Time\nPMF is not a static state. As you change the product, audience, or positioning, the score drifts. Run the test quarterly and compare cohorts. A rising score means your changes are working; a falling score means something just broke.\n\n## Sean Ellis Test in Practice: How Koji Automates It\n\nSean Ellis originally distributed the test via static survey tools (SurveyMonkey, Typeform, Google Forms). The mechanics were clunky: send a link, wait, manually code open-ended responses, and hope respondents answered the follow-ups.\n\nWith platforms like Koji, the entire workflow is faster and richer:\n\n- **Structured PMF question types:** Use Koji's six structured question types (open_ended, scale, single_choice, multiple_choice, ranking, yes_no) to capture the \"very disappointed\" rating as a `single_choice` question and the rest as `open_ended` with AI follow-up probing.\n- **AI follow-up probing:** When a user says \"very disappointed,\" Koji's AI moderator asks *why* in their own words — surfacing the language of love that becomes your marketing copy.\n- **Voice or text:** Run the same PMF survey via voice for higher response quality and emotional nuance, or text for speed and asynchronicity.\n- **Real-time analysis:** Auto-extracted themes, quality scores, and quote tags appear as soon as interviews complete. No manual coding required.\n- **Cohort segmentation:** Pipe responses to your CRM with the Koji API, a CSV or JSON export, or an automation platform like Zapier calling the API on a schedule. Outbound webhooks are not available yet. Either route lets you segment \"very disappointed\" users for ICP refinement and lookalike targeting.\n\nA typical workflow that used to take 3 weeks (build survey, recruit, manually analyze) now takes 2 to 3 days end-to-end.\n\n## How to Interpret Your Sean Ellis Score\n\n![Table of Sean Ellis test score bands from under 25% to 50% and above, with what each means for product-market fit and the next step.](/visuals/sean-ellis-test-score-interpretation-table/sean-ellis-test-score-interpretation-table.png \"The share answering very disappointed decides whether you iterate or scale. · Source: Sean Ellis, product-market fit survey\")\n\n| Score | Interpretation | Recommended Next Step |\n|---|---|---|\n| **< 25%** | No PMF yet | Pivot positioning or audience. Re-run customer discovery interviews. |\n| **25–39%** | Approaching PMF | Tighten ICP. Double down on the use case that resonates with \"very disappointed\" users. |\n| **40–49%** | PMF achieved | Begin scaling growth investment. Continue tracking quarterly. |\n| **50%+** | Strong PMF | Focus on retention, expansion, and category leadership. |\n\nFamous PMF scores at scale:\n\n- **Slack** reportedly hit 51% in early days\n- **Superhuman** famously rebuilt the product over 18 months to climb from 22% to 58%\n- **Hubspot** ran the test as part of their early growth motion\n\n## Common Mistakes That Skew the Score\n\n1. **Surveying users who haven't activated.** Their answer to \"would you be disappointed\" is \"no\" because they never got value. Filter ruthlessly.\n2. **Treating \"somewhat disappointed\" as a win.** It isn't. Only \"very disappointed\" counts in the 40% calculation.\n3. **Sampling power users only.** Running the survey to your top 10 customers will inflate the score. Use a representative cohort.\n4. **Ignoring open-ended responses.** The score is the headline; the qualitative data is the actual playbook.\n5. **Treating the score as static.** PMF is a moving target. Pricing changes, competitor launches, and audience shifts all move the number.\n\n## Sean Ellis Test vs. NPS vs. Other PMF Metrics\n\n| Metric | What It Measures | When to Use |\n|---|---|---|\n| **Sean Ellis Test** | Emotional dependence on product | PMF check, especially pre-scaling |\n| **NPS** | Willingness to recommend | Ongoing loyalty tracking |\n| **Retention curve** | Behavioral stickiness | Long-term PMF validation |\n| **CES (Customer Effort Score)** | Friction in specific tasks | UX-level optimization |\n| **DAU/MAU stickiness ratio** | Usage frequency | Engagement health |\n\nThe Sean Ellis Test is the best *leading indicator* — it tells you whether your retention curve and growth metrics *will* be healthy. NPS and retention are *trailing indicators* — they confirm what your PMF score already showed.\n\n## Frequently Asked Questions\n\n### How many users do I need for a valid Sean Ellis Test?\nA minimum of 40 responses is needed for the 40% threshold to be statistically meaningful. Below 40 responses, treat the result as directional only. Aim for 100+ responses for confident decisions.\n\n### Should I include inactive users in the survey?\nNo. Only survey users who have used the product enough to form an opinion. A typical filter is \"used the product 2 or more times in the past 14 days.\" Sending the survey to inactive users will deflate your score with non-engaged opinions.\n\n### Is the 40% threshold universal across industries?\nThe 40% rule is a general benchmark, not an industry-specific number. For high-frequency consumer products (messaging, social), strong PMF often shows up at 50%+. For B2B enterprise tools, 40% from the right ICP is a strong signal. The trend over time matters more than hitting a specific number.\n\n### Can I use the Sean Ellis Test on a free product?\nYes — in fact, it's especially useful for free-tier products where willingness-to-pay isn't a useful signal yet. Filter for users who have hit your activation milestone, then run the test on that cohort.\n\n### How does the Sean Ellis Test compare to a PMF survey?\nThe Sean Ellis Test *is* a PMF survey — specifically, it's the most widely-adopted version. Other PMF surveys exist (Superhuman PMF Engine, Pirate Metrics retention proxies), but the Sean Ellis question remains the gold standard for its simplicity.\n\n### Does AI moderation change the validity of the Sean Ellis Test?\nNo — if anything, it improves it. Static surveys are subject to respondent fatigue and shallow answers. AI moderators (like Koji's) ask follow-up questions in plain language, capture richer \"why\" responses, and reduce survey abandonment by 30 to 50%.\n\n## Related Resources\n\n- [Find Your ICP in 3 Rounds of Interviews](/docs/find-ideal-customer-profile-three-rounds-of-interviews): a founder playbook with questions and scoring for each round\n- [Product Market Fit Interviews](/docs/product-market-fit-interviews) — Deeper qualitative methodology that pairs with the Sean Ellis Test\n- [Structured Questions Guide](/docs/structured-questions-guide) — Use the six question types to design your PMF survey\n- [Customer Validation Guide](/docs/customer-validation-guide) — End-to-end framework for validating product-market fit\n- [Mom Test User Interviews](/docs/mom-test-user-interviews) — How to ask better follow-up questions during PMF research\n- [NPS Survey Guide](/docs/nps-survey-guide) — Trailing-indicator metric to track once PMF is achieved\n- [Customer Discovery Interviews](/docs/customer-discovery-interviews) — Pre-PMF research methodology for finding the right audience\n\nRun a Sean Ellis Test in under an hour with [Koji's AI interview platform](/) — structured PMF questions, AI follow-up probing, and auto-analysis included.\n","category":"Research Methods","lastModified":"2026-10-30T05:58:27.390923+00:00","metaTitle":"Sean Ellis Test: The 40% Rule for Product-Market Fit (2026 Guide)","metaDescription":"Run the Sean Ellis Test to measure product-market fit. Learn the 40% threshold, the 4 core questions, how to interpret scores, and run conversational PMF surveys with AI moderation.","keywords":["sean ellis test","product market fit test","40% rule product market fit","pmf survey","pmf score","how to measure product market fit","sean ellis pmf survey","product market fit questions","pmf interviews","product market fit measurement"],"aiSummary":"The Sean Ellis Test is a 4-question survey that measures product-market fit by asking active users how they would feel without the product. If 40% or more say 'very disappointed', you have PMF and can scale growth. Below 40%, keep iterating. AI interview platforms like Koji automate the survey with conversational follow-up probing, voice or text modes, and auto-analysis — turning a 3-week manual process into 2 to 3 days.","aiPrerequisites":["Basic understanding of product-market fit concept","At least 30-40 active product users to survey","Defined activation event or usage threshold for filtering respondents"],"aiLearningOutcomes":["Understand the 40% threshold and why it matters","Run a Sean Ellis Test in 6 steps from sample selection to score interpretation","Interpret Sean Ellis scores against industry benchmarks","Avoid the 5 most common mistakes that skew PMF scores","Compare the Sean Ellis Test with NPS, retention, and other PMF metrics","Automate the test with AI-moderated interviews for richer qualitative data"],"aiDifficulty":"intermediate","aiEstimatedTime":"14 min read"},{"type":"documentation","id":"8b1a3b9d-18cf-49f1-a1a5-da05d5a5bd54","slug":"customer-discovery-interviews","title":"Customer Discovery Interviews: The Complete Guide","url":"https://www.koji.so/docs/customer-discovery-interviews","summary":"Customer discovery interviews are structured conversations with potential customers designed to test whether a real problem exists before building a solution. This guide covers the Steve Blank methodology, optimal sample sizes of 15–20 interviews, proven question frameworks based on the Mom Test, and the six most common mistakes founders make — including pitching instead of listening and asking hypothetical future questions.","content":"\nCustomer discovery interviews are one-on-one conversations with potential customers designed to test your assumptions about a problem before you invest in building a solution. According to CB Insights, **42% of startups fail because they built something nobody wanted** — customer discovery interviews are how you avoid that outcome.\n\n## What Are Customer Discovery Interviews?\n\nCustomer discovery interviews test a single hypothesis: does a real problem exist, and is it painful enough that people actively seek a solution? The goal is not to pitch your idea. It is to understand the customer's world.\n\nSteve Blank, the Stanford professor who created the Customer Development methodology, framed it directly: \"There are no facts inside the building, so get the hell outside.\" Until you have heard from real people about real problems — in their own words — everything you believe about your market is an assumption.\n\nCustomer discovery is distinct from customer validation (testing whether people will buy your specific solution) and usability testing (evaluating whether an existing product is easy to use). Discovery comes first. It answers the question: *is there a real problem worth solving?*\n\n## Why Customer Discovery Interviews Matter\n\nThe evidence is unambiguous:\n\n- **42% of startups fail due to no market need**, making it the single most common startup failure cause, according to CB Insights analysis of over 100 startup post-mortems.\n- **20–30 interviews surface 90–95% of customer needs.** Griffin and Hauser's landmark 1993 study in *Marketing Science* demonstrated that a structured interview program reliably uncovers nearly all discoverable needs in a well-defined customer segment.\n- **Thematic saturation typically occurs between interviews 9 and 17**, according to PMC research on qualitative saturation methods — meaning most early-stage teams need far fewer interviews than they assume.\n- **The 40% \"very disappointed\" threshold** established by Sean Ellis provides a measurable benchmark for product-market fit: when 40% or more of your users say they would be very disappointed without your product, you have found it.\n\n## When to Use Customer Discovery Interviews\n\n| Situation | Right Approach? |\n|-----------|-----------------|\n| You have an idea and want to test if the problem is real | ✅ Yes — this is the core use case |\n| You are entering a new market segment you do not understand | ✅ Yes — discovery interviews first |\n| You want to understand why customers are churning | ✅ Yes — interviews reveal the why |\n| You need statistically representative data across 1,000+ customers | ❌ Better to use a survey after discovery |\n| You already have a product and want usability feedback | ❌ Better to use usability testing |\n| You want to track satisfaction scores over time | ❌ Better to use CSAT or NPS surveys |\n\n## How to Conduct Customer Discovery Interviews: Step by Step\n\n### Step 1: Write Down Your Assumptions First\n\nBefore talking to anyone, document your hypotheses in writing. Who is your customer? What problem do they have? How severe is it? What do they currently do to solve it? Writing your assumptions forces you to confront what you do not know — and gives you a benchmark to compare against what you actually hear.\n\n### Step 2: Define Your Target Segment Precisely\n\n\"Small business owners\" is too broad. \"B2B SaaS founders with fewer than 10 employees who manage their own sales pipeline\" is testable. The more precisely you define your segment, the more coherent and comparable your data will be. Interviewing a heterogeneous group produces contradictory patterns; a focused segment produces clear signal.\n\n### Step 3: Recruit 15–20 Participants\n\nFor a focused customer segment, 15–20 interviews conducted with genuine open-ended listening will reach saturation for most early-stage discovery. Recruit through:\n- Your professional network (with caution around courtesy bias)\n- LinkedIn outreach targeting your specific segment\n- Online communities and Slack groups relevant to your problem space\n- Customer research platforms that maintain panels of pre-screened participants\n\n### Step 4: Prepare Your Discussion Guide\n\nA discussion guide is not a script — it is a set of topic areas to explore. Keep it to 5–7 questions, all open-ended. Every question should invite a story, not a yes or no.\n\n**Strong opening questions:**\n- \"Walk me through how you currently handle [problem area]. What does a typical week look like?\"\n- \"When was the last time [the problem] came up? What happened?\"\n\n**Problem depth questions:**\n- \"What is the most frustrating part of dealing with [X]?\"\n- \"How much time or money does this problem cost you per week?\"\n- \"What have you already tried to solve this? What worked or did not work?\"\n\n**The key principle from Rob Fitzpatrick's *The Mom Test*:** ask about specific past behavior, not hypothetical future behavior. \"Would you use this?\" is unreliable. \"Can you show me what you use today?\" reveals the truth.\n\n### Step 5: Conduct the Interview — Listen More Than You Talk\n\nCindy Alvarez, author of *Lean Customer Development*, captured the discipline in one sentence: \"The more you talk, the less you will learn from your customer.\" Aim to speak less than 20% of the time.\n\nLet silences breathe — participants often fill them with their most honest observations. Never mention your solution in the first half of the interview. You are there to understand their world, not pitch yours.\n\n**Practical tips:**\n- Record the session (with permission) so you can listen without taking notes\n- Use a neutral second observer to watch while you moderate\n- Probe with \"Tell me more about that\" — it works in almost every situation\n- Note body language and energy shifts, not just words\n\n### Step 6: Analyze Patterns Across All Interviews\n\nThe insight is never in a single interview — it lives in the patterns across 10 or 20 conversations. After each session, write a brief note: what new things did you hear? What surprised you?\n\nAfter all interviews are complete, do a thematic analysis:\n- Group recurring problems, phrases, and workarounds\n- Identify which problems appeared in more than half your interviews (these are likely real)\n- Look for specific situations, motivations, and outcomes that cluster together\n\nEric Ries framed the underlying principle: \"We must learn what customers really want, not what they say they want or what we think they should want.\"\n\n### Step 7: Update and Re-Test Your Hypotheses\n\nCustomer discovery is iterative. Each round of interviews should update your hypotheses and inform the next round. If early conversations reveal an unexpected problem, go deeper on it. Continue until you have conviction — or until you have invalidated your original idea and need to pivot.\n\n## Common Mistakes to Avoid\n\n**1. Pitching instead of listening.** The moment you introduce your solution, the conversation becomes a pitch. Interviewees want to be helpful — they will respond to your idea rather than describe their actual problem. Keep your solution out of the conversation until you have heard everything they have to say.\n\n**2. Asking hypothetical future questions.** Rob Fitzpatrick identified this as the world's most misleading feedback: \"I would definitely buy that.\" People are systematically more positive about imagined future behavior than actual behavior. Ask about what they have done in the past, not what they would do hypothetically.\n\n**3. Collecting compliments, not data.** Generic enthusiasm — \"This sounds really interesting!\" — is noise. The signal is: demonstrated past behavior, existing workarounds they have already paid for, and concrete expressions of pain in time, money, or emotional cost.\n\n**4. Interviewing the wrong people.** Friends, co-workers, and enthusiastic supporters who have a social obligation to be kind are the wrong sample. Recruit people who fit your target segment precisely and have no reason to spare your feelings.\n\n**5. Stopping too soon.** Five conversations feel like significant work, but they rarely reach saturation. Commit to at least 10–15 before drawing conclusions.\n\n**6. Not recording or systematically analyzing.** Memory is selective and self-serving. Without recordings or structured notes compared across all interviews, the data degrades into impressions shaped by whoever spoke most forcefully.\n\n## Real-World Example\n\nImagine you are building a tool to help independent consultants track project profitability. You believe the core problem is that consultants do not know their hourly rate.\n\nAfter 10 discovery interviews, a different picture emerges:\n- 8 of 10 consultants know their rate precisely\n- The actual pain is **scope creep** — projects consistently expand without additional compensation\n- Most are losing 20–30% of project revenue to undocumented scope changes\n- Their current workaround is a combination of spreadsheets and awkward email threads\n\nYou have just avoided building the wrong product. The discovery interviews did not just refine your idea — they revealed a completely different, far more acute problem to solve.\n\n## How Many Interviews Are Enough?\n\n| Source | Recommendation |\n|--------|----------------|\n| Griffin & Hauser, *Marketing Science* (1993) | 20–30 interviews to surface 90–95% of customer needs |\n| Guest et al., qualitative research methods | Saturation commonly reached by 12 interviews |\n| PMC thematic saturation research | Full saturation typically between interviews 9–17 |\n| Cindy Alvarez, *Lean Customer Development* | 10–15 interviews per week during active discovery |\n\n**The practical rule:** Start with 5–6, analyze as you go, and keep going until new conversations stop producing new themes. That is saturation — and your signal to move to the next phase.\n\n## Tools and Resources\n\nRunning 15–20 discovery interviews manually creates a significant analysis burden. Identifying patterns across transcripts, tagging themes, and synthesizing insights traditionally takes days of careful work.\n\nAI-native research platforms like Koji can streamline this significantly. Koji conducts AI-moderated voice or text interviews with participants at scale, automatically surfaces themes and patterns across all conversations, and generates synthesized reports — so teams can focus analytical energy on interpreting findings rather than manually coding transcripts. For teams that need to move fast, this compresses discovery timelines from weeks to days.\n\n## Key Takeaways\n\n- Customer discovery interviews test your core assumption: does this problem exist, and is it painful enough to solve?\n- Conduct 15–20 interviews with a precisely defined target segment to reach saturation\n- Ask about specific past behavior, not hypothetical future intent\n- Listen 80% of the time — let silences breathe\n- Patterns across interviews, not individual conversations, are where insight lives\n- Stop when new conversations stop surprising you\n\n## Frequently Asked Questions\n\n**Q: How long should a customer discovery interview be?**\nA: 30–45 minutes is the sweet spot. Long enough to go deep on the problem, short enough to maintain participant attention and respect their time.\n\n**Q: Should I record customer discovery interviews?**\nA: Yes, always with permission. Recording lets you be fully present during the conversation and gives you accurate data for analysis rather than reconstructed impressions.\n\n**Q: What if participants keep asking about my solution?**\nA: Redirect gently: \"That is helpful context — I would love to share what we are exploring later. First, I want to make sure I understand the problem as you experience it.\" Keep the focus on their world.\n\n**Q: How do I find participants if I do not have customers yet?**\nA: LinkedIn, online communities, Reddit, and Slack groups relevant to your problem space are all viable starting points. Offer a small incentive — a gift card or expert consulting feedback — to increase response rates.\n\n**Q: How is customer discovery different from user research?**\nA: Customer discovery is pre-product — you are testing whether a problem worth solving exists. User research typically evaluates an existing product or prototype. Discovery comes first; user research comes after you have built something to test.\n\n**Q: How many customer discovery interviews do I actually need?**\nA: Most teams reach saturation at 15–20 interviews with a focused segment. Start with 5–6, analyze as you go, and continue until new conversations stop producing new themes.\n\n\n---\n\n## Related Resources\n\n- [Find Your ICP in 3 Rounds of Interviews](/docs/find-ideal-customer-profile-three-rounds-of-interviews): a founder playbook with questions and scoring for each round\n- [Thematic Analysis Guide](/docs/thematic-analysis-guide) — Analyze interview data systematically\n- [How Many Interviews Are Enough](/docs/how-many-interviews-enough) — Sample size guidance\n- [Customer Discovery at Scale](/docs/customer-discovery-interviews-at-scale) — Run discovery at scale with AI\n- [Koji for Founders](/docs/koji-for-founders) — Startup research workflows\n- [Startup Idea Validation](/docs/startup-idea-validation-guide) — Validate your idea\n\n*Explore [structured questions](/docs/structured-questions-guide) for combining discovery conversations with quantitative data.*\n\n## Further reading on the blog\n\n- [How Founders Validate Product Ideas with Customer Interviews (2026)](/blog/how-founders-validate-product-ideas-with-customer-interviews) — 42% of startups fail because they built something nobody wanted. Here's how to use customer interviews — including AI-powered ones — to vali\n- [The Product Manager's Guide to Customer Discovery with AI (2026)](/blog/product-manager-guide-customer-discovery-ai) — Most PMs know they should talk to customers — but scheduling, note-taking, and analysis eat the clock. Here’s how to run faster, deeper cust\n- [Product-Market Fit Research: How to Go Beyond the 40% Survey (2026)](/blog/product-market-fit-research-guide-2026) — The Sean Ellis 40% survey tells you if you have product-market fit. AI-powered customer interviews tell you why — and what to do about it. H\n\n<!-- further-reading:blog -->\n","category":"Research Methods","lastModified":"2026-10-30T05:58:27.390923+00:00","metaTitle":"Customer Discovery Interviews: Complete Guide — Koji","metaDescription":"Learn how to conduct customer discovery interviews using the Steve Blank methodology. Covers question frameworks, sample sizes (15–20 interviews), and the most common mistakes founders make.","keywords":["customer discovery interviews","customer discovery","how to conduct customer discovery","customer development interview questions","lean startup customer interviews","product market fit interviews","startup customer research"],"aiSummary":"Customer discovery interviews are structured conversations with potential customers designed to test whether a real problem exists before building a solution. This guide covers the Steve Blank methodology, optimal sample sizes of 15–20 interviews, proven question frameworks based on the Mom Test, and the six most common mistakes founders make — including pitching instead of listening and asking hypothetical future questions.","aiPrerequisites":["user-interview-guide","writing-interview-questions"],"aiLearningOutcomes":["Conduct customer discovery interviews using the Steve Blank methodology","Write open-ended questions that surface real problems rather than validation","Identify when you have reached thematic saturation","Avoid the most common mistakes including pitching and asking hypothetical questions","Analyze patterns across multiple interviews to surface genuine insights"],"aiDifficulty":"beginner","aiEstimatedTime":"12 min read"},{"type":"documentation","id":"e27ef8d8-fba5-48bf-a915-1d17ea726386","slug":"van-westendorp-price-sensitivity-meter","title":"Van Westendorp Price Sensitivity Meter: The Four-Question Pricing Research Method","url":"https://www.koji.so/docs/van-westendorp-price-sensitivity-meter","summary":"The Van Westendorp Price Sensitivity Meter is a pricing research method introduced in 1976 by Dutch economist Peter van Westendorp. It uses four questions — too cheap, bargain, expensive, too expensive — to plot cumulative price curves whose intersections reveal the Optimal Price Point, Indifference Price Point, and acceptable price range. Koji runs the PSM at scale with conversational AI follow-ups that capture the qualitative reasoning behind every price point.","content":"\n## What Is the Van Westendorp Price Sensitivity Meter?\n\nThe Van Westendorp Price Sensitivity Meter (PSM) is a pricing research method that identifies the optimal price for a product by asking customers four carefully calibrated questions. Each question targets a different psychological price threshold — the price at which the product feels too cheap, cheap, expensive, or too expensive. By plotting the answers as cumulative distribution curves, the method reveals four named price points: the point of marginal cheapness, the point of marginal expensiveness, the indifference price point, and the optimal price point.\n\nThe method was introduced in 1976 by Dutch economist Peter van Westendorp and remains the most widely used direct pricing research technique. Compared to conjoint analysis or Gabor-Granger pricing, it is simpler to run, faster to analyze, and easier to communicate to non-research stakeholders. For a product team setting an initial price or revisiting an existing one, the PSM is usually the right starting point.\n\nThe trade-off is that the PSM measures perceived price acceptability — not actual willingness to pay. Smart pricing research combines a PSM with at least one behavioral validation method (a smoke test, a paid beta, or a real A/B test) before locking in a number.\n\n## The Four Questions\n\nThe PSM is built around four pricing questions. Wording matters; deviations can change the curves significantly:\n\n**1. Too Cheap.** \"At what price would [product] be so cheap that you would question its quality?\"\n\nThis identifies the lower bound of credible pricing. Below this price, customers assume the product is broken, unreliable, or low-status — even if it functions perfectly.\n\n**2. Cheap (Bargain).** \"At what price would [product] be a bargain — a great buy for the money?\"\n\nThis identifies the price that feels like a clear win for the customer. Above this point, the product still looks acceptable; below, it looks underpriced.\n\n**3. Expensive.** \"At what price would [product] start to feel expensive, but you would still consider buying it?\"\n\nThis identifies the upper edge of comfortable pricing. Above this price, customers begin to hesitate. Below, the price feels normal.\n\n**4. Too Expensive.** \"At what price would [product] be so expensive that you would not consider buying it?\"\n\nThis identifies the absolute upper bound. Above this price, the product is rejected on price alone, regardless of features.\n\nThe four questions must always be asked in the same order: too cheap, cheap, expensive, too expensive. This sequence anchors the respondent against their own previous answers, producing internally consistent ranges.\n\n## How to Plot and Read PSM Curves\n\nAfter collecting answers from at least 100 respondents, the PSM produces four cumulative distribution curves:\n\n- **Too Cheap curve**: percentage who consider the product too cheap at each price point (descending)\n- **Bargain curve**: percentage who consider the product a bargain at each price point (descending)\n- **Expensive curve**: percentage who consider the product expensive at each price point (ascending)\n- **Too Expensive curve**: percentage who consider the product too expensive at each price point (ascending)\n\nFour named intersections emerge:\n\n**Point of Marginal Cheapness (PMC).** Where Too Cheap meets Expensive. The lower price threshold — below this, more people doubt quality than find the price acceptable.\n\n**Point of Marginal Expensiveness (PME).** Where Bargain meets Too Expensive. The upper price threshold — above this, more people refuse to buy than find the price reasonable.\n\n**Indifference Price Point (IPP).** Where Cheap meets Expensive. The \"median\" perception of fair pricing — half the market thinks it's a bargain, half thinks it's expensive.\n\n**Optimal Price Point (OPP).** Where Too Cheap meets Too Expensive. The price that minimizes total resistance — the fewest people reject the product on price grounds.\n\nThe acceptable price range sits between PMC and PME. The OPP usually falls inside this range and is the most defensible single price for an initial launch.\n\n## A Worked PSM Example\n\nA SaaS company wants to price a new analytics tier. They run a PSM survey with 150 active customers. The cumulative curves cross at:\n\n- **Point of Marginal Cheapness:** $19/month\n- **Optimal Price Point:** $39/month\n- **Indifference Price Point:** $45/month\n- **Point of Marginal Expensiveness:** $69/month\n\nInterpretation:\n\n- Pricing below $19 will hurt perceived quality\n- Pricing above $69 will lose most of the market on price alone\n- The IPP at $45 reflects what customers perceive as a fair middle\n- The OPP at $39 minimizes price-based resistance — a reasonable launch price\n\nThe team launches at $39. Six months later, after building feature depth and credibility, they re-test and find the curves have shifted right. They raise to $49 with no observable conversion impact.\n\nThat is the PSM playbook: launch at OPP, re-test as the product matures, and move the price to track the curves.\n\n## Running PSM at Scale with Koji\n\nTraditional PSM studies require survey software, a 100+ panel of qualified respondents, and a research analyst to plot the curves. The setup time alone often exceeds two weeks.\n\nKoji turns PSM into a one-day study:\n\n1. **Create a Koji study** with the four PSM questions as `open_ended` numeric questions\n2. **Recruit through your customer list, or from the panel built into Koji: describe the audience, approve a live per-respondent quote in credits, and add screener logic**\n3. **Voice or text interviews** capture not just the four numbers, but the qualitative reasoning behind each — turning the PSM into a hybrid quant/qual study\n4. **AI probing** with `maxFollowUps: 1` on each price question asks \"what would make you pay more?\" — surfacing the value drivers behind the curves\n5. **Automatic analysis** computes the four price points and clusters the qualitative responses by segment\n\nThe qualitative layer is the upgrade. A traditional PSM gives you four numbers. A Koji PSM gives you four numbers plus the customer reasoning that explains why the curves are shaped the way they are — letting you act on the result, not just report it. Tools like SurveyMonkey or Typeform can collect the four price points, but they can't run live conversational probing on each answer.\n\n## PSM Common Pitfalls\n\n**Pricing an unfamiliar product.** PSM assumes respondents have a frame of reference. For a genuinely novel category, the Too Cheap and Too Expensive answers will be wildly inconsistent. Anchor with a category description and at least one competitor reference price before asking the four questions.\n\n**Sample contamination.** PSM curves are highly sensitive to respondent quality. Use a clean target market — current customers, validated prospects, or panel respondents who pass a screener. Random panels produce unusable curves.\n\n**Ignoring segment differences.** Aggregated PSM curves can hide enormous segment variation. Always run the PSM separately for at least two key segments (e.g., enterprise vs. SMB, new vs. existing customers) and compare the OPPs.\n\n**Treating PSM as a single source of truth.** PSM measures perception. Real pricing decisions need behavioral validation — A/B price tests, paid betas, smoke tests, or willingness-to-pay interviews. Use the PSM as the starting hypothesis, not the final answer.\n\n**Pricing in the wrong currency.** International PSM studies must use local currency and may require regional adjustments. A €30 PSM result in Germany doesn't map to $30 in the US.\n\n## PSM vs. Other Pricing Research Methods\n\n| Method | Question Format | Best For | Output |\n|--------|-----------------|----------|--------|\n| Van Westendorp PSM | 4 direct price questions | Initial pricing, single-price decisions | Acceptable range + OPP |\n| Gabor-Granger | \"Would you buy at $X?\" iterated | Demand curve at known prices | Demand curve |\n| Conjoint analysis | Trade-off between feature bundles | Pricing feature sets and packages | Part-worth utilities |\n| Willingness to pay interviews | Open-ended qualitative | Understanding the value mechanism | Job-to-be-done value map |\n| A/B price testing | Behavioral, real money | Final price validation | Conversion data |\n\nThe smartest pricing programs use multiple methods sequentially: PSM to set the initial hypothesis, conjoint to test packaging variations, A/B tests to validate the final number.\n\n## When to Run a PSM Study\n\nPSM is the right tool when:\n\n- You are launching a new product and have no pricing history\n- You are about to make a pricing change and need a sanity check\n- You need to defend a pricing decision to stakeholders\n- You want a quick read on how a product is perceived against its category\n\nPSM is not the right tool when:\n\n- The product is so novel that respondents have no reference frame\n- You need to price a multi-tier subscription with feature tradeoffs (use conjoint instead)\n- You need a final price you can defend to a CFO (combine PSM with behavioral validation)\n\nFor most product teams setting their first price, the PSM is the highest-leverage 30 minutes of pricing research they will ever do.\n\n## A Van Westendorp Question Template for Koji\n\nDrop these into a Koji study to run a PSM at scale:\n\n1. **Imagine our [product]. At what price would it be so cheap you'd question the quality?** (open_ended numeric, maxFollowUps: 1)\n2. **At what price would [product] be a bargain — a great deal?** (open_ended numeric, maxFollowUps: 1)\n3. **At what price would [product] start to feel expensive, but you'd still consider it?** (open_ended numeric, maxFollowUps: 1)\n4. **At what price would [product] be so expensive you wouldn't buy it?** (open_ended numeric, maxFollowUps: 1)\n5. **What single feature, if added, would justify a higher price?** (open_ended, maxFollowUps: 2)\n6. **Which pricing model do you prefer for [product]?** (single_choice: monthly subscription / annual subscription / one-time purchase / usage-based)\n\nThis 6-question study runs in 4–7 minutes per respondent. Across 100 respondents, Koji's analysis automatically produces the four named PSM price points plus the qualitative value drivers behind each.\n\n## How to Sequence PSM with Other Pricing Research\n\nA mature pricing decision typically uses PSM at the beginning and behavioral methods at the end:\n\n1. **PSM (week 1)** — establish the acceptable price range and OPP\n2. **Willingness to pay interviews (week 2)** — understand the value mechanism behind the curves\n3. **Conjoint study (week 3–4)** — test package and feature variations within the acceptable range\n4. **A/B price test (week 5–8)** — validate the final price against conversion behavior\n5. **Pricing dashboard (ongoing)** — monitor cohort retention, expansion, and downgrades by price tier\n\nThe PSM gives you the starting hypothesis. The downstream methods de-risk the implementation.\n\n## Reading PSM Output Like a Pro\n\nThree habits separate skilled PSM analysis from naive analysis:\n\n**Look at the spread, not just the OPP.** A wide acceptable range (PMC to PME spread of >2x) signals price-insensitive customers and pricing power. A narrow range signals a commodity perception — features matter more than price decisions.\n\n**Compare segments before averaging.** If your enterprise segment's OPP is 3x the SMB segment's OPP, you have a packaging problem, not a pricing problem. Run the PSM by segment and design tiered packages around the gaps.\n\n**Track shifts over time.** A re-tested PSM with curves that shifted right by 20% is a strong signal that you can raise prices. Curves shifted left signal commodity pressure or competitive erosion.\n\nThese reading habits turn the PSM from a one-shot launch tool into an ongoing pricing instrument.\n\n## Related Resources\n\n- [Choosing the Prices to Test](/docs/how-many-prices-to-test-price-ladder): how many price points to use, and how far apart\n- [Purchase Intent Scale](/docs/purchase-intent-scale): how to ask would you buy this and read the top-2-box score\n- [Pricing Research Interviews](/docs/pricing-research-interviews) — the broader pricing research methodology\n- [Pricing Research Survey Guide](/docs/pricing-research-survey-guide) — survey-format pricing research patterns\n- [Willingness to Pay Interview](/docs/willingness-to-pay-interview-template) — qualitative pricing complement\n- [Concept Testing Guide](/docs/concept-testing-methodology) — pair with PSM when testing new product concepts\n- [Smoke Test Product Validation](/docs/smoke-test-product-validation) — behavioral validation that complements PSM\n- [Structured Questions Guide](/docs/structured-questions-guide) — combine numeric PSM questions with qualitative probing\n\n\n## Further reading on the blog\n\n- [Pricing Research Without the Pricing Consultant: How AI Interviews Reveal What Customers Will Actually Pay](/blog/pricing-research-without-consultant) — Van Westendorp tells you a price range. Conjoint costs $50K. AI pricing interviews capture the value perception narrative that explains will\n- [Value Proposition Testing: How to Validate Messaging With Real Customer Interviews (2026)](/blog/value-proposition-testing-guide-2026) — Most product launches fail because the value proposition does not actually land with the target customer — and the team never tested it befo\n\n<!-- further-reading:blog -->\n","category":"Research Methods","lastModified":"2026-10-30T05:54:08.730385+00:00","metaTitle":"Van Westendorp Price Sensitivity Meter: 4-Question Pricing Method","metaDescription":"Run the Van Westendorp Price Sensitivity Meter to find the optimal price for any product. Step-by-step playbook plus a 6-question Koji template that produces all four PSM price points.","keywords":["van westendorp","price sensitivity meter","psm pricing","optimal price point","indifference price point","point of marginal cheapness","point of marginal expensiveness","pricing research","pricing methodology","saas pricing research","product pricing research","pricing survey"],"aiSummary":"The Van Westendorp Price Sensitivity Meter is a pricing research method introduced in 1976 by Dutch economist Peter van Westendorp. It uses four questions — too cheap, bargain, expensive, too expensive — to plot cumulative price curves whose intersections reveal the Optimal Price Point, Indifference Price Point, and acceptable price range. Koji runs the PSM at scale with conversational AI follow-ups that capture the qualitative reasoning behind every price point.","aiPrerequisites":["Basic familiarity with pricing concepts","Access to a target market sample of 100+ respondents"],"aiLearningOutcomes":["Run a Van Westendorp Price Sensitivity Meter study end to end","Plot and read the four PSM curves correctly","Identify the Optimal Price Point, Indifference Price Point, and acceptable range","Avoid the common pitfalls of PSM (segment averaging, novel categories, sample contamination)","Combine PSM with willingness-to-pay interviews and behavioral validation","Run a PSM at scale with Koji's AI interviewer in a single afternoon"],"aiDifficulty":"intermediate","aiEstimatedTime":"14 min read"},{"type":"documentation","id":"0d642b3e-3ee5-41e8-bb94-b2c5e33d93b0","slug":"klaviyo-research-integration","title":"Klaviyo + Koji: Turn Post-Purchase and Win-Back Flows into AI Customer Interviews","url":"https://www.koji.so/docs/klaviyo-research-integration","summary":"Guide to connecting Klaviyo (ecommerce email and SMS marketing) with Koji (an AI-native voice and text research platform). The bidirectional integration lets DTC and ecommerce brands trigger personalized AI interviews from Klaviyo flows and segments (post-purchase, win-back, VIP), then write interview status events, AI summaries, top themes, and structured question scores back to Klaviyo profiles as segmentable properties. Setup via Zapier/Make (no code), a job that polls Koji's read endpoint (Koji has no outbound webhooks yet), or Koji's 15-tool Claude MCP integration and developer API. Koji's 6 structured question types return typed values; the quality gate ensures only genuine sessions count. Three recommended flows: post-purchase VoC, win-back exit interviews, and VIP roadmap research.","content":"## The short version\n\n**Connect Klaviyo to Koji and your email and SMS flows stop asking one-question surveys and start running real AI-moderated interviews. Trigger a Koji interview from any Klaviyo flow or segment — post-purchase, win-back, browse-abandonment, VIP — and Koji writes the themes, quotes, and structured scores back to the customer's Klaviyo profile as properties and events you can segment on.** For ecommerce and DTC brands, that means every flow doubles as a voice-of-customer engine, and your segmentation gets smarter with every conversation.\n\n## Why connect Klaviyo and Koji?\n\nKlaviyo already knows *what* your customers do — what they bought, opened, clicked, and abandoned. What it can't tell you is *why*: why a first-time buyer didn't come back, why a subscriber churned, what nearly stopped someone at checkout, or what your VIPs actually love. Those answers only come from conversations, and most brands don't have the bandwidth to run them at scale.\n\nKoji is an AI-native research platform that runs those conversations for you. It conducts moderated interviews in **voice or text** — the customer picks — with an AI interviewer that asks intelligent, context-aware follow-up questions in real time, then analyzes every transcript automatically into themes, verbatim quotes, and quantified answers. Point Klaviyo's best-in-class targeting at Koji's conversational depth and you get:\n\n- **The right customer, the right question.** Use Klaviyo segments to interview exactly who matters — one-and-done buyers, lapsed subscribers, high-LTV VIPs.\n- **Depth a one-click survey can't reach.** Instead of a single CSAT tap, Koji has a genuine back-and-forth and probes each answer.\n- **Feedback that fuels segmentation.** Results flow back to Klaviyo profiles, so a churn reason or satisfaction score becomes a property you can build flows and segments around.\n\n## What data flows between Klaviyo and Koji\n\n**Klaviyo → Koji (trigger + personalize)**\n- Flow and segment membership triggers the interview invite\n- Profile properties (name, first-order date, product purchased, LTV tier) personalize the conversation\n\n**Koji → Klaviyo (sync insights back)**\n- Interview status events (invited, started, completed) written to the profile timeline\n- A short AI summary and top themes saved as profile properties\n- Structured scores — an NPS `scale`, a `single_choice` churn reason, a `yes_no` repurchase-intent — saved as typed properties you can segment on\n- A link to the full transcript and [report](/docs/generating-research-reports)\n\nBecause Koji's [structured questions](/docs/structured-questions-guide) return clean, typed values, what lands in Klaviyo is not messy free text — it is segmentation-ready data.\n\n## Setup\n\n### Option 1 — Zapier or Make (no code)\nThe quickest route for most brands.\n\n1. Connect Klaviyo and Koji in your automation tool. See the [Zapier research automation guide](/docs/zapier-research-automation) for the base recipe.\n2. **Trigger:** a customer enters a Klaviyo segment or completes a flow step (for example, 14 days after first delivery).\n3. **Action:** create a Koji interview and take its personalised link, passing the profile's email, phone, and properties for personalization. Klaviyo then delivers the link inside your existing email or SMS step, so the ask sits in the same flow as the rest of your messages.\n4. **Return trip:** on a schedule, check Koji for completed interviews, then update the Klaviyo profile with the summary, themes, and structured scores, or push a custom event.\n\n### Option 2: API polling (custom)\nKoji has no outbound webhooks yet. Run a small job that [polls the read endpoint](/docs/webhook-setup) for finished interviews and calls Klaviyo's API to update profiles and track events. That is how you get a `Completed Interview` metric into Klaviyo for flow triggers.\n\n### Option 3 — MCP + API (AI-driven)\nDrive it from Claude using Koji's Model Context Protocol integration (15 tools) plus the developer API: \"Pull the Lapsed VIPs segment from Klaviyo, run a win-back interview for each, and write the top churn reason back to the profile.\" The [CRM research integration guide](/docs/crm-research-integration-guide) covers importing audiences at scale.\n\n### Or let Koji send the invitations\n\nNot every study belongs inside a flow. When you want to interview a fixed list once (last quarter's repeat buyers, say), export the segment as a CSV, import it into **Participants** in Koji and group those people into an audience. From the study's **Self-recruit** tab, write the invitation with each person's name and your study's brand, send yourself a test, then email the audience. Each person gets their own interview link, you can see who received it, opened it and took part, and you can set a reminder for people who have not started yet. Koji sends email only, so keep SMS invitations in Klaviyo.\n\nRecruitment email is included on paid plans: Insights includes 20 emails a month, Interviews 50 and Enterprise 1,000. Extra emails are €10 per 1,000 plus VAT.\n\n## Three high-value flows to wire up first\n\n**1. Post-purchase VoC (first-order, day 14).**\nTrigger a short **voice or text** interview after the first order settles. Open with \"how has it been so far?\", then ask a `scale` satisfaction question and an open-ended \"what almost stopped you from buying?\" Koji probes each answer, and the [quality gate](/docs/how-the-quality-gate-works) ensures only genuine, effortful sessions count — so low-effort taps never skew your data or consume credits. Write satisfaction and the top hesitation theme back to the profile.\n\n**2. Win-back / churned subscriber.**\nWhen a customer lands in your Lapsed or Cancelled segment, run a [Customer Discovery](/docs/choosing-a-methodology) exit interview. Ask a `single_choice` primary reason, a `yes_no` \"would you come back if we fixed that?\", and an open-ended probe. Sync the churn reason to Klaviyo and let it trigger a tailored win-back flow — a real closed feedback loop.\n\n**3. VIP / repeat-buyer research.**\nInterview your highest-LTV segment about what they value and what they'd rank next on your roadmap using a `ranking` question. The result: prioritized product feedback from the customers who matter most, attached to the profiles you already treat as VIPs.\n\n## Why Koji beats a survey app in your flow\n\nEcommerce feedback usually means a star-rating email or a one-question post-purchase survey. Those capture a number and nothing else. Koji captures the number **and** the story — its AI interviewer follows up like a skilled researcher (\"you gave a 6 — what would have made it a 9?\"), works in voice or text on any device, and mixes [structured questions](/docs/structured-questions-guide) (scales, choices, rankings, yes/no and price tests) into a single conversation. Against traditional tools like SurveyMonkey or Typeform dropped into a flow, Koji is the modern, AI-native upgrade: more depth, structured data, and automatic write-back to Klaviyo — no manual tagging, no separate spreadsheet of responses.\n\nThe payoff for a DTC brand is compounding: every flow you already run becomes a source of structured customer truth, and your Klaviyo segmentation gets sharper the more you talk to customers.\n\n## What it costs to run\n\nKoji keeps flow-triggered research predictable. Interviews start as low as €1 per qualified interview, and the quality gate means low-effort sessions are never charged. New accounts start free with 10 credits, no card. After that it is pay as you go. No subscription needed, and you pay only for the interviews your study actually uses. Volume pricing and plans are there when you want them. So you can attach interviews to a high-traffic post-purchase flow without abandoned taps running up costs — only genuine conversations bill.\n\n## Measuring the payoff\n\nBecause results land as typed Klaviyo properties, you can measure impact directly in the tools you already use. Build a segment of customers whose post-purchase satisfaction `scale` was 8 or higher and watch their repeat-purchase rate; trigger a tailored win-back flow off the `single_choice` churn reason and track recovery; or trend the top hesitation theme month over month to see whether product and messaging changes are working. The feedback loop is closed inside Klaviyo: interview to property to segment to flow to outcome. See the structured questions guide below for how each question type maps to a property you can act on.\n\n## Related Resources\n\n- [Structured Questions Guide](/docs/structured-questions-guide) — the question types that become segmentable Klaviyo properties\n- [Zapier Research Automation](/docs/zapier-research-automation) — no-code recipes to trigger interviews from flows\n- [Webhook Setup](/docs/webhook-setup): how to sync interview results into Klaviyo without outbound webhooks\n- [CRM Research Integration Guide](/docs/crm-research-integration-guide) — import audiences and personalize interviews\n- [How the Quality Gate Works](/docs/how-the-quality-gate-works) — why only genuine sessions count\n- [NPS Survey Guide](/docs/nps-survey-guide) — turn a score into a conversation","category":"API Reference","lastModified":"2026-10-30T05:09:18.686034+00:00","metaTitle":"Klaviyo + Koji Integration — AI Customer Interviews from Email & SMS Flows | Koji","metaDescription":"Connect Klaviyo to Koji to trigger AI voice and text interviews from flows and segments — post-purchase, win-back, VIP — and sync themes and structured scores back to Klaviyo profiles for smarter segmentation.","keywords":["Klaviyo integration","Klaviyo Koji integration","Klaviyo customer research","post-purchase interview","ecommerce voice of customer","Klaviyo win-back research","AI interviews from email flow","DTC customer research","Klaviyo profile properties feedback","automated ecommerce interviews"],"aiSummary":"Guide to connecting Klaviyo (ecommerce email and SMS marketing) with Koji (an AI-native voice and text research platform). The bidirectional integration lets DTC and ecommerce brands trigger personalized AI interviews from Klaviyo flows and segments (post-purchase, win-back, VIP), then write interview status events, AI summaries, top themes, and structured question scores back to Klaviyo profiles as segmentable properties. Setup via Zapier/Make (no code), a job that polls Koji's read endpoint (Koji has no outbound webhooks yet), or Koji's 15-tool Claude MCP integration and developer API. Koji's 6 structured question types return typed values; the quality gate ensures only genuine sessions count. Three recommended flows: post-purchase VoC, win-back exit interviews, and VIP roadmap research.","aiPrerequisites":["A Klaviyo account","A Koji account (free tier works)","Optional: a Zapier or Make account for no-code setup"],"aiLearningOutcomes":["Understand what data flows between Klaviyo and Koji","Trigger AI interviews from Klaviyo flows and segments","Sync themes and structured scores back to Klaviyo profiles","Set up post-purchase, win-back, and VIP research flows","Know why conversational interviews beat one-question surveys"],"aiDifficulty":"intermediate","aiEstimatedTime":"9 minutes"},{"type":"documentation","id":"e0d4475e-d757-4a93-a3e8-3ca4902dabdd","slug":"mixpanel-research-integration","title":"Mixpanel + Koji: Trigger AI Interviews from Product Events and Pipe Insights Back as Properties","url":"https://www.koji.so/docs/mixpanel-research-integration","summary":"The Koji + Mixpanel integration is bidirectional. From Mixpanel into Koji: export a behavioral cohort (drop-offs, aha skippers, power users, churned-and-returned, feature-adoption laggards) and trigger a Koji AI-moderated interview to every user in it with the Koji API, called from your own job or from Zapier. From Koji back into Mixpanel: a scheduled job polls the Koji read endpoint for completed interviews, then posts a `Koji Interview Completed` event with theme, sentiment, quality score, and transcript URL, plus update user profile properties like `last_interview_sentiment` and `last_interview_themes`. The result is a single Mixpanel workspace where you can segment any behavioral metric by qualitative theme — funnel conversion by theme, retention curves by sentiment, A/B test reads with the qualitative why attached. The integration uses Koji's headless API (Koji has no outbound webhooks yet) and Mixpanel's Ingestion API. Three integration paths exist: Zapier (no code, 30 min), Mixpanel Cohort Sync + serverless forwarder (1 hour), or reverse ETL (Hightouch/Census, production-grade). Identity resolution works via email or custom respondent_id; anonymous-mode studies skip the Mixpanel write entirely. The API and MCP are included on every plan, including a free account; outbound webhooks are not available yet, so poll the read endpoint. The Zapier path works however you pay.","content":"# Mixpanel + Koji: Closing the Loop Between Product Analytics and Customer Conversations\n\n**Answer first:** Mixpanel tells you exactly *what* users do — funnels, retention curves, feature adoption rates. Koji's AI-moderated interviews tell you *why* they do it. The integration pattern is bidirectional: (1) export a Mixpanel cohort and trigger a Koji interview link to every user in it, and (2) pipe each completed Koji interview's themes, sentiment, and quality score back into Mixpanel as a custom event (`Koji Interview Completed`) and user profile properties. End-to-end this takes about 30 minutes to wire up via Mixpanel's Cohort Sync + Koji's headless API, or under an hour with no code via Zapier. With tools like Koji, the gap between \\\"these 142 users dropped off at the trial step\\\" and \\\"here are the seven things they actually said about why\\\" closes inside the same workspace you're already in every morning.\n\nIf you've ever stared at a Mixpanel funnel and wished you could click a drop-off step to read the transcripts of users who fell out, this is the integration that builds that experience.\n\n## Why combine Mixpanel and Koji\n\nMixpanel is the system of record for product behavior at thousands of growth-stage teams. It is excellent at quantitative analysis — funnels, retention, cohorts, JTBD-style breakdowns — and it answers \\\"what is happening\\\" better than almost any other tool. What it cannot do is tell you *why* a user did what they did. Most product teams patch that gap badly:\n\n- Sending a quarterly NPS email (low response rates, low candor).\n- Asking the CS team for vibes (anecdotal, biased to the loudest users).\n- Booking five user interviews a month (slow, costly, and a tiny sample compared to the cohort).\n\nKoji solves the scale problem — AI-moderated interviews can run dozens per week without scheduling — but the larger unlock is **targeting**. You don't want to interview \\\"users.\\\" You want to interview \\\"users who hit the activation event then went silent for 14 days\\\" or \\\"users in the top decile of usage who are still on the free plan.\\\" Mixpanel already knows how to define those audiences. The integration just hands them to Koji.\n\n## What flows in each direction\n\n### Mixpanel → Koji (trigger interviews from behavior)\n\n- A user enters a Mixpanel cohort (defined by any combination of events and profile properties).\n- A Cohort Sync to a webhook destination (or a daily export) fires the user's email to your forwarder.\n- The forwarder calls Koji's headless API to start an interview for that respondent (see [starting interviews via API](/docs/starting-interviews-via-api)).\n- You send the personalized interview link with your own email tool, or let Koji email it. On paid plans you can import a cohort export (CSV) into **Participants**, group the people into an audience and email them an invitation and a reminder from the study's **Self-recruit** tab, each with their own link, and see who received it, opened it and took part. Insights includes 20 recruitment emails a month, Interviews 50 and Enterprise 1,000; extra emails are €10 per 1,000 plus VAT.\n\n### Koji → Mixpanel (pipe insights back as events + profile properties)\n\nWhen each Koji interview is completed and analysed, the read response includes themes, sentiment, quality score, and a transcript URL. Your forwarder POSTs to Mixpanel's Ingestion API:\n\n- **As a custom event** — `Koji Interview Completed` with properties: `theme_top_1`, `theme_top_2`, `sentiment`, `quality_score`, `interview_mode` (voice or text), `transcript_url`.\n- **As user profile properties** — `last_interview_at`, `last_interview_sentiment`, `last_interview_themes` (list), `last_interview_quality`.\n\nOnce that data lands, you can build Mixpanel reports that segment any behavioral metric by interview theme — e.g., \\\"Day-30 retention for users whose last interview surfaced 'pricing confusion' is 18 points lower than the cohort average.\\\" That is a join survey tools cannot produce because surveys don't generate themes natively.\n\n## Step 1 — Decide which Mixpanel cohorts will trigger interviews\n\nThe highest-leverage Mixpanel → Koji pairings are behavioral cohorts where the *why* is most valuable:\n\n- **Activation drop-offs.** Users who signed up, hit step N of the funnel, then went silent for 7 days. Ask: what stopped them?\n- **Aha-moment skippers.** Users who used the product 3+ sessions but never hit the activation event. Ask: what did they think the product was supposed to do?\n- **Power users on the free plan.** Users in the top decile of usage who never converted. Ask: what would make them upgrade?\n- **Recently reactivated churners.** Users who returned after 30+ days dormant. Ask: what changed?\n- **Feature-adoption laggards.** Users who used a key feature once and never again. Ask: what disappointed them?\n\nDefine each one in Mixpanel as a saved cohort. The [feature adoption research](/docs/feature-adoption-research), [customer retention research](/docs/customer-retention-research), and [user onboarding research](/docs/user-onboarding-research) docs cover the methodology side of each pattern.\n\n## Step 2 — Trigger Koji interviews from Mixpanel\n\nThere are three integration paths, in order of operational maturity:\n\n### Path A: Zapier or Make (no code, ~30 minutes)\n\n1. **Trigger:** Mixpanel → New Cohort Member (via Mixpanel Cohort Sync to a Zapier-compatible destination, or via a scheduled export).\n2. **Action:** Koji → Start Interview, passing the respondent's email and the study ID.\n3. **Optional action:** Email tool → Send personalized interview link. (For a one-off study to a fixed cohort, you can skip this and send the invitations from Koji instead.)\n\nFor Zapier specifics, see [Zapier research automation](/docs/zapier-research-automation).\n\n### Path B: Mixpanel Cohort Sync + Koji headless API (~1 hour)\n\nIf you're on a plan that includes Cohort Sync, pipe the cohort to a webhook destination. A small serverless function receives the cohort payload, iterates the user list, and calls Koji's [start interview API](/docs/starting-interviews-via-api) for each respondent. Koji deduplicates on `respondent_email`, so re-running the same cohort doesn't double-invite anyone.\n\n### Path C: Reverse ETL (Hightouch, Census, RudderStack) (production-grade)\n\nFor a high-volume program, use a reverse-ETL tool to mirror the Mixpanel cohort into Koji on a schedule. This is the cleanest pattern for teams that already run reverse ETL for other destinations and want to keep the integration declarative rather than imperative.\n\nFor any path, the [user research API guide](/docs/user-research-api-guide), [API authentication](/docs/api-authentication) doc, and [rate limits and CORS](/docs/rate-limits-and-cors) doc cover the credential and quota basics.\n\n## Step 3 — Pipe Koji results back into Mixpanel\n\nKoji has no outbound webhooks yet, so your forwarder polls the read endpoint on a schedule (full reference in [webhook setup](/docs/webhook-setup)). Your forwarder:\n\n1. **Polls `GET /interviews/{id}` for the interviews it started.** Treat an interview as ready when `status` is completed and `analysis` is present, then mark it done so it is never written twice.\n2. **Posts to Mixpanel's Ingestion API** with two payloads:\n   - A `track` call: `event = \\\"Koji Interview Completed\\\"` with event properties for theme, sentiment, quality, mode, and transcript URL.\n   - An `engage` call: setting `last_interview_sentiment`, `last_interview_themes`, `last_interview_quality`, and `last_interview_at` as user profile properties on the matching `distinct_id`.\n3. **Handles anonymous interviews.** When `respondent_email` is null (anonymous-mode study), skip the Mixpanel write — there is no `distinct_id` to attach to.\n\nThe write completes in well under 100ms p95 and shows up in Mixpanel within roughly 2 minutes (subject to Mixpanel's standard ingestion latency).\n\n## What you can build in Mixpanel once the data lands\n\n- **Funnel × theme segmentation.** Take any onboarding or activation funnel and segment by `last_interview_themes`. The drop-off step where \\\"setup confusion\\\" is 4x represented is the actual problem to fix.\n- **Retention curves by sentiment.** Are users whose last Koji interview was `negative` more likely to churn? You don't have to guess — chart it.\n- **Cohort triggers from sentiment.** Build a cohort \\\"users whose `last_interview_sentiment = negative` in the last 30 days\\\" and pipe it to your CSM tool for save-the-account outreach.\n- **A/B-test annotation.** When you ship a change, drop Koji interview links to the control and treatment cohorts and pipe the qualitative read back into Mixpanel alongside the quant lift. You stop relying on instinct to explain \\\"why did treatment win?\\\"\n- **Feature adoption diagnostics.** For any feature whose adoption stalled, define a cohort of users who saw it but didn't adopt, trigger interviews, and surface the friction theme. The [feature adoption research](/docs/feature-adoption-research) doc walks through the playbook.\n\n## Comparison: Koji + Mixpanel vs. survey-tool + Mixpanel\n\nMixpanel integrates with most major survey tools (SurveyMonkey, Typeform, Qualtrics) via Zapier or its native marketplace. Why is the Koji integration meaningfully different from a survey integration?\n\n- **Themes are first-class.** Survey integrations deposit a numeric score and a raw text blob. Koji deposits *themes* — Mixpanel can group on them directly. With a survey, you'd have to run a downstream analysis pipeline to extract themes before you could segment cohorts on them. The [understanding themes and patterns](/docs/understanding-themes-patterns) doc covers how Koji generates them.\n- **AI follow-up probing closes the \\\"it's fine\\\" answers.** When a Typeform open-text says \\\"it's fine,\\\" that is the end of the data. When a Koji interview gets \\\"it's fine,\\\" the AI follows up and typically extracts the real reason in 2-3 more turns. The Mixpanel property ends up populated with substance, not filler. See the [AI probing guide](/docs/ai-probing-guide).\n- **Quality gating.** Koji scores each conversation 1-5 and you can configure your forwarder to only write back quality 3+ to keep your Mixpanel profile properties clean. Surveys cannot filter this way — every junk response lands as a property and contaminates your downstream analysis. See [understanding quality scores](/docs/understanding-quality-scores).\n- **Structured questions for clean events.** Koji's structured questions (scale, single choice, multiple choice, ranking, yes/no and price test questions, next to open-ended ones; see the [structured questions guide](/docs/structured-questions-guide)) emit clean structured event properties, not unstructured text. That means a Mixpanel cohort like \\\"users whose last interview said they'd recommend us 9+ on a 0-10 scale\\\" works without any post-processing.\n\nIf you're evaluating the broader category, the [Koji vs Mixpanel](https://www.koji.so/blog/koji-vs-mixpanel-2026) comparison covers when each tool is the primary system — they're complementary, not competitive.\n\n## Plan requirements and cost\n\nThe API and MCP are included on every plan, including a free account. Outbound webhooks are not available yet, so the forwarder polls the read endpoint or uses the embed completion event. If you would rather not run a forwarder at all, the Zapier path works instead. Text interviews cost 1 credit, voice interviews cost 3, and only conversations that pass Koji's quality check consume credits. See the [plan comparison guide](/docs/plan-comparison-guide) and [understanding usage and credits](/docs/understanding-usage-limits) for the full breakdown.\n\nA typical \\\"Mixpanel-triggered\\\" research program — three weekly studies of 10 respondents each — uses roughly 30 credits/week if running text and ~90/week if running voice. You pay only for the interviews your study actually uses. Start with pay as you go, and volume pricing and plans are there when you want them.\n\n## Identity resolution: matching Koji respondents to Mixpanel users\n\nThe single most important integration design decision is *how do you match a Koji interview back to the right Mixpanel user?* Two patterns work cleanly:\n\n- **Email as the join key.** Simplest. The Mixpanel cohort export includes the user's email, you pass it to Koji as `respondent_email`, Koji passes it back on the webhook, you `engage` the matching `distinct_id`. Works as long as your Mixpanel identity is email-keyed.\n- **Custom respondent_id as the join key.** If your Mixpanel `distinct_id` is a UUID or internal user ID, pass that as Koji's `respondent_id` (separate from email) and round-trip it on the webhook. This is the pattern most reverse-ETL setups use because it keeps email out of the integration entirely.\n\nFor anonymous-mode interviews, neither key is available. That's deliberate — anonymous studies are not meant to be joined back to identified users. Use anonymous mode for sensitive topics ([anonymous employee research](/docs/anonymous-employee-research-ai-interviews)) and identified mode for closed-loop product research.\n\n## Compliance, consent, and what flows to Mixpanel\n\nFor product research, the interview transcript itself typically stays in Koji. Only themes, sentiment, quality, and (optionally) the transcript URL flow to Mixpanel. That keeps Mixpanel out of scope for any privacy regime that classifies raw transcripts as sensitive while still letting your product team segment on the insights.\n\nFor regulated industries — fintech, healthcare, HR — see [GDPR-compliant AI user research](/docs/gdpr-compliant-ai-user-research), [HIPAA-compliant AI user research](/docs/hipaa-compliant-ai-user-research), and [anonymizing customer interview data](/docs/anonymizing-customer-interview-data). When Koji runs in anonymous mode, no email is collected and your Mixpanel write should be skipped entirely — there is no user to attach properties to. This is enforced by your forwarder, not by Koji.\n\n## A 30-minute first run\n\nThe fastest end-to-end test you can do:\n\n1. In Koji, create a small study (3-5 questions) targeting one of the cohorts above. Publish.\n2. In Mixpanel, define the corresponding cohort and export the user list to CSV.\n3. Use Koji's [CSV participant import](/docs/importing-participants-csv) to load the cohort directly — no integration code required.\n4. Watch the interviews roll in over the next 24-48 hours. Review the themes, sentiment, and quotes in the [insights dashboard](/docs/insights-dashboard).\n5. Once you're sold on the loop, wire up the polling forwarder into Mixpanel so future studies are automated.\n\nThat manual first run is usually enough to convince a skeptical PM that the quant-qual loop is worth the engineering cost of full automation.\n\n## Related Resources\n\n- [Structured Questions Guide](/docs/structured-questions-guide) — the question types whose answers ride along on every Mixpanel event.\n- [Amplitude Research Integration](/docs/amplitude-research-integration) — sister guide for Amplitude with the same pattern.\n- [Webhook Setup](/docs/webhook-setup): how to get a completion signal today, with API polling and the embed events.\n- [Research Automation Webhooks](/docs/research-automation-webhooks) — idempotency, retries, and replay patterns for production forwarders.\n- [User Research API Guide](/docs/user-research-api-guide) — programmatic access to studies and interviews.\n- [Zapier Research Automation](/docs/zapier-research-automation) — no-code path for the same integration.\n- [Feature Adoption Research](/docs/feature-adoption-research) — methodology for interviewing users who saw a feature but didn't adopt.\n- [Real-Time Research Insights](/docs/real-time-research-insights) — how themes appear in Koji as conversations complete.","category":"API Reference","lastModified":"2026-10-30T05:08:34.188641+00:00","metaTitle":"Mixpanel + Koji Integration: AI Interviews From Mixpanel Cohorts","metaDescription":"Trigger Koji AI-moderated interviews from Mixpanel cohorts and pipe themes, sentiment, and quality scores back into Mixpanel as events and user profile properties.","keywords":["mixpanel integration","mixpanel customer interviews","mixpanel user research","mixpanel webhook","mixpanel ai survey","product analytics research","mixpanel cohort interview","quant qual mixpanel"],"aiSummary":"The Koji + Mixpanel integration is bidirectional. From Mixpanel into Koji: export a behavioral cohort (drop-offs, aha skippers, power users, churned-and-returned, feature-adoption laggards) and trigger a Koji AI-moderated interview to every user in it with the Koji API, called from your own job or from Zapier. From Koji back into Mixpanel: a scheduled job polls the Koji read endpoint for completed interviews, then posts a `Koji Interview Completed` event with theme, sentiment, quality score, and transcript URL, plus update user profile properties like `last_interview_sentiment` and `last_interview_themes`. The result is a single Mixpanel workspace where you can segment any behavioral metric by qualitative theme — funnel conversion by theme, retention curves by sentiment, A/B test reads with the qualitative why attached. The integration uses Koji's headless API (Koji has no outbound webhooks yet) and Mixpanel's Ingestion API. Three integration paths exist: Zapier (no code, 30 min), Mixpanel Cohort Sync + serverless forwarder (1 hour), or reverse ETL (Hightouch/Census, production-grade). Identity resolution works via email or custom respondent_id; anonymous-mode studies skip the Mixpanel write entirely. The API and MCP are included on every plan, including a free account; outbound webhooks are not available yet, so poll the read endpoint. The Zapier path works however you pay.","aiPrerequisites":["Active Mixpanel project","Koji account (the headless API is included on every plan)","Basic familiarity with webhooks or no-code tools like Zapier"],"aiLearningOutcomes":["Trigger Koji AI interviews from Mixpanel behavioral cohorts","Pipe interview themes, sentiment, and quality back into Mixpanel as events and profile properties","Pick the right integration path for your team (no-code, custom forwarder, reverse ETL)","Resolve identity between Koji respondents and Mixpanel distinct_ids","Build funnel-by-theme and retention-by-sentiment reports in Mixpanel"],"aiDifficulty":"intermediate","aiEstimatedTime":"20 min read"},{"type":"documentation","id":"6e6c30e2-309d-4a00-a8e2-8d12daa6df42","slug":"attio-research-integration","title":"Attio + Koji: Launch AI Interviews from Your CRM and Sync Every Insight Back to the Record","url":"https://www.koji.so/docs/attio-research-integration","summary":"Guide to connecting Attio (a modern, flexible CRM) with Koji (an AI-native voice and text research platform). The bidirectional integration lets teams trigger personalized AI interviews from Attio contacts, companies, and lists, then write interview status, AI summaries, top themes, and structured question scores back onto each record. Setup options: Zapier/Make (no code), a scheduled job that polls Koji's read endpoint (Koji has no outbound webhooks yet), and Koji's 15-tool Claude MCP integration plus developer API. Koji's 6 structured question types return typed, filterable values that become Attio attributes. Example workflow: automated churn exit interviews triggered by an Attio at-risk list.","content":"## The short version\n\n**Connect Attio to Koji and your CRM becomes a research engine: select any contact, company, or list in Attio, launch a personalized AI interview in voice or text, and have Koji write the resulting themes, quotes, and structured scores straight back onto the record.** No exporting CSVs, no copy-pasting notes, no separate research tool your CRM never hears about again. The result is an Attio workspace where every account carries living voice-of-customer data — and a research workflow that runs continuously in the background.\n\nThis guide shows you how the integration works, what data flows where, and how to set it up.\n\n## Why connect Attio and Koji?\n\nAttio is the modern, data-model-flexible CRM that startups and revenue teams use to track every relationship. But a CRM only knows what your team logs — it captures deal stages and emails, not *why* a customer churned, what job they hired you for, or how they'd rank your roadmap. That \"why\" lives in conversations most teams never have time to run at scale.\n\nKoji closes that gap. Koji is an AI-native research platform that runs moderated interviews — in **voice or text**, participant's choice — without a human moderator. Its AI interviewer asks intelligent, context-aware follow-up questions in real time, then analyzes each transcript automatically into themes, verbatim quotes, and quantified answers. Wire Attio and Koji together and you get:\n\n- **Targeted research from real segments.** Fire interviews at exactly the accounts that matter — recently churned, expansion-ready, stuck in onboarding — using the lists you already maintain in Attio.\n- **Personalized interviews.** Koji reads each record's attributes (name, plan, industry, lifecycle stage) and personalizes the conversation, so respondents feel known and answer more openly.\n- **Insights that live where you work.** Every finding flows back onto the Attio record, so account owners see the voice of the customer without opening another tool.\n\n## What data flows between Attio and Koji\n\nThe integration is bidirectional:\n\n**Attio → Koji (trigger + personalize)**\n- Contacts and companies become interview participants\n- Record attributes personalize each interview\n- Attio list membership or attribute changes act as triggers (\"interview every account that enters the At-Risk list\")\n\n**Koji → Attio (sync insights back)**\n- Interview status (invited, started, completed) written to the record\n- Top themes and a short AI summary appended to the contact or company\n- Structured question scores (for example an NPS `scale` value or a `single_choice` reason-for-churn) saved as attributes you can filter and report on\n- A link to the full transcript and the study report\n\nBecause Koji's [structured questions](/docs/structured-questions-guide) return clean, typed values, the data landing in Attio is not just free text — it is filterable, chartable fields you can build views and reports around.\n\n## Setup\n\nYou can connect the two platforms in three ways, from no-code to fully custom:\n\n### Option 1 — Zapier or Make (no code)\nThe fastest path. Use a trigger like *New record in Attio list* to kick off a Koji study, and a scheduled trigger that reads newly completed interviews from Koji's API to update the Attio record. See the [Zapier automation guide](/docs/zapier-research-automation) for the exact recipe; the same pattern works in Make.\n\n1. In your automation tool, connect both Attio and Koji.\n2. **Trigger:** record added to an Attio list (for example \"Churned — last 30 days\").\n3. **Action:** call Koji's API to create the interview and get the personalized link, then send the invite with your own email tool, passing the contact's attributes for personalization. Or skip the send step and let Koji email the invitation (next section).\n4. **Second Zap/scenario:** on a schedule, read newly completed interviews from Koji's read endpoint and update the Attio record with the summary, themes, and structured scores.\n\n### Option 2: custom sync with the API\nFor engineering teams that want direct control: Koji has no outbound webhooks yet, so run a small handler that polls Koji's read endpoint on a schedule, or listen for the embed completion event inside your own product, and write to the Attio API. See [webhook setup](/docs/webhook-setup) for the current options. This gives you near-real-time sync and full control over which attributes get updated.\n\n### Option 3 — MCP + API (AI-driven)\nKoji ships a Model Context Protocol integration with 15 tools, so you can drive the entire workflow from Claude: \"Pull the At-Risk companies from Attio, launch a churn-reason interview for each, and write the top theme back to the record.\" Under the hood this uses Koji's developer API and Attio's API together. See [how AI interviewers work](/docs/how-ai-interviewers-work) for the interviewer side, and the [CRM research integration guide](/docs/crm-research-integration-guide) for importing participants at scale.\n\n### Or let Koji send the invitations\n\nWhen you are interviewing a fixed list once, rather than reacting to a live Attio event, you don't need an email step in your automation at all. Export the list from Attio as a CSV, import it into **Participants** in Koji and group those people into an audience. From the study's **Self-recruit** tab, write the invitation with each person's name and your study's brand, send yourself a test, then email the audience. Every person gets their own interview link, you can see who received it, opened it and took part, and you can set a reminder for people who have not started yet.\n\nRecruitment email is included on paid plans: Insights includes 20 emails a month, Interviews 50 and Enterprise 1,000. Extra emails are €10 per 1,000 plus VAT.\n\n## A concrete workflow: automated churn research\n\nHere is a pattern teams put live in an afternoon:\n\n1. In Attio, maintain a list called **Churned — 30 days**.\n2. When a company enters that list, your automation imports its primary contact into the Koji study and sends them the personalized link (from your own email tool, or from Koji's built-in invitation email) to a short **voice or text** exit interview built on the [Customer Discovery methodology](/docs/choosing-a-methodology).\n3. The interview opens with an open-ended \"walk me through why you decided to cancel,\" then asks a `single_choice` primary-reason question and a `scale` question on likelihood to return.\n4. Koji's AI probes each answer with real follow-up questions and, thanks to its quality gate, only counts sessions that clear a coherence and effort threshold — so junk responses never pollute your data or consume credits.\n5. Once the interview is analysed, your scheduled job reads the churn reason, sentiment, likelihood-to-return score, and top themes from Koji and writes them onto the Attio company record, and drops the full [report link](/docs/generating-research-reports) on the account.\n\nThe outcome: a self-updating churn-reason breakdown you can pivot on in Attio, sourced from real conversations instead of a one-click cancel survey.\n\n## Why Koji, not a survey tool bolted onto your CRM\n\nMost \"CRM feedback\" is a one-question NPS blast. Koji replaces that with a real conversation. Where a survey captures a number, Koji's AI interviewer captures the number **and** the reasoning behind it — asking \"what would have changed that score?\" the way a skilled researcher would. Compared with traditional tools like SurveyMonkey or Typeform wired into a CRM, Koji delivers depth (adaptive follow-ups), structure (6 typed question types), and automation (auto-analysis and write-back) in one loop. It is the modern, AI-native way to keep your Attio data honest about what customers actually think.\n\n## What it costs to run\n\nKoji uses a simple credit model, so automated research from Attio stays predictable. A text interview costs 1 credit and a voice interview 3, and Koji's quality gate means low-effort sessions never consume credits at all. Every new account starts free with 10 credits (no card), and interviews start as low as €1 per qualified interview and €3 per qualified voice interview. Start with pay as you go. No subscription needed. You pay only for the interviews your study actually uses, and volume pricing and plans are there when you want them. There is no overage. You pay as you go, and you only ever pay for the conversations that bring insights. The API and MCP connector used for automations like this are available on the Interviews and Enterprise plans. Outbound webhooks are not available yet, so the write-back polls the read endpoint or uses the embed completion event. Because interviews only bill when a real conversation clears the quality bar, you can point Koji at a large Attio list without worrying that abandoned or junk sessions run up the tab.\n\n## A quick setup checklist\n\n- Confirm the Attio attributes you want to write back to (for example `Churn Reason`, `Last NPS`, `Interview Summary`) exist on the object\n- Decide your trigger: a specific list, a lifecycle-stage change, or a manual \"research this account\" action\n- Map Koji's structured question IDs to the matching Attio attributes so scores land in typed fields\n- Test with a single record end to end before enabling the automation for a whole segment\n- Put the report link somewhere prominent on the record so account owners actually see it\n\nKoji's MCP integration can even run this checklist conversationally — ask Claude to create the study, wire the fields, and dry-run one record before you flip it on.\n\n## Related Resources\n\n- [Structured Questions Guide](/docs/structured-questions-guide) — the question types that become filterable Attio attributes\n- [CRM Research Integration Guide](/docs/crm-research-integration-guide) — import participants and personalize every interview\n- [Zapier Research Automation](/docs/zapier-research-automation) — no-code recipes to trigger interviews and sync results\n- [Webhook Setup](/docs/webhook-setup): how to get interview results out of Koji today, since outbound webhooks are not available yet\n- [How AI Interviewers Work](/docs/how-ai-interviewers-work) — the technology behind Koji's follow-up questions\n- [Generating Research Reports](/docs/generating-research-reports) — turn interviews into shareable reports","category":"API Reference","lastModified":"2026-10-30T05:08:06.572123+00:00","metaTitle":"Attio + Koji Integration — Launch AI Customer Interviews from Your CRM | Koji","metaDescription":"Connect Attio to Koji to trigger AI voice and text interviews from CRM records and lists, then auto-sync themes, quotes, and structured scores back onto every contact and company. Setup via Zapier, the REST API, or MCP.","keywords":["Attio integration","Attio Koji integration","Attio customer research","AI interviews from CRM","Attio voice of customer","CRM research automation","Attio churn research","automated customer interviews","Attio API research","AI customer interviews CRM"],"aiSummary":"Guide to connecting Attio (a modern, flexible CRM) with Koji (an AI-native voice and text research platform). The bidirectional integration lets teams trigger personalized AI interviews from Attio contacts, companies, and lists, then write interview status, AI summaries, top themes, and structured question scores back onto each record. Setup options: Zapier/Make (no code), a scheduled job that polls Koji's read endpoint (Koji has no outbound webhooks yet), and Koji's 15-tool Claude MCP integration plus developer API. Koji's 6 structured question types return typed, filterable values that become Attio attributes. Example workflow: automated churn exit interviews triggered by an Attio at-risk list.","aiPrerequisites":["An Attio workspace","A Koji account (free tier works)","Optional: a Zapier or Make account for no-code setup"],"aiLearningOutcomes":["Understand what data flows between Attio and Koji","Set up the integration via Zapier, the REST API, or MCP","Trigger personalized AI interviews from Attio lists","Sync interview themes and structured scores back onto CRM records","Build an automated churn-research workflow"],"aiDifficulty":"intermediate","aiEstimatedTime":"9 minutes"},{"type":"documentation","id":"625cdd1d-ad19-4d87-bdf6-9e1abb82e6e8","slug":"mailchimp-research-integration","title":"Mailchimp + Koji: Turn Email Journeys and Audiences into AI Customer Interviews","url":"https://www.koji.so/docs/mailchimp-research-integration","summary":"Guide to connecting Mailchimp (email and SMS marketing, Customer Journey Builder) with Koji (an AI-native voice and text research platform). The bidirectional integration lets marketers trigger personalized AI interviews from Mailchimp journeys, tags, and segments (post-purchase, win-back, VIP), then write interview status, AI summaries, top themes, and structured question scores back to Mailchimp contacts as merge fields, tags, or segment membership. Setup via Zapier/Make (no code, polling Koji's read endpoint), a server job on the developer API, or Koji's 15-tool Claude MCP integration; Koji generates the personalised links and Mailchimp sends them. Koji 6 structured question types return typed values; the quality gate ensures only genuine sessions count. Three recommended journeys: post-purchase VoC, win-back exit interviews, and VIP roadmap research.","content":"**Connect Mailchimp to Koji and every Customer Journey becomes a doorway to a real conversation.** Instead of dropping a static one-question survey into a broadcast email, you trigger a Koji AI interview — voice or text — the moment a contact hits a journey point, then write the themes and structured scores back onto their Mailchimp profile so your next segment is built on *why*, not just opens and clicks.\n\nMailchimp is where marketers already orchestrate the customer relationship: welcome series, post-purchase flows, win-back campaigns, and audience segments. Koji is the AI-native research layer that turns any of those moments into a moderated interview at scale — no recruiter, no scheduling, no moderator sitting in on calls. This guide shows you exactly how the two connect and which journeys are worth wiring up first.\n\n## What data flows between Mailchimp and Koji\n\nThe integration is bidirectional:\n\n- **Mailchimp → Koji (trigger):** When a contact enters a Customer Journey, gets a tag, joins a segment, or completes an e-commerce action, Mailchimp fires an event, your automation creates a personalized Koji interview link through the API, and Mailchimp emails or texts them that link. For a one-off study to a fixed list, Koji can also send the invitation itself (see below).\n- **Koji → Mailchimp (write-back):** When the interview finishes, your automation reads the result from Koji's API and writes an interview status, a short AI summary, the top themes, and typed scores from your structured questions back onto the Mailchimp contact, as merge fields, tags, or segment membership you can immediately target.\n\nBecause Koji's outputs come from **structured questions with typed values** (an NPS scale, a single-choice churn reason, a yes/no repurchase intent), they land in Mailchimp as clean, segmentable data — not a blob of free text you have to read by hand.\n\n## Three ways to connect\n\n**1. No-code with Zapier or Make (recommended to start).** Mailchimp exposes triggers like *New Subscriber*, *New or Updated Subscriber*, and *Tag Added* in both Zapier and Make. Wire one of those to an HTTP step that calls Koji's start-interview endpoint. On the return trip, a scheduled step polls Koji's read endpoint for completed interviews (Koji has no outbound webhooks yet) and runs a Mailchimp *Add/Update Subscriber*, *Add Tag*, or *Update Merge Field* action. Most teams have a working loop in under an hour.\n\n**2. Your own server job with Mailchimp webhooks.** Mailchimp's per-audience webhooks push real-time POSTs for `subscribe`, `profile`, `unsubscribe`, and `campaign` events. Point them at a small handler that calls Koji's API to launch an interview. On the way back, a server job polls Koji's read endpoint for completed interviews (or, if the interview is embedded in your own pages, listens for the embed completion event) and calls the Mailchimp Marketing API to patch the contact's merge fields. This is the path to use for high-volume flows.\n\n**3. AI-driven with Koji's MCP and API.** Koji ships a Model Context Protocol integration (15 tools) plus a developer API, so you can drive the whole workflow from Claude or your own code: create a study, import respondents from a Mailchimp segment export, launch interviews, and pull structured results back — all programmatically.\n\n## Or let Koji send the invitations\n\nIf you only need Mailchimp as the source of the list, you can skip the send step entirely. Export the segment as a CSV, import it into **Participants** in Koji, and group those people into an audience. Then, from the study's **Self-recruit** tab, write the invitation with each person's name and your study's brand, send yourself a test, and email the audience. Every person gets their own interview link, and you can see who received it, opened it and took part. Set a reminder for the people who have not started yet.\n\nRecruitment email is included on paid plans: Insights includes 20 emails a month, Interviews 50 and Enterprise 1,000, and extra emails are €10 per 1,000 plus VAT. A simple rule: keep the invite in Mailchimp when it belongs inside a live journey (a purchase, a churn-risk tag), and use Koji's built-in email when you are interviewing a fixed list once.\n\n## The best Mailchimp journeys to add interviews to\n\nYou don't need to interview everyone. Add Koji to the three moments where email alone under-delivers:\n\n**Post-purchase voice-of-customer.** Add a journey point roughly 10–14 days after a purchase (or after the \"order delivered\" event from a connected store). A Koji interview captures not just a satisfaction rating but the hesitation they felt before buying, the alternative they almost chose, and the feature they wish existed. That is roadmap and messaging fuel a star-rating never gives you.\n\n**Win-back and lapsed-subscriber exit interviews.** Mailchimp's re-engagement and churn-risk journeys are the perfect trigger. Instead of one more discount email, ask the real question: *what changed?* Koji's AI follow-ups dig past \"too expensive\" to the actual job the product stopped doing — the difference between a guess and a fix.\n\n**VIP and repeat-buyer roadmap research.** Tag your highest-LTV segment and route them into a short Koji study that ranks the features they want next. Your most valuable customers get heard first, and you get prioritized input from the people whose retention matters most.\n\n## What syncs back to Mailchimp\n\nAfter each completed interview, your automation can read these from Koji's API and write them to the contact:\n\n- **Interview status** (e.g. `koji_status = completed`) as a merge field or tag\n- **A short AI summary** of the conversation\n- **Top themes** surfaced across the interview\n- **Structured question scores** — an NPS value, a CSAT scale point, a single-choice reason, a yes/no signal — each a typed field you can segment or trigger on\n- **A link** to the full transcript and Koji report\n\nThat means your next Mailchimp segment can be \"detractors who cited onboarding\" or \"repeat buyers who ranked integrations #1\" — audiences you simply cannot build from opens and clicks.\n\n## Why a conversation beats a one-question email survey\n\nTiming and relevance drive survey response far more than the survey tool itself — feedback requests sent close to a real experience consistently outperform generic embedded email surveys, which often languish in the low single digits to mid-teens. But response rate is only half the story. A rating tells you the score; it never tells you the reason.\n\nKoji closes that gap. Its AI interviewer asks adaptive follow-up questions in voice or text, so a \"6 out of 10\" becomes a two-minute explanation of exactly what went wrong. And it mixes **structured questions into a single conversation** (open-ended, scale, single choice, multiple choice, ranking, yes/no and price test questions), so you capture clean quantitative signals *and* the narrative behind them at once. See the [structured questions guide](/docs/structured-questions-guide) for how to combine them.\n\nKoji's quality gate keeps the data honest: each session is scored for effort and coherence, so one-word or nonsense responses are flagged, excluded, and never consume credits. Only genuine conversations flow back to your Mailchimp audience. Koji's free start includes 10 credits and no card, so you can test the loop. Interviews start as low as €1 per qualified interview, and €3 per qualified voice interview. Start with pay as you go. No subscription needed.\n\n## Mapping Mailchimp fields to Koji structured questions\n\nA little planning up front keeps the write-back clean. Decide which merge fields will hold Koji's results before you build the loop, because Mailchimp merge fields are defined per audience and referenced by their merge tag (for example `*|KOJINPS|*`).\n\nA typical mapping looks like this:\n\n- **`KOJISTAT`** (text) ← Koji interview status (`invited`, `started`, `completed`)\n- **`KOJINPS`** (number) ← the scale question score, so you can segment detractors vs promoters\n- **`KOJIREASON`** (text) ← the single_choice churn or hesitation reason\n- **`KOJITHEME`** (text) ← the top theme Koji surfaced\n- **`KOJILINK`** (text) ← a link to the full transcript and report\n\nCreate those merge fields once, then have your Zapier, Make, or server-job step patch them when it finds a completed interview. Because the scale and single_choice values arrive as typed data, you can build a live segment such as \"KOJINPS ≤ 6 AND KOJIREASON = onboarding\" and drop those contacts straight into a recovery journey.\n\n## Common setup mistakes to avoid\n\n**Interviewing on the broadcast, not the moment.** Blasting an interview link to a whole audience gets the same fatigue as a mass survey. Trigger from a specific journey point — a purchase, a churn-risk signal, a tag — so the ask always lands in context.\n\n**Forgetting the throttle.** Add a wait/delay and a suppression tag so the same contact is not invited to two interviews in the same month. A simple \"interviewed in last 90 days\" segment used as an exclusion keeps the experience respectful.\n\n**Storing free text instead of typed scores.** The whole advantage of Koji is that its structured questions return values you can segment on. Map those to number and single-choice merge fields, not one catch-all text field, and your Mailchimp audiences stay queryable.\n\nStart with one journey — post-purchase is the easiest win — prove the loop end to end, then expand to win-back and VIP research once the write-back is landing cleanly on your contacts.\n\n## Related Resources\n\n- [Structured Questions Guide](/docs/structured-questions-guide) — the question types that make interview data segmentable\n- [Klaviyo + Koji Integration](/docs/klaviyo-research-integration) — the same pattern for Klaviyo flows and segments\n- [Zapier Research Automation](/docs/zapier-research-automation) — build the no-code loop step by step\n- [Make Research Automation](/docs/make-research-automation) — the Make.com alternative\n- [Webhook Setup](/docs/webhook-setup): polling the API and the embed events\n- [Post-Purchase Survey Guide](/docs/post-purchase-survey-guide) — what to ask after a sale\n- [How the Quality Gate Works](/docs/how-the-quality-gate-works) — why low-effort responses never pollute your data","category":"API Reference","lastModified":"2026-10-30T05:07:30.237067+00:00","metaTitle":"Mailchimp + Koji Integration — AI Customer Interviews from Email Journeys | Koji","metaDescription":"Connect Mailchimp to Koji to trigger AI voice and text interviews from Customer Journeys, tags, and segments — post-purchase, win-back, VIP — and sync themes and structured scores back to Mailchimp merge fields for smarter targeting.","keywords":["Mailchimp integration","Mailchimp Koji integration","Mailchimp customer research","Customer Journey interview","email marketing voice of customer","Mailchimp win-back research","AI interviews from email journey","post-purchase interview","Mailchimp merge fields feedback","automated email research"],"aiSummary":"Guide to connecting Mailchimp (email and SMS marketing, Customer Journey Builder) with Koji (an AI-native voice and text research platform). The bidirectional integration lets marketers trigger personalized AI interviews from Mailchimp journeys, tags, and segments (post-purchase, win-back, VIP), then write interview status, AI summaries, top themes, and structured question scores back to Mailchimp contacts as merge fields, tags, or segment membership. Setup via Zapier/Make (no code, polling Koji's read endpoint), a server job on the developer API, or Koji's 15-tool Claude MCP integration; Koji generates the personalised links and Mailchimp sends them. Koji 6 structured question types return typed values; the quality gate ensures only genuine sessions count. Three recommended journeys: post-purchase VoC, win-back exit interviews, and VIP roadmap research.","aiPrerequisites":["A Mailchimp account","A Koji account (free tier works)","Optional: a Zapier or Make account for no-code setup"],"aiLearningOutcomes":["Understand what data flows between Mailchimp and Koji","Trigger AI interviews from Mailchimp journeys, tags, and segments","Sync themes and structured scores back to Mailchimp merge fields","Set up post-purchase, win-back, and VIP research journeys","Know why conversational interviews beat one-question email surveys"],"aiDifficulty":"intermediate","aiEstimatedTime":"9 minutes"},{"type":"blog","id":"b72db351-6c23-4a1f-9f3d-819159731fb4","slug":"koji-vs-usertesting-2026","title":"Koji vs UserTesting: AI-Powered vs Panel-Based Research (2026)","url":"https://www.koji.so/blog/koji-vs-usertesting-2026","summary":"Koji is an AI-native research platform: an AI interviewer runs voice and text interviews in 30+ languages, follows up on vague answers and generates a themed report in hours. It starts free with 10 credits. Insights is €29/month, Interviews €79/month (3 seats), Enterprise from €500/month, with built-in panel recruitment across 130+ countries on paid plans. UserTesting is a quote-only platform for recorded, task-based usability sessions with its Contributor Network, AI Insight Summary and Friction Detection. Vendr data puts the median contract at about $40,000 a year. Choose Koji for discovery and customer interviews at scale, UserTesting for video evidence and task-based usability testing.","content":"Choosing between Koji and UserTesting comes down to one question: do you need recordings of people using your product, or do you need to understand what customers think and why, quickly and at a price you can repeat every month?\n\nUserTesting is the long-standing leader in recorded, panel-based usability research. Koji is an AI-native research platform where an AI interviewer holds voice or text conversations with participants, asks follow-up questions, and writes the analysis for you. Both now recruit participants for you, so the real differences are method, speed and cost.\n\n## Quick Comparison\n\n| Factor | Koji | UserTesting |\n| --- | --- | --- |\n| Research method | AI-moderated interviews (voice and text) | Recorded sessions of people completing tasks |\n| Turnaround | Hours | Days |\n| Pricing | Free to start. Insights €29/mo, Interviews €79/mo, Enterprise from €500/mo | Quote only, median contract about $40,000/yr (Vendr) |\n| Cost per interview | 1 credit (text) or 3 credits (voice), €1 per extra credit | Bundled into annual credit packages |\n| Seats | Insights 1, Interviews 3 (+€15/mo each), Enterprise 10 | Set in the contract |\n| Participant recruitment | Built in on paid plans: describe the audience, see the cost before launch, 130+ countries. Or use your own list, link or website embed | UserTesting Contributor Network plus custom and private networks |\n| Screening | Screening questions run before every interview, on every plan | Screener questions per test |\n| Languages | 28, chosen by each participant | Depends on panel and market |\n| AI analysis | Full themed report generated automatically | AI Insight Summary and Friction Detection |\n| Video or screen recording | No | Yes |\n| Task-based usability testing | No | Yes |\n| Free plan | Yes: 10 credits, no card, one active study | No, trial only |\n| Best for | Discovery and customer interviews at scale | Usability testing with video evidence |\n\n## What Is UserTesting?\n\nUserTesting, founded in 2007, pioneered on-demand access to human testers who complete recorded research sessions. Participants come from the UserTesting Contributor Network, or from custom networks you set up. They record their screen and voice while they try tasks, answer questions or react to concepts, thinking aloud as they go. Researchers watch the videos, cut highlight reels and share them across the organisation.\n\nUserTesting has added AI on top of that workflow. Its AI Insight Summary condenses the findings from tasks and verbal responses, and Friction Detection, launched in April 2023, uses machine learning to flag moments of excessive clicking or scrolling across sessions ([UserTesting](https://www.usertesting.com/company/newsroom/press-releases/usertesting-launches-new-behavioral-analytics-capability-friction)).\n\n**UserTesting works best for:**\n\n- Watching real people use your product on video\n- Task-based usability testing with think-aloud protocols\n- Video evidence to persuade sceptical stakeholders\n- Enterprise teams with dedicated researchers and large budgets\n\n## What Is Koji?\n\nKoji is an AI-native research platform. You describe your research goals and Koji helps you write the study, including its questions and [screening questions](/docs/research-screener-questions). Participants open a link, pick text or voice and their language, and have a natural conversation with an AI interviewer that probes when an answer is vague. When interviews finish, Koji themes every conversation and produces a report you can publish and share by link.\n\nKoji also finds participants for you. Open the Panel recruit tab, describe who you want to hear from and see the cost before you launch. Credits are set aside at launch and anything unused goes back to your balance. Panel recruitment is available on paid plans and reaches 130+ countries ([panel recruitment](/docs/panel-recruitment)). Recruited participants land in the same study and report as people you invite yourself.\n\nBefore you launch, the Preview tab lets you take your own interview by text or voice. Previews are free: they never use credits and never appear in your report.\n\n**Koji works best for:**\n\n- Customer discovery and product research interviews\n- Running dozens or hundreds of interviews at once\n- Teams without a dedicated researcher\n- Multi-market studies, with 30+ languages in one report\n- Continuous research that runs every week, not once a quarter\n\n## The Cost Gap\n\nUserTesting does not publish prices. Every plan is quoted by sales and priced by annual credit volume, plan tier and participant sourcing. Vendr's buyer data puts annual contracts between roughly **$12,000 and $113,642**, with a **median of about $40,000 a year** ([Vendr](https://www.vendr.com/marketplace/usertesting)). There is no free plan, only a trial.\n\nKoji publishes its pricing. You can start free with 10 credits and no card. Insights costs €29 a month with 29 credits, Interviews €79 a month with 79 credits and 3 seats, and Enterprise starts from €500 a month. A text interview uses 1 credit and a voice interview 3 credits, extra credits are €1 on every plan, and low-quality conversations are never charged. Panel participants are paid for in credits, with the price shown before you launch.\n\nFor a team that mainly uses UserTesting for interviews and concept feedback, the gap is large. On Interviews at €79 a month, a year of Koji costs under €1,000 before any extra credits. Even a heavy year of panel recruitment usually leaves the total far below a typical UserTesting contract.\n\n## Speed: Hours vs Days\n\nA UserTesting study involves writing a screener, waiting for participants to complete sessions, and then watching the recordings before findings emerge. The AI summary helps, but someone still reviews the video.\n\nKoji interviews run in parallel. Fifty participants can finish on the same afternoon, and their conversations are analysed together, so a themed report is ready in hours. Because the interviewer asks follow-up questions itself, you get the reasons behind answers without scheduling calls.\n\nThe wider field is moving the same way. In Qualtrics' 2025 market research trends study, **89% of researchers** said they were already using AI tools regularly or experimentally, and **83%** said their organisations planned to significantly increase AI investment ([Research Live](https://www.research-live.com/article/news/researchers-predict-synthetic-responses-will-dominate-in-coming-years/id/5131923)).\n\n## Where UserTesting Has the Edge\n\n### Video evidence\n\nNothing moves a sceptical executive like watching a real person fail at checkout. UserTesting's recordings and highlight reels are a uniquely persuasive format. Koji produces transcripts, quotes, themes and charts, not video.\n\n### Task-based usability testing\n\nUserTesting was built for task completion and think-aloud protocols on live sites and prototypes. If your question is \"can people complete this flow?\", UserTesting or a lighter tool such as Maze is the better fit. You also may not need many sessions: Nielsen and Landauer's model suggests about five users uncover roughly 85% of the problems that affect a large share of users on a task ([MeasuringU](https://measuringu.com/five-user-85/)).\n\n### Enterprise maturity\n\nUserTesting has nearly two decades of enterprise procurement, security reviews and governance behind it. Koji's Enterprise plan covers SSO, data residency and [Bring Your Own Key](/docs/bring-your-own-key) for teams that need them, but some procurement teams prefer a long track record.\n\n## When to Use Each Tool\n\n**Choose Koji when:**\n\n- You need insight this week, not next month\n- You are running discovery, concept or customer interviews\n- You want published pricing and a free start\n- You do not have a dedicated research team\n- You want to run research continuously, in several languages\n\n**Choose UserTesting when:**\n\n- You need video of real people using your product\n- You are running task-based usability tests\n- Stakeholders need visual evidence\n- You have an enterprise research budget\n\n**Use both:**\nMany teams run Koji for high-volume discovery and concept work, and keep a small usability budget for the handful of recorded sessions that need video. Our guide to [AI usability testing](/docs/ai-usability-testing-guide) shows how the two fit together, and [voice vs text interviews](/docs/voice-vs-text-interviews) helps you pick the right mode in Koji.\n\n## Final Verdict\n\n**UserTesting** is a mature platform for recorded usability research. If you need video evidence, task-based testing and enterprise governance, and have a budget in the tens of thousands a year, it delivers.\n\n**Koji** is the better choice for most product teams that need more customer conversations, not more recordings. It is priced in the open, starts free, recruits participants for you on paid plans, runs in 30+ languages and turns every interview into a report within hours.\n\nIf you are comparing these tools because you want to talk to more customers, [start your first study free](https://www.koji.so) with 10 credits and no card.\n\n_Pricing data from Vendr buyer benchmarks and Koji's pricing page. Last verified: October 2026._\n\n## Frequently Asked Questions\n\n**Q: Is Koji a UserTesting alternative?**\nA: For interview-based research such as discovery, concept feedback and churn interviews, yes. Koji is faster and far cheaper. For task-based usability testing with screen recordings, UserTesting still has the edge.\n\n**Q: How much does UserTesting cost compared to Koji?**\nA: UserTesting is quote only. Vendr's buyer data shows a median contract of about $40,000 a year. Koji starts free with 10 credits. Insights is €29 a month and Interviews €79 a month, with extra credits at €1.\n\n**Q: Can Koji recruit participants like UserTesting?**\nA: Yes. On paid plans, describe your audience on the Panel recruit tab, see the cost before launch, and Koji recruits across 130+ countries. You can also invite your own customers by link, list or website embed.\n\n**Q: Does Koji record video?**\nA: No. Participants talk to an AI interviewer by voice or reply by text. There are no screen or camera recordings.\n\n**Q: Which tool is better for early-stage startups?**\nA: Koji. It has a free plan, published pricing, needs no research expertise and returns findings in hours. UserTesting's annual contracts are usually out of reach for early-stage teams.\n","category":"Research","lastModified":"2026-10-29T15:33:39.910337+00:00","metaTitle":"Koji vs UserTesting: AI vs Panel Research (2026)","metaDescription":"Koji vs UserTesting compared: AI-moderated interviews vs recorded sessions. UserTesting median about $40k/yr (Vendr), Koji from free, €29 or €79 a month.","keywords":["koji vs usertesting","usertesting alternative","usertesting pricing","ai moderated interviews","usability testing","user research platform 2026"],"aiSummary":"Koji is an AI-native research platform: an AI interviewer runs voice and text interviews in 30+ languages, follows up on vague answers and generates a themed report in hours. It starts free with 10 credits. Insights is €29/month, Interviews €79/month (3 seats), Enterprise from €500/month, with built-in panel recruitment across 130+ countries on paid plans. UserTesting is a quote-only platform for recorded, task-based usability sessions with its Contributor Network, AI Insight Summary and Friction Detection. Vendr data puts the median contract at about $40,000 a year. Choose Koji for discovery and customer interviews at scale, UserTesting for video evidence and task-based usability testing.","aiKeywords":["koji vs usertesting","usertesting alternative","usertesting pricing","ai interviews","panel recruitment"],"aiContentType":"comparison","faqItems":[{"answer":"For interview-based research such as discovery, concept feedback and churn interviews, yes. Koji is faster and far cheaper. For task-based usability testing with screen recordings, UserTesting still has the edge.","question":"Is Koji a UserTesting alternative?"},{"answer":"UserTesting is quote only. Vendr's buyer data shows a median contract of about $40,000 a year. Koji starts free with 10 credits. Insights is €29 a month and Interviews €79 a month, with extra credits at €1.","question":"How much does UserTesting cost compared to Koji?"},{"answer":"Yes. On paid plans, describe your audience on the Panel recruit tab, see the cost before launch, and Koji recruits across 130+ countries. You can also invite your own customers by link, list or website embed.","question":"Can Koji recruit participants like UserTesting?"},{"answer":"No. Participants talk to an AI interviewer by voice or reply by text. There are no screen or camera recordings.","question":"Does Koji record video?"},{"answer":"Koji. It has a free plan, published pricing, needs no research expertise and returns findings in hours. UserTesting's annual contracts are usually out of reach for early-stage teams.","question":"Which tool is better for early-stage startups?"}],"relatedTopics":["usertesting alternatives","ai moderated interviews","usability testing","user research pricing"]},{"type":"blog","id":"da00b952-220f-4723-a7c2-d38df5e190d9","slug":"ai-moderated-vs-human-moderated-interviews","title":"AI-Moderated vs Human-Moderated Interviews: Which Should You Choose?","url":"https://www.koji.so/blog/ai-moderated-vs-human-moderated-interviews","summary":"AI-moderated interviews deliver equivalent research quality to human-moderated interviews for most exploratory research questions, while offering dramatically lower cost, zero scheduling overhead, and automatic analysis. Human moderation remains superior for emotionally sensitive topics, highly exploratory research, and complex task-based usability studies. Most high-performing research teams use AI moderation for continuous discovery and human moderation for periodic deep strategic research.","content":"\nAI-moderated interviews and human-moderated interviews each have a time and a place. The question is not \"which is better\" — it is \"which is right for your specific research question and constraints.\"\n\n**The direct answer:** For most exploratory research — why do users churn, what problems do they face, how do they make decisions — AI-moderated interviews deliver equivalent insight quality with dramatically lower cost and time investment. For research requiring deep empathetic probing, culturally sensitive topics, or complex task observation, human moderators still add irreplaceable value.\n\n## The Core Trade-off\n\nHuman-moderated interviews offer maximum flexibility and empathy. A skilled moderator can follow a surprising thread deep into territory the discussion guide never anticipated, sense when a participant is holding back, and adapt tone for sensitive subjects.\n\nAI-moderated interviews offer unlimited scale, zero scheduling overhead, and consistent application of the same protocol across every participant. The AI never has a bad day, never leads witnesses, and can run 1,000 conversations simultaneously across every timezone.\n\nAccording to the Lyssna Research Synthesis Report 2025 (surveying 300 research practitioners), the number one frustration with research workflows is time — 60.3% of researchers cite manual, time-consuming work as their biggest challenge. AI moderation directly addresses this at every stage of the research cycle.\n\n## Quick Comparison\n\n| Dimension | Human-Moderated | AI-Moderated (Koji) |\n|-----------|-----------------|---------------------|\n| Time to schedule | Days to weeks | None (async) |\n| Cost per interview | $50–$400+ | Fraction of human cost |\n| Simultaneous sessions | 1 or a few | Unlimited |\n| Consistency across sessions | Varies by moderator | 100% consistent |\n| Intelligent follow-up probing | Deep and adaptive | Contextual and adaptive |\n| Sensitive or emotional topics | High empathy possible | May miss emotional depth |\n| Moderator bias risk | Present, even in good moderators | Eliminated |\n| Participant availability | Requires calendar coordination | Anytime, any timezone |\n| Transcription | Manual or third-party tool | Automatic |\n| Analysis | Manual or tool-assisted | Automatic themes and sentiment |\n| Time from question to insight | Days to weeks | Hours |\n\n## When AI-Moderated Interviews Win\n\n### Research at Scale\n\nThe most immediate advantage of AI moderation is scale. Running 50 user interviews would require months of scheduling and moderation work from a human researcher. With AI moderation, those 50 interviews can be completed in 48–72 hours.\n\nAccording to User Interviews State of User Research 2025, 70% of large software companies cite time and bandwidth as their top research challenges. Teams using AI-native platforms like Koji can conduct 10x more research with the same headcount — or run continuous discovery without a dedicated researcher.\n\nThe Maze Future of User Research Report 2026 found that 66% of research teams reported increased research demand in 2026, but most organizations had not grown headcount to match. AI moderation is how teams bridge that gap without budget increases.\n\n### Eliminating Moderator Bias\n\nEven skilled moderators introduce bias. Research consistently shows participants respond differently based on perceived expectations — picking up on the interviewer's micro-expressions, intonation, and word choice. A moderator who is enthusiastic about a feature unconsciously signals that enthusiasm. A moderator who is skeptical telegraphs that too.\n\nAI moderation eliminates this variable. Every participant hears the same questions with the same neutral framing. This is particularly important in concept testing and competitive research, where you want uncontaminated first impressions from participants who have not been influenced by the researcher.\n\n### Speed to Insight\n\nResearch is increasingly happening on the timeline of business decisions. When a product review is happening on Thursday and you need customer perspective today, a human-moderated study is not an option.\n\nAccording to the Maze Future of User Research Report 2026, 22% of organizations now use research at every level of business strategy — up from just 8% a year prior. That shift requires tools that deliver insights in hours, not weeks. AI moderation makes same-week research the norm rather than the exception.\n\n### Enabling Continuous Discovery\n\nContinuous product discovery — running small research studies on an ongoing basis — requires a different operational model than traditional research. According to User Interviews State of User Research 2024, only 44% of companies were running continuous research. The biggest barrier is not willingness — it is operational overhead.\n\nAI moderation removes that overhead. Studies can run continuously in the background, collecting new participant responses as users encounter your product. The approach that was previously reserved for well-resourced research teams becomes accessible to any product team.\n\n### No Research Expertise Required\n\nAI-moderated platforms like Koji include AI consultants that guide any team member through study design and question writing. A product manager, founder, or customer success lead can run rigorous research without needing a research methodology background. This democratizes insight-gathering across the entire organization.\n\n## When Human-Moderated Interviews Win\n\n### Emotionally Sensitive Topics\n\nAI moderators excel at structured and semi-structured research. But for research involving significant emotional weight — bereavement, health conditions, trauma, or complex social situations — human empathy is essential. A skilled moderator knows when to pause, when to offer reassurance, and when a participant is struggling.\n\nFor research in these categories, human moderation is not optional — it is the ethical standard.\n\n### Highly Exploratory Research in Uncharted Territory\n\nIf your research question is genuinely wide open and you do not know what you do not know, a human moderator ability to follow unexpected threads can be uniquely valuable. When a participant says something that no one anticipated, an expert moderator can spend 20 minutes on a thread that was never in the discussion guide.\n\nAI moderation is best when you have a reasonably clear hypothesis to explore at scale. It is less suited to the kind of open anthropological exploration where the researcher instinct steers the entire conversation in real time.\n\n### Complex Task-Based Usability Research\n\nIf you need to observe a participant attempting a multi-step task — navigating a workflow, completing a checkout process, using a new interface — human moderation allows for real-time intervention, probing confusion as it happens, and observing behavior that goes beyond verbal responses.\n\nFor this type of research, tools like Lookback or UserTesting with screen recording are better choices than conversational interview platforms.\n\n### Deep Domain Expert Interviews\n\nWhen researching complex technical subjects with expert practitioners, a human interviewer with domain expertise can build credibility quickly and push past surface-level answers. A researcher who understands the field can challenge an assumption or ask a second-order question in a way that reveals deeper insight than a generalist AI moderator might reach.\n\n## The Cost Comparison\n\nTraditional human-moderated research typically costs:\n- Participant recruitment: $40–$100 per participant\n- Moderator time: 1–2 hours per interview at $80–$200/hour\n- Transcription: $1–$3 per minute of audio\n- Analysis and synthesis: 4–8 hours per study\n\nA 10-participant study can easily run $3,000–$8,000 in direct costs, plus 20–40 hours of researcher time for analysis. At that price, most teams can only run research occasionally — which means major decisions happen without customer input.\n\nAI-moderated research with Koji costs a fraction of that and delivers the analysis automatically. The same 10-participant study can be conducted and analyzed for as low as €1 per qualified interview, and €3 per qualified voice interview, in a fraction of the time. Start free with 10 credits, then pay as you go. No subscription needed.\n\nResearch from InsightMark Research (2025) shows the AI-based research services market growing at 16.1% CAGR — driven precisely by this cost and time efficiency advantage.\n\n## How AI Moderation Works in Practice\n\nHere is the Koji workflow, from question to insight:\n\n**Step 1: Design your study.** Work with the AI consultant to define your research objective, choose a methodology, and build your conversation guide. The AI helps you write questions that generate actionable answers — not just interesting conversation.\n\n**Step 2: Share the interview link.** Distribute it via email, in-app messaging, Intercom, or any channel you use to reach participants. If you do not have an audience of your own, Koji recruits the participants for you: describe who you need and approve a live per-respondent quote in credits before anything launches. There is no scheduling required. Participants take the interview at a time that works for them.\n\n**Step 3: AI conducts the conversations.** Each participant has a one-on-one conversation with Koji AI interviewer. It follows the discussion guide, asks contextual follow-ups based on what the participant says, and keeps conversations focused on the research objective. Voice and text modes are both supported.\n\n**Step 4: Review and share insights.** After interviews complete, Koji automatically surfaces themes, analyzes sentiment across all conversations, and generates a research report. Share it with stakeholders in one click.\n\nThe full cycle from \"we have a research question\" to \"we have findings\" can take less than 48 hours for a 20-participant study. The equivalent human-moderated study typically takes 2–4 weeks.\n\n## A Practical Decision Framework\n\nUse this to guide your moderation approach for any research project:\n\n| Research goal | Moderation approach |\n|--------------|-------------------|\n| Understand why users churn | ✅ AI-moderated |\n| Validate a product hypothesis with 50 users | ✅ AI-moderated |\n| Run continuous discovery on a rolling basis | ✅ AI-moderated |\n| Explore decision-making for a purchase | ✅ AI-moderated |\n| Research involving grief, illness, or trauma | Human-moderated |\n| Observe users navigating a specific UI | Human-moderated (or screen recording tool) |\n| Deep expert interviews in niche technical domains | Human-moderated |\n| Understand general user frustrations at scale | ✅ AI-moderated |\n\n## Our Recommendation\n\nFor the majority of user research questions facing product teams in 2026 — understanding churn, validating ideas, exploring user needs, gathering feedback on features — AI-moderated interviews are the right choice. You get more interviews, faster, at lower cost, with automatic analysis.\n\nReserve human moderation for research involving sensitive topics, truly open exploratory work, or complex task-based usability testing where observation of behavior is more valuable than verbal responses.\n\nThe two approaches are not mutually exclusive. The highest-performing research practices use AI moderation for continuous discovery and rapid hypothesis testing, and bring in human moderators periodically for strategic research requiring deeper empathy or specialized domain expertise.\n\n## Frequently Asked Questions\n\n**Are AI-moderated interviews as rigorous as human-moderated?**\nFor most structured and semi-structured research, yes. AI moderation delivers consistent, unbiased questioning and the scale to run far more interviews than a human team could manage. A 50-participant AI-moderated study often surfaces more robust patterns than a 10-participant human-moderated study because of the larger sample — and both cost roughly the same with modern AI tools.\n\n**Can AI interviewers ask follow-up questions?**\nYes. Koji AI interviewer adapts to participant responses and asks contextual follow-up questions based on what each participant says. If a participant mentions an unexpected problem or use case, the AI explores it. The experience is a genuine conversation, not a scripted survey.\n\n**How many interviews do I need for AI-moderated research?**\nThe same thematic saturation principles apply: 5–10 interviews per distinct user segment typically surfaces the majority of key themes for exploratory research. Because AI moderation makes running more interviews easy and affordable, teams often run 20–50 interviews where they would have previously run 5–8 — and they get significantly more robust findings as a result.\n\n**Does AI moderation work for B2B research?**\nYes. B2B research with Koji works well for understanding enterprise decision-making processes, evaluating product-market fit, understanding purchasing criteria, and gathering feedback on specific product capabilities. The async format works particularly well for busy buyers and practitioners who cannot commit to scheduled calls.\n\n**Is AI-moderated research transparent to participants?**\nYes. Participants are informed they are speaking with an AI interviewer before beginning the session. Informed consent and privacy standards apply exactly as they do with human-moderated research. Koji is GDPR-compliant and participants can exit the study at any time.\n  ","category":"Research","lastModified":"2026-10-29T15:33:38.85162+00:00","metaTitle":"AI vs Human-Moderated Interviews: Which to Choose? — Koji Blog","metaDescription":"AI-moderated interviews deliver equivalent insight quality at 10x the scale and a fraction of the cost. Here is when to use each approach for your user research.","keywords":["ai moderated interviews","human moderated interviews","ai vs human interviewer","automated user interviews","AI interview tool","qualitative research moderation","user interview ai"],"aiSummary":"AI-moderated interviews deliver equivalent research quality to human-moderated interviews for most exploratory research questions, while offering dramatically lower cost, zero scheduling overhead, and automatic analysis. Human moderation remains superior for emotionally sensitive topics, highly exploratory research, and complex task-based usability studies. Most high-performing research teams use AI moderation for continuous discovery and human moderation for periodic deep strategic research.","aiKeywords":["AI-moderated interviews","human-moderated interviews","AI vs human research","automated interviews","user interview scale","qualitative research methods","continuous discovery","moderator bias"],"aiContentType":"comparison","faqItems":[{"answer":"For most structured and semi-structured research, yes. AI moderation delivers consistent, unbiased questioning at a scale human teams cannot match. A 50-participant AI-moderated study often surfaces more robust patterns than a 10-participant human study at the same cost.","question":"Are AI-moderated interviews as rigorous as human-moderated?"},{"answer":"Yes. Koji AI interviewer adapts to participant responses and asks contextual follow-ups based on what each participant says. If a participant mentions an unexpected problem, the AI explores it. The experience is a genuine conversation, not a scripted survey.","question":"Can AI interviewers ask follow-up questions?"},{"answer":"5 to 10 interviews per distinct user segment typically surfaces the majority of key themes for exploratory research. Because AI moderation is affordable and fast, teams often run 20 to 50 interviews and get significantly more robust findings.","question":"How many interviews do I need for AI-moderated research?"},{"answer":"Yes. B2B research with Koji works well for understanding enterprise decision-making, product-market fit, purchasing criteria, and feature feedback. The async format works particularly well for busy buyers who cannot commit to scheduled calls.","question":"Does AI moderation work for B2B research?"},{"answer":"Yes. Participants are informed they are speaking with an AI interviewer before beginning. Informed consent and privacy standards apply exactly as they do with human-moderated research. Koji is GDPR-compliant and participants can exit at any time.","question":"Is AI-moderated research transparent to participants?"}],"relatedTopics":["User Research Methods","Qualitative Research","AI Research","Interview Techniques","Continuous Discovery","Research Operations"]},{"type":"blog","id":"f51d3355-9a02-4305-aa53-60f34a220b4c","slug":"how-founders-validate-product-ideas-with-customer-interviews","title":"How Founders Validate Product Ideas with Customer Interviews (2026)","url":"https://www.koji.so/blog/how-founders-validate-product-ideas-with-customer-interviews","summary":"This guide explains how founders can validate product ideas using customer interviews before writing any code. It covers the core failure modes (pitching instead of exploring), a 5-stage interview playbook based on YC methodology, how many interviews are actually needed (20-30 minimum), synthesis techniques, and how AI-powered tools like Koji make the entire process 10x faster for resource-constrained founding teams.","content":"42% of startups fail because they built something nobody wanted.\n\nNot because they ran out of money. Not because the team was not good enough. Because they made a product customers did not care about — and they did not find out until it was too late.\n\nThat statistic comes from CB Insights' analysis of 101 startup post-mortems. \"No market need\" was the top cause of failure — nearly twice as common as the second cause (running out of cash), which is itself often downstream of the same problem.\n\nThe antidote is customer interviews. But most founders run them wrong, or skip them entirely. This guide covers how to validate your idea with customer conversations that actually tell you whether your idea has legs.\n\n## Why Founder Validation Interviews Usually Fail\n\nThe problem is not that founders skip talking to potential customers. It is that the conversations they have are structured to confirm what they already believe.\n\nRob Fitzpatrick identified the core failure mode in *The Mom Test* (2013): founders pitch their idea and ask \"what do you think?\" This almost always generates polite, misleading enthusiasm. The person you are interviewing does not want to crush your dream.\n\nGood validation interviews do not ask about your idea at first. They ask about the person's actual behavior, struggles, and existing solutions. The data you want lives in their past, not their hypothetical future.\n\nY Combinator has been teaching this principle for 20 years. Eric Migicovsky's canonical YC guide, *How to Talk to Users*, identifies five questions that unlock real validation data:\n\n1. What is the hardest part of doing [the thing you are targeting]?\n2. Tell me about the last time you encountered that problem.\n3. Why was that hard?\n4. What did you do to try to fix it?\n5. What did you not like about the solutions you tried?\n\nThese questions surface real pain, real workarounds, and real willingness to switch — none of which you can fake.\n\n## How Many Interviews Do You Actually Need?\n\nMost founders stop at five to seven conversations and declare their idea validated. This is almost always too few to identify reliable patterns.\n\nThe most cited benchmarks from credible sources:\n\n- **Rob Fitzpatrick** (*The Mom Test*) recommends a minimum of 20–30 structured discovery interviews before drawing conclusions about your target audience's core problem.\n- **Steve Blank**, who created the Customer Development methodology, recommends 50–100 conversations across the discovery and validation phases.\n- **Cindy Alvarez** (*Lean Customer Development*) suggests starting with 5 interviews to sharpen your hypotheses, then doing 5–10 more per iteration.\n- **Teresa Torres** (*Continuous Discovery Habits*, 2021) recommends at minimum one customer interview per week as an ongoing practice — not a one-time sprint.\n\nA useful practical benchmark: **keep interviewing until you stop hearing new things**. When three interviews in a row surface no new problems or insights, you have likely reached thematic saturation for that hypothesis.\n\n## The 5-Stage Founder Interview Playbook\n\n### Stage 1: Define Your Hypothesis First\n\nBefore booking a single call, write down your core assumption in one sentence:\n\n*\"I believe [target customer] struggles with [specific problem] because [reason], and they would pay to solve it.\"*\n\nThis hypothesis is what your interviews will test. Without it, you will gather interesting anecdotes but no actionable signal. Most founder validation fails not during the interviews — it fails before them, because there was no clear hypothesis to test.\n\n### Stage 2: Find the Right People\n\nThe single biggest validation mistake: talking to friends, family, or anyone who agrees to take your call.\n\nFor your interviews to be valid, you need to talk to people who:\n- **Actually have the problem** you think you are solving (not just people in your target demographic)\n- **Have tried to solve it** — people who have struggled enough to look for solutions are your most informative source\n- **Are not emotionally invested in your success**\n\nWhere to find them: LinkedIn outreach to relevant job titles, niche Slack communities and subreddits, industry Facebook groups, and warm intros from advisors. Or let Koji find them: open Panel recruit, describe the person who has the problem, add a screening question so only people who have actually tried to solve it get through, and Koji finds them across 130+ countries at a price per completed interview in credits you see before you launch, on paid plans. Anyone you reach yourself gets a personalised link from a CSV import or the generic study link, and everyone lands in the same study.\n\nCold outreach for interviews typically gets 2–10% response rates (User Interviews platform data, 2023). According to the Maze *State of User Research 2024*, 61% of researchers struggle with finding the right participants. Recruiting is genuinely hard — but talking to the wrong people is worse than not talking at all.\n\n### Stage 3: Run the Interview\n\n**Duration:** 30–45 minutes is the sweet spot. Longer risks fatigue; shorter rarely surfaces the real insight.\n\n**Opening frame:** Set context without biasing the participant. \"I am working on a problem in [space] and trying to understand how people currently deal with it. I am not selling anything — I just want to understand your experience.\"\n\n**Structure:**\n1. **Warm-up:** What is their role? What does their day look like?\n2. **Problem exploration:** Use the 5 YC questions above, focused on your hypothesis area\n3. **Solution exploration:** What have they tried? Why did it not work?\n4. **Concept reaction (optional):** Only after fully exploring the problem space — share your idea and observe their reaction without asking \"would you use this?\"\n\n**What not to do:** Lead with your solution. Ask hypothetical \"would you\" questions. Pitch. Interpret silence as agreement. All four are documented failure modes.\n\n### Stage 4: Synthesize Across Interviews\n\nThis is where most founder research dies. After a round of interviews, you are sitting on hours of notes with no clear pattern.\n\nA simple synthesis process:\n1. Write one insight per note (physical or digital in Miro/FigJam)\n2. Cluster similar insights together\n3. Name each cluster: \"The [Problem] Problem\"\n4. Rank clusters by frequency (how many interviews mentioned this?) and intensity (how much pain did people express?)\n5. Your top 2–3 clusters are your validated opportunity space\n\nTraditional manual synthesis of 10 interviews takes 10–20 hours of focused work. This is where AI research tools can dramatically accelerate the process. Koji's automatic thematic analysis collapses this to minutes across any number of interviews — so the synthesis bottleneck stops being the reason founders skip this step.\n\n### Stage 5: Decide What the Data Tells You\n\nThe output of a validation round is not a feature list. It is a clearer hypothesis — confirmed, invalidated, or refined.\n\n**If validated:** You heard the same core problem, with real emotional intensity, from the majority of your interviewees. Proceed to solution exploration.\n\n**If invalidated:** The problem does not exist at the scale you thought, is not as painful as assumed, or belongs to a different customer segment. Pivot your hypothesis before spending a day on product.\n\n**If unclear:** You heard conflicting signals. Segment your interviews by role, company size, or industry — you may have been talking to two different customer types with different problems.\n\n## The Real Cost of Skipping Validation\n\nA full traditional customer research cycle — recruiting, scheduling, interviewing, transcribing, and synthesizing — costs $3,000–$8,000 for 10 interviews and takes 3–8 weeks for a lean team without a dedicated researcher.\n\nThat is a real barrier for early-stage founders simultaneously doing sales, hiring, and product work. The Maze 2024 survey found that 65% of product teams do research less than once a month, with time as the top reason.\n\nBut consider the alternative cost. The Startup Genome Project found that 74% of high-growth internet startups fail due to premature scaling — defined as scaling before validating what customers actually want. That is months of engineering time, investor capital, and team energy spent on something that could have been caught in a 30-minute conversation.\n\nIn a 2024 survey by Lenny Rachitsky, 78% of product professionals said they wished they had done more customer interviews in the early stages of their product. The most common regret: \"We built the wrong thing for 6–12 months because we assumed instead of asked.\"\n\n## How AI Changes the Equation for Founders\n\nAI-native research platforms like Koji change what is operationally possible for early-stage teams. Instead of scheduling 20 individual calls, you configure your research study once and let the AI interviewer conduct all conversations asynchronously. The AI analyzes the full dataset and surfaces themes automatically.\n\nWhat used to take 6 weeks can take 48 hours. This is not a hypothetical — it is the core design principle behind tools built for exactly this workflow.\n\n**The important caveat:** AI interviews excel at hypothesis testing and pattern detection at scale. For the earliest, most exploratory discovery — when you do not yet know what questions to ask — a handful of live human conversations is still the best starting point. The ideal founder research stack combines a few exploratory human conversations early on with AI-powered interview rounds for validation and iteration.\n\n## What Good Validation Looks Like\n\n**Dropbox (2007):** Drew Houston validated demand for a product that did not exist yet by posting a three-minute explainer video to Hacker News. Overnight, 75,000 people signed up for early access. Not a line of production code written. Full demand validation.\n\n**Y Combinator standard:** By Demo Day, YC expects founders to have had hundreds of customer conversations — not prototype tests, but conversations about the problem. Paul Graham pushed early-batch founders to talk to users every single day.\n\n**Michael Seibel (YC CEO)** has publicly stated that the most common mistake he sees in YC companies is founders who are \"too embarrassed to talk to users\" or who interpret one positive conversation as validation.\n\nThe research is unambiguous. Companies that practice regular customer research are 60% more likely to exceed their revenue goals, according to Pragmatic Institute's annual Product Management Survey (1,000+ product managers surveyed). That gap is not explained by luck — it is explained by decisions that were grounded in real customer signal instead of internal assumption.\n\n## Getting Started With Koji for Validation\n\nKoji is built for exactly this workflow. You configure your research brief — the hypothesis, the target audience, the questions — and Koji's AI interviewer handles the rest. Participants complete voice or text interviews on their own schedule. You get thematic analysis across all responses, surfaced automatically.\n\nFor founders who need to validate fast without a research team, this changes what is possible. You can run a 20-interview validation round in the time it used to take to schedule five Zoom calls.\n\n[Start your first validation study on Koji →](/docs/getting-started)\n\n## Frequently Asked Questions\n\n**How many customer interviews do I need to validate a startup idea?**\nMost practitioners recommend 20–30 structured discovery interviews as the minimum to identify reliable patterns for a specific hypothesis. Start with 5–10 to refine your hypothesis, then run another round to validate it. Keep going until you stop hearing new problems — that is thematic saturation.\n\n**What is the biggest mistake founders make in customer interviews?**\nPitching their idea instead of exploring the problem. Founders who ask \"would you use this?\" receive polite, unreliable answers. The right approach is to ask about the person's past behavior, existing struggles, and what they have already tried — not their hypothetical future intentions.\n\n**Should I use AI interviews or live video calls for validation?**\nBoth have a place. Live exploratory conversations are best at the very beginning, when you do not yet know what questions to ask. AI-moderated interviews are more efficient once you have a defined hypothesis to test — you can run dozens of conversations in the time it takes to schedule one Zoom call.\n\n**How do I find people to interview for product validation?**\nTwo routes. If you do not have the people yet, Koji's Panel recruit finds them: describe the audience, add a screening question for people who actually have the problem, and see the price per completed interview in credits before you launch. If you are recruiting yourself: LinkedIn outreach to your target job titles, niche Slack communities, industry subreddits, and warm intros from advisors. Cold outreach typically gets 2–10% response rates, so plan for volume. Avoid friends and family: social pressure distorts the feedback you will receive.\n\n**What should I do if validation interviews show my idea will not work?**\nThis is a success, not a failure. You have avoided spending months building something customers will not buy. Use the interviews to understand what problem they actually have and what they have already tried — this is the raw material for a better hypothesis. Most successful products emerged from a pivot driven by exactly this kind of early signal.","category":"Tutorial","lastModified":"2026-10-29T15:33:37.487562+00:00","metaTitle":"How Founders Validate Product Ideas with Customer Interviews — Koji","metaDescription":"Learn how to run customer validation interviews that actually work. The 5-stage playbook used by YC founders, with AI-powered tools to go faster.","keywords":["customer validation interviews","how to validate product ideas","founder customer discovery","product validation startup","customer interview guide","startup idea validation","customer discovery interviews","product market fit research"],"aiSummary":"This guide explains how founders can validate product ideas using customer interviews before writing any code. It covers the core failure modes (pitching instead of exploring), a 5-stage interview playbook based on YC methodology, how many interviews are actually needed (20-30 minimum), synthesis techniques, and how AI-powered tools like Koji make the entire process 10x faster for resource-constrained founding teams.","aiKeywords":["product validation","customer discovery","founder interviews","startup research","customer development","product market fit","user interviews","AI research tools"],"aiContentType":"tutorial","faqItems":[{"answer":"Most practitioners recommend 20-30 structured discovery interviews as the minimum to identify reliable patterns. Start with 5-10 to refine your hypothesis, then run another round to validate it. Keep going until you stop hearing new problems — that is thematic saturation.","question":"How many customer interviews do I need to validate a startup idea?"},{"answer":"Pitching their idea instead of exploring the problem. The right approach is to ask about the person's past behavior, existing struggles, and what they have already tried — not their hypothetical future intentions.","question":"What is the biggest mistake founders make in customer interviews?"},{"answer":"Both have a place. Live exploratory conversations are best at the start when you do not know what questions to ask. AI-moderated interviews are more efficient once you have a defined hypothesis to test — you can run dozens of conversations in the time it takes to schedule one Zoom call.","question":"Should I use AI interviews or live video calls for validation?"},{"answer":"Two routes. If you do not have the people yet, Koji's Panel recruit finds them: describe the audience, add a screening question for people who actually have the problem, and see the price per completed interview in credits before you launch. If you are recruiting yourself: LinkedIn outreach to target job titles, niche Slack communities, industry subreddits, and warm intros from advisors. Cold outreach gets 2-10% response rates, so plan for volume. Avoid friends and family: social pressure distorts feedback.","question":"How do I find people to interview for product validation?"},{"answer":"This is a success. Use the interviews to understand what problem they actually have and what they have already tried — this is the raw material for a better hypothesis. Most successful products emerged from a pivot driven by this kind of early signal.","question":"What should I do if validation interviews show my idea will not work?"}],"relatedTopics":["startup validation","customer discovery","product market fit","founder research","customer development","AI interviews","user research for startups"]},{"type":"blog","id":"98a5a627-a25c-4371-a48d-a875533b9fe3","slug":"continuous-discovery-handbook-weekly-customer-interviews","title":"The Continuous Discovery Handbook: How Product Teams Run Weekly Customer Interviews (2026)","url":"https://www.koji.so/blog/continuous-discovery-handbook-weekly-customer-interviews","summary":"Continuous discovery is the practice of running weekly customer interviews, conducted by the full product team, to maintain a persistent feedback loop from real customers. This guide covers Teresa Torres' Opportunity Solution Tree framework, how to set up a weekly interview pipeline, story-based interviewing techniques, and how AI tools like Koji make continuous discovery operationally achievable for any team size.","content":"64% of software features are rarely or never used.\n\nThat statistic, from Standish Group research confirmed by Pendo's analysis of 6,800 products, represents roughly $29.5 billion in wasted R&D annually. Teams built things. They shipped them. Customers ignored them.\n\nThe root cause is not poor engineering. It is disconnected product development: teams making roadmap decisions based on assumptions, stakeholder requests, and internal opinions — without a continuous signal from the people they are building for.\n\nContinuous discovery is the antidote. And for modern product teams, AI is finally making it operationally achievable.\n\n## What Is Continuous Discovery?\n\nTeresa Torres, product coach and author of *Continuous Discovery Habits* (2021), defines it precisely:\n\n> \"At a minimum, weekly touchpoints with customers by the team that's building the product, where they conduct small research activities in pursuit of a desired product outcome.\"\n\nThe key words are **weekly**, **the team** (not just the researcher), and **in pursuit of a desired outcome** (not generic insight-gathering).\n\nContinuous discovery is not a quarterly research project. It is not a usability test before shipping. It is a structural habit built into how product teams operate every single week.\n\nThe current reality falls far short of this standard. According to the UserInterviews *State of User Research 2024*, only 44% of teams conduct continuous research. The majority — 56% — do research only project-by-project or not at all. Meanwhile, 46% of product strategy is driven by senior leadership direction or sales and support feedback, not customer data (ProductPlan, 2024).\n\nThis is the gap continuous discovery closes.\n\n## The Opportunity Solution Tree\n\nTeresa Torres developed the Opportunity Solution Tree (OST) as the primary organizing tool for continuous discovery. It maps four levels:\n\n1. **Desired outcome** — the business or product metric your team is trying to move\n2. **Opportunity space** — customer needs, pain points, and desires discovered through interviews\n3. **Solutions** — product ideas mapped to specific opportunities\n4. **Assumptions and experiments** — the cheapest test to validate each solution before building\n\nThe OST prevents the most common failure mode in product development: jumping from \"a customer mentioned this\" directly to \"let's build it,\" skipping the step of understanding whether that need is widespread, addressable, and worth solving.\n\nWithout this structure, teams build features that solve one person's problem — or worse, features that nobody specifically requested but that seemed logical in a planning meeting.\n\n## The Weekly Interview Habit\n\nThe foundational practice is simple: **one customer interview per week, every week**.\n\nNot one per sprint. Not one before a major release. Every week.\n\nTorres argues this cadence is achievable by any product team, even those without dedicated research staff. One interview per week generates approximately 50 customer conversations per year — a meaningful, growing knowledge base that informs decisions in real time.\n\nUnlike episodic research (big sprint, long gap, another sprint), weekly interviews give your team live signal that can influence decisions as they are being made. The team that interviewed three customers this week has better context for tomorrow's prioritization meeting than the team that ran a big study three months ago.\n\nA 2022 CDH Benchmark Survey of nearly 2,000 respondents found that even in Teresa Torres' own practitioner community — which skews heavily toward teams trying to do this well — only 45% had talked to a customer in the past week. The gap between intention and practice is real. But it is closeable.\n\n## Setting Up Your Continuous Discovery System\n\n### Step 1: Define Your Outcome\n\nBefore scheduling any interviews, name the specific metric your team is trying to move. \"Improve the product\" is not an outcome. \"Increase week-2 retention from 42% to 55%\" is an outcome.\n\nYour interviews should be focused on understanding the customer behaviors, needs, and pain points that connect to this outcome. Unfocused discovery generates interesting notes. Outcome-focused discovery generates decisions.\n\n### Step 2: Build Your Interview Pipeline\n\nThe operational bottleneck for most teams is finding people to talk to. The UserInterviews *State of User Research 2024* found that 61% of researchers struggle with recruiting — and for teams without a dedicated researcher, this is even harder.\n\nStrategies that work in practice:\n\n**In-product recruitment:** Intercept users after a key action. \"You just completed your first project — would you answer a few questions?\" Convert a small percentage into interview participants on an ongoing basis.\n\n**Customer success referrals:** CS teams talk to customers constantly. A monthly ask — \"Can you introduce us to someone willing to talk to the product team?\" — builds a steady pipeline over time.\n\n**Research panel building:** After any interview, ask participants if they would be willing to talk again in 3–6 months. Most say yes. This builds your own private panel for the questions only existing customers can answer.\n\n**AI-powered async interviews:** Configure an AI interviewer to run conversations at scale — participants complete interviews on their own schedule, without any scheduling or moderator overhead. Koji is built specifically for this model, and it recruits too: import the customers from your CRM so each gets a personalized link, or open Panel recruit when the outcome depends on people you do not have yet, such as prospects in a segment you are entering, and approve the price per completed interview in credits before you launch.\n\n### Step 3: Run Story-Based Interviews\n\nThe most common error in product team interviews: asking customers what they want. According to Nielsen Norman Group research, leading questions, closed questions, and hypothetical questions are the three most frequent mistakes — and all three produce unreliable data.\n\nStory-based interviews focus on past behavior, not future intentions:\n- \"Tell me about the last time you [did the thing you are researching].\"\n- \"What were you trying to accomplish?\"\n- \"What made that hard?\"\n- \"What did you do to work around it?\"\n- \"What did that cost you in time or money?\"\n\nThese questions surface the actual opportunity space — real problems customers have encountered and cared enough to work around — rather than wish lists that often do not reflect actual behavior or willingness to pay.\n\n### Step 4: Update Your Opportunity Solution Tree\n\nAfter each interview, spend 15–20 minutes updating your OST:\n\n1. What new opportunities (customer needs, pain points, desires) did you hear?\n2. Do they map to existing opportunities on your tree, or are they genuinely new?\n3. Did any interviews challenge or invalidate existing assumptions?\n4. Do new opportunities suggest solution directions worth exploring?\n\nThe OST is a living document. It should evolve weekly as you gather new signal. Teams that use the OST as a static artifact defeat its purpose.\n\n### Step 5: Test Assumptions Before Building\n\nFor every solution on your OST, identify the riskiest assumption — the one that, if wrong, would cause the solution to fail. Design the cheapest possible test:\n\n- A landing page that describes the feature and measures sign-ups\n- A prototype walkthrough with three customers\n- A \"fake door\" button in the UI that surfaces intent before the feature is built\n- A short async interview asking specifically about the assumption\n\nResearch shows teams can validate 80% of critical assumptions using experiments that cost just 5–10% of full development investment. Teams using this approach validate assumptions 5–8x faster than teams that build complete solutions before testing.\n\n## The Real Cost of Skipping Discovery\n\nThe data on what happens when teams skip continuous discovery is sobering.\n\n**$29.5 billion in wasted R&D annually** — Pendo's analysis of 6,800 products found that only 6.4 out of every 100 features drive 80% of user activity. Best-in-class products reach 15.6 per 100. The rest is shipped and ignored.\n\n**40–50% of development effort** goes to features customers rarely or never use, according to multiple studies cited by the Standish Group and corroborated by Pendo's benchmarks.\n\n**$2 trillion in annual global software rework costs** — the downstream cost of building without validating. Teams that invest 2–3 days in rigorous problem framing upstream reduce scope changes by 60%.\n\nThe 1:10:100 Rule, widely cited across product and engineering disciplines, captures the underlying logic: every $1 spent on research upstream saves $10 in development costs and $100 in post-launch support. This is not a precise formula — it is a directional principle that every experienced product leader has felt.\n\n## How AI Is Transforming Continuous Discovery in 2026\n\nThe biggest barrier to continuous discovery is operational: finding time for recruiting, scheduling, note-taking, and synthesis. These are not strategic activities, but they consume the hours most teams do not have.\n\nAccording to the UserInterviews *State of User Research 2025*, AI usage in research jumped from 56% to 80% in a single year — the fastest adoption rate in any research practice area. The most common applications:\n\n- Automated transcription (the first and most universal use case)\n- Thematic synthesis across multiple interviews simultaneously\n- AI-generated discussion guides and research briefs\n- Automatic tagging and clustering of feedback across sources\n\nFor teams practicing continuous discovery, AI interview platforms like Koji change the operational model entirely. Instead of scheduling and moderating individual sessions, you configure your study once and the AI interviewer conducts conversations asynchronously. Thematic analysis runs automatically across all sessions — not one at a time, but across your entire interview dataset.\n\nDovetail reported that customers including Amazon, Canva, Meta, Notion, and Mayo Clinic saved over 38 hours per week using AI-powered research features. Manual interview synthesis — which typically takes 2–3 hours per session — compresses to minutes.\n\nTeams using AI-assisted research are running 5x more studies in the same time window as teams using traditional manual methods, according to early 2025 data from Sprig.\n\nThis does not replace the human judgment needed to interpret findings and make product decisions. But it removes the operational friction that causes most teams to do less research than they know they should.\n\n## Common Pitfalls in Continuous Discovery\n\n**Pitfall 1: Interviewing for confirmation, not discovery**\nThe most dangerous interview is one where you go in believing you already know the answer. Confirmation bias — filtering what you hear through your existing hypothesis — is the dominant error in founder and PM-run research (documented extensively by Rob Fitzpatrick in *The Mom Test*). The antidote: write down your assumptions before the interview, then actively look for evidence that contradicts them.\n\n**Pitfall 2: Building an insight graveyard**\nTeams that do research but do not connect it to decisions create an insight graveyard — a growing archive of reports nobody reads. The OST prevents this by mapping every opportunity to a specific product outcome. If an insight does not connect to a desired outcome, it is filed, not acted upon.\n\n**Pitfall 3: Researcher as gatekeeper**\nOne of Torres' most important contributions is the principle that the entire product trio — PM, designer, and engineer — should attend customer interviews, not just the researcher. Only 16.8% of product trios reported that all three members participated in their last interview (CDH Benchmark Survey, 2022). Engineers who have directly heard a customer's pain point make very different technical decisions than engineers who received a secondhand summary.\n\n**Pitfall 4: Waiting for certainty before acting**\nSome teams fall into analysis paralysis — waiting until they have heard the same thing from 20 customers before acting. Continuous discovery is not about eliminating uncertainty. It is about maintaining a steady signal that reduces uncertainty over time. Ship small bets, measure, learn, iterate.\n\n## Measuring the Impact of Continuous Discovery\n\nAccording to the Maze *State of User Research 2024*, teams that conducted research bi-weekly or more frequently were 2.4x more likely to report high confidence in their product roadmap decisions.\n\nTeams with higher research frequency also correlated with stronger organizational buy-in: 75% of active research teams reported high peer buy-in (up from 66% the prior year), and 57% had strong leadership buy-in (up from 43%), per UserInterviews 2024 data.\n\nThe simplest case for continuous discovery is the one built from absence: companies that do not talk to customers regularly build the wrong things. The Standish Group documented this in 2002. Pendo confirmed it in 2024 with 6,800 products. The teams that avoid the $29.5 billion waste problem are the ones that maintained the habit.\n\n## Getting Started This Week\n\nYou do not need to redesign your entire research practice to start. One interview this week is better than a perfect system next quarter.\n\n1. **Name your current outcome:** What metric is your team trying to move in the next 90 days?\n2. **Identify three customers to talk to:** CS referrals, in-product intercept, or LinkedIn outreach.\n3. **Schedule 30 minutes each:** Use the story-based questions above. Take notes. Do not pitch.\n4. **Update or create your OST:** Map what you heard to opportunities.\n5. **Repeat next week.**\n\nFor teams that want to scale this habit without adding moderator overhead, Koji's AI interviewer lets you run continuous research asynchronously — gathering and synthesizing customer conversations without scheduling a single call. You get from question to insight in hours, not weeks.\n\n[Start your first continuous discovery study on Koji →](/docs/getting-started)\n\n## Frequently Asked Questions\n\n**What is continuous discovery in product management?**\nContinuous discovery is the practice of running at minimum one customer interview per week, conducted by the full product team, to maintain a persistent feedback loop with real customers. Defined by Teresa Torres in *Continuous Discovery Habits* (2021), it uses an Opportunity Solution Tree to connect customer insights to specific product outcomes.\n\n**How many teams actually practice weekly customer interviews?**\nAccording to the UserInterviews *State of User Research 2024*, only 44% of teams conduct continuous research. The majority still research on a project-by-project basis. Even in Teresa Torres' own practitioner community, only 45% had talked to a customer in the past week.\n\n**Why do most teams struggle to maintain a continuous discovery cadence?**\nTime is the top barrier (cited by 43% of teams in the Maze 2024 survey), followed by difficulty recruiting the right participants. Manual recruiting, scheduling, and synthesis each consume significant hours. AI-powered interview tools are now the primary way teams reduce this overhead — AI adoption in research reached 80% in 2025.\n\n**What is the difference between continuous discovery and traditional user research?**\nTraditional research is project-based: a big study at the start, perhaps a usability test before launch. Continuous discovery is an always-on practice that generates ongoing insight as the product evolves. Traditional research informs a specific decision; continuous discovery shapes how the team thinks about every decision.\n\n**How does Koji support continuous discovery?**\nKoji's AI interviewer conducts voice and text conversations with participants asynchronously, without requiring a human moderator. Automatic thematic analysis surfaces patterns across all interviews simultaneously. This lets product teams maintain a weekly research cadence without adding scheduling or synthesis overhead to their workflow.","category":"Research","lastModified":"2026-10-29T15:33:35.965732+00:00","metaTitle":"Continuous Discovery Handbook: Weekly Customer Interviews — Koji","metaDescription":"The complete guide to continuous discovery for product teams. How to run weekly customer interviews, use Opportunity Solution Trees, and use AI to scale your research cadence.","keywords":["continuous discovery","weekly customer interviews","product discovery","opportunity solution tree","Teresa Torres continuous discovery","product research cadence","continuous user research","AI product discovery"],"aiSummary":"Continuous discovery is the practice of running weekly customer interviews, conducted by the full product team, to maintain a persistent feedback loop from real customers. This guide covers Teresa Torres' Opportunity Solution Tree framework, how to set up a weekly interview pipeline, story-based interviewing techniques, and how AI tools like Koji make continuous discovery operationally achievable for any team size.","aiKeywords":["continuous discovery","product discovery","customer interviews","Opportunity Solution Tree","Teresa Torres","weekly research cadence","AI research tools","product team practices"],"aiContentType":"guide","faqItems":[{"answer":"Continuous discovery is the practice of running at minimum one customer interview per week, conducted by the full product team, to maintain a persistent feedback loop with real customers. Defined by Teresa Torres in Continuous Discovery Habits (2021), it uses an Opportunity Solution Tree to connect customer insights to specific product outcomes.","question":"What is continuous discovery in product management?"},{"answer":"According to the UserInterviews State of User Research 2024, only 44% of teams conduct continuous research. The majority still research on a project-by-project basis. Even in Teresa Torres' own practitioner community, only 45% had talked to a customer in the past week.","question":"How many teams actually practice weekly customer interviews?"},{"answer":"Time is the top barrier (cited by 43% of teams in the Maze 2024 survey), followed by difficulty recruiting the right participants. Koji removes the second barrier by recruiting the participants itself: describe the audience, approve a per-respondent quote in credits, and the interviews land in the same study as the customers you invite yourself. AI-powered interview tools are now the primary way teams reduce this overhead — AI adoption in research reached 80% in 2025.","question":"Why do most teams struggle to maintain a continuous discovery cadence?"},{"answer":"Traditional research is project-based: a big study at the start, perhaps a usability test before launch. Continuous discovery is an always-on practice that generates ongoing insight as the product evolves. Traditional research informs a specific decision; continuous discovery shapes how the team thinks about every decision.","question":"What is the difference between continuous discovery and traditional user research?"},{"answer":"Koji's AI interviewer conducts voice and text conversations with participants asynchronously, without requiring a human moderator. Automatic thematic analysis surfaces patterns across all interviews simultaneously, letting product teams maintain a weekly research cadence without adding scheduling or synthesis overhead.","question":"How does Koji support continuous discovery?"}],"relatedTopics":["continuous discovery","product discovery","user research practices","product management","customer interviews","Teresa Torres","AI research","research operations"]},{"type":"blog","id":"deeb25a4-fe6b-41f4-aa62-9c216a7caa36","slug":"koji-vs-grain-2026","title":"Koji vs Grain: AI Research Platform vs Meeting Recorder (2026)","url":"https://www.koji.so/blog/koji-vs-grain-2026","summary":"Koji and Grain serve fundamentally different workflows. Grain records and summarizes sales calls for revenue teams. Koji runs AI-moderated user research interviews and generates structured insights. This comparison breaks down features, pricing, and ideal use cases for each tool in 2026.","content":"Choosing between Koji and Grain? Here's the honest answer upfront: they're built for different jobs. Grain is a conversation recorder that sales and CS teams love. Koji is an AI-powered research platform that conducts, transcribes, and analyzes interviews without a human moderator. If you need to understand *what customers said* in a call you just had, Grain is excellent. If you need to run structured research, synthesize patterns across 20+ participants, and generate insight reports — Koji is built for that.\n\n## Quick Comparison\n\n| Feature | Koji | Grain |\n|---------|------|-------|\n| AI-moderated interviews (voice + text) | ✅ No human moderator needed | ❌ Requires a human-led call |\n| Automatic thematic analysis | ✅ Across all interviews | ⚠️ Single-session AI summaries only |\n| Research brief / study design | ✅ AI-guided study design | ❌ No study design workflow |\n| Participant-facing async interface | ✅ Any time, any device | ❌ Requires scheduling a live call |\n| Cross-interview insight reports | ✅ One-click aggregate reports | ❌ Not built for this |\n| Video call recording | ❌ | ✅ Zoom, Meet, Teams, Webex |\n| CRM sync (HubSpot, Salesforce) | ❌ | ✅ Native deep sync |\n| Sales coaching / conversation intelligence | ❌ | ✅ Core use case |\n| Embed widget / API | ✅ Full REST API on every plan, including free | ✅ Business plan API |\n| Starting price | From €1 per qualified interview | $15/seat/month |\n| Free plan | ✅ 10 credits, no card, one active study | ✅ 20 recordings/month |\n\n## The Core Difference: Research Platform vs. Meeting Recorder\n\nGrain is built around a simple premise: you're having calls, and you want to remember what was said. It joins your Zoom or Google Meet call automatically, records everything, generates a summary, lets you clip highlights, and syncs key moments to your CRM. For sales teams doing 30+ calls a week, this is genuinely transformative.\n\nKoji starts at a different question: *how do I learn what my customers actually think, without scheduling 30 calls?* Koji's AI interviewer conducts conversations autonomously — via voice or text — with your participants. You design the research brief, set your goals, and Koji handles the rest: asking probing follow-up questions, exploring unexpected responses, and synthesizing findings across all participants into a structured report.\n\nThe key distinction: Grain captures calls you're already having. Koji runs research you couldn't otherwise afford to run.\n\n## Where Grain Shines\n\n### Meeting Intelligence for Revenue Teams\n\nGrain excels at conversation intelligence for go-to-market teams. Sales reps can search their entire call library for the moment a prospect mentioned a pain point. Customer success managers can clip a customer quote and share it in Slack. Product managers can build a voice-of-customer library from sales calls without being on every call.\n\nGrain's integrations are genuinely impressive — bidirectional HubSpot sync, Salesforce support, Productboard connections, and a full MCP server with pre-built reports for sales frameworks like SPICED and MEDDICC. If your team speaks in pipeline language, Grain speaks it fluently.\n\nAccording to a 2025 market report, 75% of professionals now use an AI notetaker in their work meetings — and Grain is a leading option for teams that need CRM-connected meeting intelligence at an accessible price point ($15–$29/seat vs. Gong's $100+/user).\n\n### Stakeholder Democratization Through Video Clips\n\nOne of Grain's strongest use cases is getting insights out of recorded calls and into the hands of people who weren't on the call. The video clip and highlight reel features are best-in-class — anyone can watch a 90-second clip of exactly what a customer said, without sitting through a 45-minute recording. With 31,000+ teams using the platform and a 97% 12-month retention rate, it clearly delivers real value.\n\n### Price-to-Value for Sales Teams\n\nAt $15–$29/seat/month, Grain undercuts conversation intelligence incumbents like Gong ($100+/user) and Clari ($60+/user) dramatically. For teams that want CRM-integrated call intelligence without enterprise pricing, it's a compelling option.\n\n## Where Koji Shines\n\n### Running Research Without Scheduling Calls\n\nAccording to the 2025 State of User Research report, 80% of UX researchers now use AI tools — but the core bottleneck remains time. The average researcher spends 30% of their total research time just recruiting and scheduling participants. Koji removes both halves of that cost. If you do not have participants lined up, describe the audience you want and Koji recruits them for you, with a live per-respondent quote in credits you approve before launch. And there is nothing to schedule: participants complete AI-moderated voice or text interviews on their own time, from any device, with no Zoom link required.\n\nA study that would take 3 weeks to schedule with live moderation can run in 72 hours with Koji. You send a link, participants interview with the AI on their schedule, and Koji synthesizes the findings automatically.\n\n### Structured Research, Not Just Notes\n\nGrain gives you a transcript and a summary. Koji gives you a research deliverable. There's a meaningful difference.\n\nWith Koji, every study starts with a research brief — your goals, hypotheses, and target participant profile. Koji's AI consultant helps you design the question framework, anticipates follow-up paths, and ensures conversations stay on-topic while still exploring unexpected threads. After interviews complete, Koji surfaces themes, sentiment patterns, and key quotes across all participants — not per-session, but as an aggregated research report.\n\nThis is the gap Grain can't fill. Grain is excellent at preserving individual conversations. Koji is built to synthesize meaning across many.\n\n### No Moderator Bias\n\nHuman moderators, however skilled, introduce variability into sessions — through tone, follow-up patterns, and inadvertent leading questions. Koji's AI conducts every interview with the same structure, the same curiosity, and the same rigor, eliminating moderator bias as a confound in your findings.\n\n### Scale Without Headcount\n\nThe UX Research Software market is projected to grow from $577M in 2025 to $1.23B by 2032, driven by demand to scale research without proportional headcount growth. Koji directly addresses this: a single researcher can run 50-participant studies without hiring moderators or scheduling coordinator support.\n\n## When to Choose Grain\n\n- You're in a sales, CS, or account management role and want intelligence from calls you're already having\n- Your primary need is CRM sync — connecting customer conversations to pipeline data\n- You want to create shareable video clips for stakeholder alignment\n- You're replacing a more expensive conversation intelligence tool like Gong or Clari\n\n## When to Choose Koji\n\n- You need to run structured qualitative research — discovery, concept testing, churn analysis, feature validation\n- You want to interview 20+ participants without scheduling 20+ calls\n- You need a synthesized research report, not just individual call notes\n- You're doing product research, UX research, or market research (not capturing sales conversations)\n\n## When You Might Use Both\n\nSome teams use both tools for different jobs. Grain captures the ongoing stream of customer conversations from sales and CS calls. Koji runs structured research projects when you need to go deeper on a specific question. Your Grain library provides qualitative signal; Koji provides research-grade evidence. They're not really competing for the same workflow.\n\n## Pricing Comparison\n\n| | Koji | Grain |\n|--|------|-------|\n| Free plan | 10 credits, no card, one active study | 20 recordings/month |\n| Entry paid | From €1 per qualified interview | $15/seat/month |\n| Mid-tier | Volume pricing and plans when you want them | $29/seat/month |\n| Primary use case | Research studies | Meeting recording |\n| Seat model | Per workspace | Per recorder seat |\n\nGrain's per-seat model works well for sales teams where many people record calls. Koji's workspace model works better for research teams where a few researchers run studies for the broader organization.\n\n## The Bottom Line\n\nGrain helps you remember what customers said. Koji helps you understand what it means.\n\nThe AI Meeting Assistants market is projected to hit $34.3 billion by 2035 — and tools like Grain are riding that wave. They're genuinely useful for what they do. But recording a call and conducting research are different activities with different outputs. If you're running qualitative user research — understanding motivations, validating ideas, diagnosing problems — you need structured study design, AI moderation with adaptive probing, and synthesis that turns raw conversations into insight you can act on.\n\n---\n\n*Last verified: April 2026. Competitor features and pricing are subject to change — check each vendor's website for the latest information.*\n\n## Start Running Research Without Scheduling 30 Calls\n\nKoji's AI interviewer conducts voice and text interviews with your participants automatically — no moderator, no scheduling, no manual synthesis. From research question to structured report in hours, not weeks. [Try Koji free](/docs/getting-started).\n\n## Frequently Asked Questions\n\n**Can Grain replace user research tools like Koji?**\nNot for structured research. Grain is optimized for sales and CS teams — it doesn't have study design workflows, AI moderation, or cross-interview synthesis. For qualitative research with adaptive AI interviews and automated reports, Koji is the more capable platform.\n\n**Does Grain do qualitative research analysis?**\nGrain provides per-session AI summaries and lets you clip highlights and search transcripts. It does not do cross-interview thematic analysis or structured insight synthesis. Each recording is largely self-contained — it's a library of calls, not a research findings engine.\n\n**Is Koji cheaper than Grain for a small research team?**\nGrain starts at $15/seat/month per recorder. Koji has no seats at all: you pay per qualified interview, as low as €1. For a research team of 1–3 researchers running 5–10 studies per month, Koji's workspace model is typically more cost-effective than per-seat pricing across multiple recorders.\n\n**Can I use Grain for async user research?**\nNo. Grain requires a live video call — there's no participant-facing async interface. Koji interviews happen asynchronously: participants click a link and complete the AI interview on their own schedule, without scheduling a call.\n\n**What's the main advantage of Koji over meeting recorders for research?**\nSynthesis at scale. Grain preserves individual call recordings. Koji synthesizes meaning across all interviews in a study — surfacing common themes, sentiment patterns, and key insights without any manual analysis. For research with 10+ participants, this is the difference between a library of transcripts and a usable research report.","category":"Research","lastModified":"2026-10-29T15:33:34.884747+00:00","metaTitle":"Koji vs Grain: AI Research vs Meeting Recorder (2026)","metaDescription":"Koji vs Grain compared: AI-moderated research interviews vs meeting recording. Find out which tool fits your user research workflow in 2026.","keywords":["koji vs grain","grain alternative","user research tools","meeting recorder vs research platform","ai interview tool","conversation intelligence"],"aiSummary":"Koji and Grain serve fundamentally different workflows. Grain records and summarizes sales calls for revenue teams. Koji runs AI-moderated user research interviews and generates structured insights. This comparison breaks down features, pricing, and ideal use cases for each tool in 2026.","aiKeywords":["Koji","Grain","comparison","meeting recorder","AI interviews","conversation intelligence","user research","qualitative research","research platform"],"aiContentType":"comparison","faqItems":[{"answer":"Not for structured research. Grain is optimized for sales and CS teams and lacks study design workflows, AI moderation, and cross-interview synthesis. For qualitative research with adaptive AI interviews and automated reports, Koji is the more capable platform.","question":"Can Grain replace user research tools like Koji?"},{"answer":"Grain provides per-session AI summaries and lets you clip highlights and search transcripts. It does not do cross-interview thematic analysis or structured insight synthesis — each recording is self-contained.","question":"Does Grain do qualitative research analysis?"},{"answer":"No. Grain requires a live video call. Koji interviews happen asynchronously: participants click a link and complete the AI interview on their own schedule, without scheduling a call.","question":"Can I use Grain for async user research?"},{"answer":"Synthesis at scale. Koji synthesizes meaning across all interviews in a study — surfacing common themes and key insights without manual analysis. For 10+ participant studies, this is the difference between a library of transcripts and a usable research report.","question":"What is the main advantage of Koji over meeting recorders for research?"},{"answer":"For a research team of 1–3 researchers running multiple studies per month, Koji's seat-free model, priced from €1 per qualified interview, is typically more cost-effective than Grain's per-seat pricing across multiple recorders.","question":"Is Koji cheaper than Grain for a small research team?"}],"relatedTopics":["User Research Tools","AI Interviews","Conversation Intelligence","Qualitative Research","Research Platform Comparison","Meeting Intelligence"]},{"type":"blog","id":"9e13ec41-6cb5-4bcc-a540-553002959808","slug":"koji-vs-marvin-2026","title":"Koji vs Marvin: Full-Stack AI Research vs Analysis Repository (2026)","url":"https://www.koji.so/blog/koji-vs-marvin-2026","summary":"Koji and Marvin serve different points in the research lifecycle. Marvin is a research repository for organizing and analyzing existing qualitative data. Koji is an end-to-end platform that conducts AI-moderated interviews and synthesizes findings into structured reports. This comparison breaks down when to use each and when you might need both.","content":"Koji and Marvin both have \"AI research\" in their positioning — but they sit at very different points in the research workflow. Marvin is an AI-powered research repository built to organize and analyze data you've already collected. Koji is an end-to-end research platform that conducts AI-moderated interviews and synthesizes findings automatically. If you need a place to store and tag existing research, Marvin is worth evaluating. If you want to run new research without a human moderator and get structured insights without manual analysis, Koji does that job.\n\n## Quick Comparison\n\n| Feature | Koji | Marvin |\n|---------|------|--------|\n| AI-moderated interviews (voice + text) | ✅ Core product | ⚠️ Newer add-on feature |\n| Research repository / data storage | ❌ Not a repository | ✅ Core product |\n| Automatic thematic analysis | ✅ Across all interviews | ✅ Cross-project AI analysis |\n| Study design / research brief | ✅ AI-guided | ❌ Not included |\n| Participant-facing async interface | ✅ Any device, any time | ❌ Requires bringing recordings |\n| Import external recordings/transcripts | ❌ | ✅ 30+ integrations |\n| One-click aggregate reports | ✅ | ✅ |\n| CRM / support tool integrations | ❌ | ✅ Zendesk, Salesforce, Slack |\n| 40+ language transcription | ❌ | ✅ |\n| Participant recruitment | ✅ Built-in, quoted in credits, or use your own list | ❌ (bring your own) |\n| HIPAA / SOC2 compliance | ✅ | ✅ |\n| Transparent pricing | ✅ From €1 per qualified interview | ❌ Contact sales |\n\n## The Core Difference: Run New Research vs. Analyze Existing Data\n\nMarvin was built to solve a specific problem: qualitative data is scattered everywhere — in Notion docs, Zoom recordings, Salesforce calls, Zendesk tickets, Slack threads — and there's no single place to find what customers have said about a topic. Marvin ingests all of it, auto-tags and categorizes it, and lets you search and query across your entire research archive.\n\nKoji solves a different problem: you need structured research insights but don't have time to schedule, moderate, and manually analyze interviews. Koji runs AI-moderated voice and text interviews with your participants, then synthesizes findings into a structured report — without any human moderation, manual coding, or synthesis work.\n\nThe practical implication: Marvin is downstream research infrastructure. Koji is upstream research execution. They're not really competing products — they're tools for different points in the research lifecycle.\n\n## Where Marvin Shines\n\n### The Research Repository\n\nIf your organization has years of research data spread across Zoom recordings, Google Drive docs, Notion pages, Salesforce notes, and Zendesk tickets — Marvin is excellent at consolidating it. With 30+ native integrations, it can ingest the full breadth of customer conversation data your organization already has.\n\nThe cross-project analysis feature is genuinely powerful for organizations with large research archives: Marvin can surface themes and quote evidence across multiple studies simultaneously, helping you see patterns you'd miss by reading projects individually.\n\n### AI Analysis for Existing Qualitative Data\n\nMarvin's \"Deep Research\" mode and \"Ask AI\" feature let you query your research library conversationally. You can ask \"what did enterprise customers say about the onboarding process?\" and get an answer with citations from source transcripts — a meaningful time-saver for research leads who need to surface evidence quickly without re-reading everything.\n\nThe 40-language transcription support is a genuine advantage for global organizations conducting multilingual research programs.\n\n### Organizational Knowledge Infrastructure\n\nFor research teams at larger companies, Marvin functions as organizational memory: a place where insights don't disappear when a researcher leaves, where stakeholders can self-serve without filing a research request, and where patterns accumulate over time. This is genuinely hard to replicate with general-purpose tools like Notion or Confluence.\n\n## Where Koji Shines\n\n### Running Research You Couldn't Otherwise Run\n\nAccording to the 2025 State of User Research report, 80% of UX researchers now use AI tools — but researcher time remains the primary constraint. Demand for research increased for 66% of teams while junior headcount declined, meaning researchers are being asked to do significantly more with fewer people.\n\nKoji addresses this directly. A researcher using Koji can run a 30-participant study in the time it used to take to schedule five moderated sessions. Koji's AI interviewer conducts every conversation autonomously — asking follow-up questions based on participant responses, probing unexpected threads, and maintaining conversational depth without the moderator's schedule as a bottleneck.\n\n### No Recordings Required\n\nMarvin is fundamentally a tool for data you already have — you bring the recordings, transcripts, or documents, and Marvin analyzes them. Koji generates that data from scratch. You share a link with participants, they complete an AI-moderated voice or text interview on their own time, and the findings arrive already synthesized. If you do not have participants of your own, Koji recruits them: describe the audience you want and Koji returns a live per-respondent quote in credits that you approve before launch.\n\nFor teams that don't have a backlog of recorded interviews to analyze, or who need research on a new question not covered in their archive, Koji is the starting point — not Marvin.\n\n### Structured Research Design\n\nKoji begins every study with a research brief. You define your goals, hypotheses, and target participant profile. Koji's AI consultant helps you design the interview structure — what to explore, how to probe, what follow-up directions to prepare for. This methodological rigor upstream produces cleaner, more actionable findings downstream.\n\nMarvin has no study design layer. It assumes you're bringing in data from studies that were already designed and conducted elsewhere.\n\n### Eliminating Moderator Bias\n\nHuman moderators, however skilled, introduce variability into sessions — through tone, follow-up patterns, and inadvertent leading questions. Koji's AI conducts every interview with identical structure and consistent probing, ensuring your findings reflect what participants actually think rather than artifacts of moderation variation.\n\n### Transparent Pricing\n\nKoji's pricing starts as low as €1 per qualified interview for the full platform — AI interviewer, study design, analysis, and reports included. Marvin's pricing requires contacting sales; multiple G2 reviewers describe it as \"quite expensive compared to competitors.\" For teams that need to build a business case for research tooling, pricing transparency matters.\n\n## A Note on Marvin's AI Interviewer\n\nMarvin recently added an AI Moderated Interviewer feature, which moves it partially into Koji's territory. This is worth acknowledging: Marvin can now conduct AI-moderated interviews, not just analyze existing data.\n\nThe distinction is about depth and primary use case. Koji was purpose-built as an interview platform — adaptive probing, voice-native interaction, research brief design, and thematic synthesis are all first-class features. Marvin's AI interviewer is a newer addition to what remains primarily a repository product. Reviews of Marvin also note AI accuracy concerns and hallucination risk on thin datasets — a critical issue when the AI's output is a research deliverable that informs product decisions.\n\nFor teams whose primary need is conducting structured AI interviews, Koji is the more mature and purpose-built choice.\n\n## When to Choose Marvin\n\n- You have a large existing library of qualitative research data (recordings, transcripts, support tickets)\n- Your primary need is organizational memory — making past research findable and queryable\n- You have diverse data sources (support calls, sales calls, user interviews) that need to be integrated\n- You're building research operations infrastructure at a large organization\n- Cross-project analysis across existing studies is your highest-priority capability\n\n## When to Choose Koji\n\n- You need to run new research, not just analyze existing data\n- You want AI-moderated interviews as the primary value, not a secondary add-on\n- You need insights from 20+ participants without scheduling 20+ calls\n- You want a structured research report as the output, not a queryable database\n- You're a small-to-medium research team that needs research ROI, not just research storage\n\n## When You Might Use Both\n\nThe most sophisticated research teams eventually need both: Koji to generate new insights efficiently at the front end of every project, and a repository tool to accumulate and search organizational knowledge over time. For teams starting from scratch, Koji provides the fastest path to actionable research insights without scheduling overhead.\n\n## The Bottom Line\n\nMarvin answers the question: \"What do we already know?\" Koji answers the question: \"What do we need to find out?\"\n\nIf your most urgent problem is running user research faster and more efficiently — conducting quality interviews without burning researcher time on scheduling and moderation — Koji is the right starting point. If your most urgent problem is that valuable research is scattered across 12 different tools and no one can find it, Marvin addresses that specific pain.\n\nThe good news: they solve different problems well enough that there's rarely a direct trade-off between them.\n\n---\n\n*Last verified: April 2026. Competitor features and pricing are subject to change.*\n\n## Run Your Next Study in Hours, Not Weeks\n\nKoji's AI interviewer conducts voice and text interviews with your participants automatically, then synthesizes findings into a structured report. No moderation, no scheduling, no manual analysis. [Try Koji free](/docs/getting-started).\n\n## Frequently Asked Questions\n\n**Does Marvin replace Koji, or do they work together?**\nThey solve different problems. Marvin is a research repository for organizing and querying data you've already collected. Koji generates new research data through AI-moderated interviews. Many teams could use both: Koji to run studies, and a repository like Marvin or Dovetail to accumulate institutional knowledge over time.\n\n**Marvin now has an AI interviewer — is it the same as Koji's?**\nMarvin recently added an AI Moderated Interviewer as a newer feature. Koji was built from the ground up as an interview platform — adaptive probing, voice-native interaction, and research brief design are all core, mature capabilities. For teams whose primary need is conducting structured AI interviews, Koji is the more purpose-built platform.\n\n**Can Koji replace Marvin as a research repository?**\nNo. Koji is not designed as a repository. It excels at running individual studies and generating reports, but doesn't function as a cross-study, cross-project data archive. For large volumes of historical research, a dedicated repository like Marvin or Dovetail provides capabilities Koji doesn't.\n\n**Which tool is better for startups doing customer discovery?**\nKoji. Startups typically need to generate new insights fast — they don't have years of accumulated research to organize. Koji's AI-moderated interviews can run a discovery study with 15 participants in 48 hours, giving founders and product teams actionable insight before making expensive product decisions.\n\n**Is Marvin or Koji more affordable?**\nKoji's pricing starts as low as €1 per qualified interview and is publicly listed. Marvin's pricing requires contacting sales, and reviewers frequently cite cost as a concern. For smaller teams, Koji's transparent pricing and all-in-one workflow is typically more accessible.","category":"Research","lastModified":"2026-10-29T15:33:33.790325+00:00","metaTitle":"Koji vs Marvin: AI Research Platform vs Repository (2026)","metaDescription":"Koji vs Marvin compared: end-to-end AI interview platform vs research repository. Learn which tool fits your team's research workflow in 2026.","keywords":["koji vs marvin","marvin alternative","heymarvin vs koji","ai research tools","research repository","qualitative research platform","ai moderated interviews"],"aiSummary":"Koji and Marvin serve different points in the research lifecycle. Marvin is a research repository for organizing and analyzing existing qualitative data. Koji is an end-to-end platform that conducts AI-moderated interviews and synthesizes findings into structured reports. This comparison breaks down when to use each and when you might need both.","aiKeywords":["Koji","Marvin","HeyMarvin","comparison","research repository","AI interviews","qualitative analysis","user research","research platform"],"aiContentType":"comparison","faqItems":[{"answer":"They solve different problems. Marvin organizes research you have already collected. Koji generates new research data through AI-moderated interviews. Many teams use both: Koji to run studies and Marvin or Dovetail to accumulate institutional knowledge over time.","question":"Does Marvin replace Koji, or do they work together?"},{"answer":"Marvin recently added an AI Moderated Interviewer as a newer feature. Koji was purpose-built as an interview platform from day one — adaptive probing, voice-native interaction, and research brief design are all core, mature capabilities.","question":"Marvin now has an AI interviewer — is it the same as Koji's?"},{"answer":"No. Koji excels at running individual studies and generating reports, but does not function as a cross-study, cross-project data archive. For large volumes of historical research, a dedicated repository like Marvin or Dovetail provides capabilities Koji does not.","question":"Can Koji replace Marvin as a research repository?"},{"answer":"Koji. Startups need to generate new insights fast and typically lack years of accumulated research to organize. Koji can run a 15-participant discovery study in 48 hours, giving founders actionable insight before making expensive product decisions.","question":"Which tool is better for startups doing customer discovery?"},{"answer":"Koji's pricing starts as low as €1 per qualified interview and is publicly listed. Marvin requires contacting sales and reviewers frequently cite cost as a concern — making Koji more accessible for smaller teams.","question":"Is Marvin or Koji more affordable?"}],"relatedTopics":["User Research Tools","AI Interviews","Research Repository","Qualitative Research","Research Platform Comparison","Research Operations"]},{"type":"blog","id":"5370d0d1-6fbe-4276-a7c2-2605cab8e08b","slug":"how-to-recruit-user-research-participants-2026","title":"How to Recruit User Research Participants: The Complete Guide (2026)","url":"https://www.koji.so/blog/how-to-recruit-user-research-participants-2026","summary":"Recruiting the right participants is the highest-leverage decision in any research project. This guide covers every stage of the recruitment process — defining participant profiles, writing screeners, choosing recruitment channels (panels, customer lists, social outreach, in-product), setting incentives, managing no-shows, and how async AI interviews reduce scheduling friction.","content":"Recruiting the right participants is the single highest-leverage decision you make in any research project. Perfect questions and rigorous analysis can't compensate for the wrong people. Yet recruitment consistently ranks as the #1 pain point for researchers: 54% cite recruitment time as a major challenge, 41% struggle with no-shows, and the average researcher spends 30% of their total research time just finding and scheduling participants. That applies to anyone a business needs to hear from: customers and users, prospects and lost deals, churned accounts, employees at every stage from onboarding to exit, students, portfolio companies, and consumers in markets where you have no list yet. The same steps work for all of them.\n\nThis guide walks you through every stage of the recruitment process — from writing a screener to managing incentives to the emerging approaches that are changing how teams find participants in 2026.\n\n## Why Recruitment Is So Hard (And Why It Matters)\n\nThe paradox of user research is that you need the *right* people, but you usually have limited access to them. Your best customers are the busiest people. Hard-to-reach professionals — executives, developers, healthcare workers — command high incentives and are difficult to schedule. And no matter how carefully you screen, an average of 11% of participants won't show up (NNGroup data), which means you need to build in backups from the start.\n\nThe cost of bad recruitment compounds quickly. Spending 2 weeks finding the wrong participants means 2 weeks of delay plus research findings that don't reflect your actual target user. 40% of UX research projects are delayed due to inefficient recruitment. Investment in good recruitment upfront pays compounding returns throughout the project.\n\n## Step 1: Define Your Participant Profile\n\nBefore you write a single screener question, get specific about who you're looking for. Vague criteria produce vague findings.\n\nA strong participant profile includes:\n\n**Behavioral criteria (most important):**\n- What have they done recently? (\"Has signed up for a B2B SaaS product in the last 6 months\")\n- How frequently do they do the relevant activity? (\"Conducts user interviews at least monthly\")\n- What's their relationship to the problem you're studying? (\"Has experienced X in the last 3 months\")\n\n**Demographic criteria (use sparingly):**\n- Only include demographics that genuinely affect the research question\n- Age, location, and job title are often proxies — look for the behavior instead\n- Be specific about professional context when relevant (\"IC designer at a company with 50+ employees\")\n\n**Exclusion criteria:**\n- People who work in your industry (competitive intelligence risk)\n- People who've participated in research in the last 3 months (research fatigue)\n- People outside the relevant experience level for your use case\n\nPro tip: Write a \"perfect participant\" description in 2–3 sentences before building your screener. If you can describe them clearly, you can screen for them reliably.\n\n## Step 2: Write Your Screener\n\nA screener survey filters candidates before you commit time to qualifying them. Good screeners are short (5–8 questions), specific, and don't telegraph the \"right\" answers.\n\n**Screener writing principles:**\n\n**Use multi-select questions, not yes/no:** Instead of \"Do you conduct user interviews?\" ask \"Which of the following research activities do you do in your current role?\" and include user interviews among several options. This prevents candidates from selecting what they think you want.\n\n**Avoid leading questions:** \"How often do you struggle with research participant recruitment?\" signals that struggling is the expected answer. Instead: \"How would you describe your experience with research participant recruitment?\" with a scale or open response.\n\n**Include disqualifying response options:** Design your screener so the wrong candidates naturally select themselves out. If you need people who've done 5+ interviews in the last month, include options \"0,\" \"1–2,\" \"3–4,\" \"5+\" — people who select the low options disqualify cleanly without any additional friction.\n\n**Keep it under 5 minutes:** Longer screeners reduce response rates significantly. Every question should do necessary work — cut anything that doesn't directly inform qualification.\n\n## Step 3: Choose Your Recruitment Channel\n\nDifferent channels work for different audiences. Use this framework to choose the right one for your study.\n\n### Participant Panels (Fastest for General Consumer Research)\n\nPlatforms like User Interviews and Respondent maintain large panels of pre-screened, incentive-ready participants. They're fast — you can have your first interviews within 24–48 hours — but costs add up:\n\n- **User Interviews:** ~$40/session (pay-as-you-go) + 50% of participant incentive as a platform fee\n- **Respondent:** ~$39/session B2C, ~$65/session B2B\n\nFor harder-to-reach professional audiences, total all-in costs can reach $200–$300 per session including incentives. According to the 2025 User Interviews Research Budget Report, participant recruitment and incentives account for roughly one-fifth of total research budgets.\n\nYou do not need a separate vendor for this channel. [Panel recruitment](/docs/panel-recruitment) is built into Koji on paid plans: describe the audience you need (market, demographics, screening) and Koji returns a live per-respondent quote. The quote is shown and paid in credits, and you approve it before launch. Interviews use the normal interview credits on top.\n\n**Best for:** Consumer studies, broad professional audiences, fast turnaround needs\n\n### Your Own Customer Base (Best Quality, Lowest Cost)\n\nRecruiting from your customer list produces the most relevant participants for product research. They understand your context, you can layer in usage data to find the right segment, and incentive costs are lower. Reaching them does not have to mean email: with Koji you can send the interview link, embed it in your product, paste phone numbers and let Koji call them, or share a call link so people enter their number and Koji calls back. Which link you send is a choice too. Export the segment from your CRM and import it so each person gets a personalized link: you see who responded, the interviewer greets them by name, and the report is already segmented by plan, tenure or whatever else you imported. If candor matters more than tracking, churned customers, an exit interview, a sensitive topic, share the generic link and leave the lead form off, so responses stay anonymous. And when the question is about non-customers or a market you have no list for, Panel recruit above finds them across 130+ countries at a price you approve first. All routes land in the same study and the same report.\n\nThe challenge: availability and self-selection bias. Your most engaged customers may not represent your struggling or churned segments. Build a research opt-in CRM — a simple list of customers who've agreed to be contacted — to make this channel reliably available.\n\n**Best for:** Product validation, feature research, churn analysis, onboarding research\n\n### Community and Social Outreach (Best for Niche Audiences)\n\nLinkedIn, Reddit, Slack communities, and Discord servers can reach specific professional niches that panels underrepresent. A post in a Slack community for research ops professionals reaches people panels don't have. A targeted LinkedIn message to product managers at fintech startups reaches a segment you can't easily filter for in a general panel.\n\nThe tradeoff: slower (1–2 weeks for responses), less predictable volume, and more manual follow-up. Works best when you have a compelling incentive and a specific enough request that the right people self-select.\n\n**Best for:** Niche professional audiences, researchers, specialized experts\n\n### In-Product Intercepts (Best for Capturing In-Context Users)\n\nIf you have an active product, in-app recruitment is powerful: you can target users while they're using relevant features, link recruitment to behavioral signals (users who just completed onboarding, users inactive for 30 days), and pre-qualify based on actual usage data.\n\nConversion rates are typically 2–5% but quality is high because you're finding participants in the moment of relevance.\n\n**Best for:** Feature-specific research, behavioral segments, product discovery\n\n## Step 4: Set Your Incentives\n\nIncentives are the most-often miscalibrated part of recruitment. Too low and response rates collapse; too high and you attract people who just want the money.\n\n**The benchmark:** Approximately $3/minute for moderated sessions — or $90–$200/hour depending on participant type. The midpoint is ~$145/hour (2025 User Interviews Incentives Report).\n\n**Incentive by audience type:**\n- General consumers / students: $50–$80 for a 30-minute session\n- Working professionals: $100–$150 for a 30-minute session\n- Senior professionals / executives: $150–$300+ for a 30-minute session\n- Niche specialists (doctors, lawyers, engineers): $200–$400+ for a 30-minute session\n\n**Format matters:** Amazon gift cards and Visa prepaid cards are the most universally accepted. For B2B audiences, charitable donations in the participant's name also work well. Avoid account credits or subscription upgrades as the only incentive — they only work for participants who already value your product.\n\n## Step 5: Manage No-Shows and Drop-offs\n\nThe average no-show rate in user research is 11% (NNGroup) — but it spikes for cold panel recruits and community outreach. Plan for it:\n\n- **Over-recruit by 20%:** If you need 10 participants, target 12 confirms\n- **Send confirmation sequences:** Email the day before + 1 hour before each session\n- **Use SMS reminders:** Text confirmations dramatically reduce no-shows compared to email alone\n- **Maintain a waitlist:** Keep 2–3 alternates on standby for last-minute drops\n\nFor in-person sessions, no-show rates can reach 15–20%. For async research (more on this below), no-shows effectively cease to exist — participants complete on their own schedule.\n\n## Step 6: The Async Alternative — How AI Interviews Change the Recruitment Equation\n\nOne of the biggest shifts in research practice since 2024 is the rise of async AI-moderated interviews. Instead of scheduling participants for live sessions, you send them a link. They complete an AI-moderated voice or text interview whenever it's convenient — during a lunch break, on a commute, at 11pm.\n\nThe impact on recruitment is significant. According to the 2025 State of User Research, 57% of research teams report AI has improved project turnaround time, and 58% report improved team efficiency. The logistical friction that makes recruitment hard — coordinating schedules, managing calendar availability, handling time zones — disappears when participants don't need to synchronize with a human moderator.\n\nWith Koji, you design the research brief, set up your AI interviewer, and share a link, embed it in your product, or paste phone numbers and let Koji call people at a time you schedule. If you need participants you do not already have, recruit them from Koji's built-in panel: describe the audience, review the live per-respondent quote in credits, and approve it before launch. Participants complete the study on their own time. Koji's AI asks adaptive follow-up questions — probing unexpected responses, exploring threads that emerge mid-conversation — so you get the depth of a human-moderated interview without the scheduling constraint.\n\nThis doesn't eliminate the need for good screening. You still need the right participants. But it dramatically expands the pool of willing participants by removing the single biggest barrier: finding a time that works.\n\n## Common Recruitment Mistakes (And How to Avoid Them)\n\n**Recruiting too late:** Build 2–3 weeks of recruitment buffer into every research plan. Starting recruitment after finalizing research questions is almost always too late for B2B research.\n\n**Screener too long:** Every additional screener question reduces completion rate. Keep it under 8 questions, under 5 minutes.\n\n**Recruiting only the happy path:** Research teams consistently over-recruit satisfied users. Deliberately include at-risk customers, churned users, and non-users in your pool — the edges of your customer base often produce the most actionable insights.\n\n**Not maintaining a participant CRM:** Every participant who completes research is a warm contact for future studies. Maintain a simple opt-in list with their role, experience level, and study history. Import it into Koji and each person gets a personalized link, so the next study is one upload away.\n\n**Ignoring screener disqualification rates:** If 80%+ of screener respondents are being disqualified, your recruitment channel is wrong for your criteria — not your screener. Revisit the channel before loosening criteria.\n\n## Key Takeaways\n\n- Define behavioral criteria before demographic criteria — behaviors predict research relevance better than demographics\n- Budget recruitment generously: average researcher labor for a 5-person study is ~7 hours on recruitment alone\n- Over-recruit by 20% and maintain a waitlist for synchronous sessions\n- Use your customer list for the highest-quality participants at lowest cost\n- Incentive benchmarks: ~$3/minute, or $90–$200/hour depending on audience type\n- Async AI interviews remove scheduling friction and expand participation — especially valuable for reaching professionals who can't commit to live calls\n\n---\n\nReady to run research without the scheduling overhead? Koji's AI interviewer conducts async voice, text and phone interviews automatically. You do not need your own participants: if you have a list, Koji interviews it, and if you do not, Koji recruits the people for you with a credit quote you approve before launch. Send a link, get structured insights, no moderator required. [Try Koji free](/docs/getting-started).\n\n## Frequently Asked Questions\n\n**How many participants do I need for user research?**\nFor qualitative research, 5–8 participants per distinct user segment typically reach thematic saturation — the point where new interviews stop producing new themes. For broader discovery or quantitative signals, 15–30 participants provide more robust patterns. The right number depends on your question: focused usability studies need fewer participants than broad market discovery.\n\n**What's the fastest way to recruit user research participants?**\nParticipant panels like User Interviews and Respondent can deliver qualified participants in 24–48 hours. For even faster turnaround, async AI-moderated interviews (like Koji) remove the scheduling bottleneck entirely: participants complete the study on their own schedule, so you can collect responses from 20+ participants in 48 hours without any calendar coordination. Koji also has panel recruitment built in, quoted per respondent in credits from live feasibility, so you can source and interview in one place.\n\n**How much should I pay research participants?**\nThe industry benchmark is approximately $3/minute for moderated sessions — roughly $90–$120 for a 30-minute session with general consumers, and $150–$300+ for senior professionals or hard-to-reach specialists. Amazon gift cards and Visa prepaid cards have the broadest acceptance.\n\n**How do I recruit hard-to-reach B2B participants?**\nStart with your own customer list — warm relationships and existing context make B2B recruitment easier than cold outreach. For outside your customer base, describe the audience in Koji and approve the live per-respondent quote in credits, or run LinkedIn outreach with a clear value proposition and strong incentive. Niche professional Slack communities can reach specialists that general panels don't cover. Expect B2B recruitment to take 2–4 weeks.\n\n**How do I reduce research participant no-shows?**\nThe industry average no-show rate is 11% (NNGroup). Over-recruit by 20%, send reminder sequences (24 hours before and 1 hour before), and use SMS reminders where possible. Async research formats eliminate no-shows entirely since participants complete on their own schedule.\n\n**What's the difference between recruitment and screening?**\nRecruitment is finding potential participants through channels (panels, customer lists, social outreach). Screening is filtering those candidates to confirm they match your participant criteria before committing research time to them. Both are necessary: great screening applied to the wrong recruitment channel still produces poor results.","category":"Tutorial","lastModified":"2026-10-29T15:33:32.445307+00:00","metaTitle":"How to Recruit User Research Participants (2026)","metaDescription":"The complete guide to recruiting user research participants: screeners, channels, incentives, no-show management, and async AI alternatives. Updated for 2026.","keywords":["how to recruit user research participants","user research recruitment","participant recruitment ux research","research screener","user interview recruitment","research incentives","ux research participants"],"aiSummary":"Recruiting the right participants is the highest-leverage decision in any research project. This guide covers every stage of the recruitment process — defining participant profiles, writing screeners, choosing recruitment channels (panels, customer lists, social outreach, in-product), setting incentives, managing no-shows, and how async AI interviews reduce scheduling friction.","aiKeywords":["participant recruitment","screener design","research incentives","user research","qualitative research","research operations","no-shows","async interviews","panel recruitment"],"aiContentType":"tutorial","faqItems":[{"answer":"For qualitative research, 5–8 participants per distinct user segment typically reach thematic saturation. For broader discovery, 15–30 participants provide more robust patterns. The right number depends on your research question.","question":"How many participants do I need for user research?"},{"answer":"Participant panels like User Interviews and Respondent can deliver qualified participants in 24–48 hours. Koji has panel recruitment built in, quoted per respondent in credits and approved before launch. Its async AI-moderated interviews remove the scheduling bottleneck entirely: participants complete the study on their own schedule, enabling 20+ responses in 48 hours.","question":"What is the fastest way to recruit user research participants?"},{"answer":"The benchmark is approximately $3/minute for moderated sessions — roughly $90–$120 for a 30-minute consumer session and $150–$300+ for senior professionals or hard-to-reach specialists.","question":"How much should I pay research participants?"},{"answer":"Over-recruit by 20%, send confirmation sequences 24 hours and 1 hour before each session, and use SMS reminders. The industry average no-show rate is 11% (NNGroup). Async research formats eliminate no-shows entirely since participants complete on their own schedule.","question":"How do I reduce research participant no-shows?"},{"answer":"Recruitment is finding potential participants through channels like panels, customer lists, or social outreach. Screening is filtering those candidates to confirm they match your criteria before committing research time to them. Both are necessary for good research quality.","question":"What is the difference between recruitment and screening?"},{"answer":"Start with your own customer list, then either describe the audience in Koji and approve a live per-respondent quote in credits, or run LinkedIn outreach with a clear value proposition and strong incentive. Niche professional Slack communities can reach specialists that general panels miss. Expect B2B recruitment to take 2–4 weeks.","question":"How do I recruit hard-to-reach B2B participants?"}],"relatedTopics":["User Research","Research Operations","Participant Recruitment","Qualitative Research","Research Methods","UX Research","Research Incentives"]},{"type":"blog","id":"0dc3325b-5775-48d6-a274-9bdf9ee45a7f","slug":"koji-vs-typeform-2026","title":"Koji vs Typeform: Which Tool Actually Uncovers Why Users Behave the Way They Do? (2026)","url":"https://www.koji.so/blog/koji-vs-typeform-2026","summary":"Comprehensive comparison of Koji and Typeform for user research in 2026, explaining why AI-moderated interviews deliver deeper qualitative insights than static survey forms, with pricing and use case analysis.","content":"Choosing between Koji and Typeform for user research? Here is the honest comparison you need.\n\nTypeform is one of the world's most recognized form builders — its conversational, one-question-at-a-time design produces beautiful surveys with high completion rates. But in 2026, the gap between \"collecting survey responses\" and \"understanding your users\" has never been wider. Koji is an AI-native research platform that conducts real interviews, probes automatically, and delivers synthesized insights — without any researcher on your team.\n\n**The short answer:** If you need to know *what* users think, Typeform is fast and polished. If you need to understand *why* they think it, Koji is built for that job.\n\n## Quick Comparison\n\n| Feature | Koji | Typeform |\n|---|---|---|\n| Interview modality | AI voice + text interviews | Static forms |\n| Follow-up probing | ✅ Real-time adaptive AI | ❌ None |\n| Qualitative analysis | ✅ Automatic thematic synthesis | ❌ Manual CSV export |\n| Response limits | Limited only by credits | 10–10,000/month |\n| Moderator bias | ✅ Eliminated | N/A |\n| One-click reports | ✅ Built-in | ❌ |\n| Best for | Deep qualitative research | Surveys, lead gen, NPS |\n| Research expertise needed | None | Low |\n| Starting price | Free: 10 credits at signup, then from €1 per qualified interview | Free (10 responses/month) |\n\n## What Is Typeform?\n\nTypeform is a form builder with a conversational interface. It presents questions one at a time in a visually clean layout, producing higher completion rates than traditional multi-question forms. It supports conditional logic, file uploads, rating scales, and a large library of integrations.\n\nTypeform is genuinely excellent for:\n- Lead generation and contact forms\n- NPS and CSAT surveys\n- Event registration and quizzes\n- Customer onboarding questionnaires\n- Simple feedback collection at scale\n\nIt is not built for qualitative research. That distinction matters enormously when you are making product decisions.\n\n## Typeform's Limitations for User Research\n\n### No Dynamic Probing\n\nTypeform's logic is configured in advance. If a participant gives an unexpected answer — something fascinating, something that opens a new line of inquiry — Typeform moves to the next pre-written question regardless. There is no mechanism for the tool to say \"tell me more about that.\"\n\nIn qualitative research, the most valuable insights almost always come from adaptive follow-up questions. A 2025 study on AI-moderated interviews found that **over 70% of breakthrough insights emerged from follow-up probing** — moments where the moderator noticed something interesting and pursued it. Typeform cannot follow threads.\n\n### Text Responses Pile Up Unanalyzed\n\nTypeform collects open-ended responses, but provides no tooling to analyze them at meaningful scale. The output is a CSV column of raw text. Extracting themes requires a separate analysis tool — NVivo, MAXQDA, or a manual spreadsheet — adding cost and workflow fragmentation that most teams simply skip.\n\nThis means the qualitative layer teams hope to capture in open-ended questions often disappears into a data export that nobody reads. The numbers sit in a spreadsheet; the story behind them is lost.\n\n### Response Limits Scale Against You\n\nTypeform's free plan allows just 10 responses per month across all forms. The Business plan at $99/month billed annually caps responses at 10,000 per month. For research programs that run regular cycles, these limits become a constraint quickly. And none of Typeform's plans include qualitative analysis — that remains a separate budget line entirely.\n\nA 2026 pricing review of Typeform noted that many teams outgrow their response limits unexpectedly, triggering costly overages or forced upgrades mid-quarter. The pricing model penalizes research scale.\n\n### No Voice Research Capability\n\nHuman communication carries information beyond words. Tone, hesitation, emotional register, and pacing all signal meaning that text responses miss entirely. Typeform has no native voice interview capability. All data collected is text-only.\n\nFor research questions that require emotional depth — \"how do you feel about your current workflow?\", \"walk me through a recent purchase decision\" — text forms are an inherently limited instrument.\n\n## What Koji Does Differently\n\nKoji is an end-to-end AI research platform. You do not need an audience of your own: if you have a customer list, Koji interviews it, and if you do not, Koji recruits the participants for you and quotes the cost per respondent in credits before anything launches. It does not compete with Typeform on form design — it competes on insight quality.\n\n### AI Voice Interviews with Real-Time Probing\n\nKoji's AI conducts live voice conversations with participants. It listens actively, and when a participant says something unexpected or revealing, Koji probes. It follows interesting threads, asks clarifying questions, and surfaces the kind of nuanced qualitative data that static forms never capture.\n\nEvery conversation adapts to the participant — the way a skilled human researcher would, but available 24/7, in any timezone, at any scale.\n\n### Automatic Thematic Analysis\n\nWhen interviews complete, Koji analyzes every conversation simultaneously — clustering themes, surfacing key quotes, and identifying patterns across all respondents. What would take a research team days of manual coding is delivered in minutes.\n\nNo spreadsheets. No affinity diagrams. No analyst required.\n\n### No Moderator Bias\n\nHuman interviewers, however experienced, introduce bias. They react differently to different answers, signal approval or skepticism, and inadvertently steer responses. Koji's AI maintains perfectly consistent, neutral moderation across every session — delivering cleaner, more comparable qualitative data.\n\nAccording to a 2025 report on research methodology, **moderator effects can account for up to 25% variance in qualitative interview outcomes**. Eliminating moderator bias is not a minor optimization — it is fundamental to data quality.\n\n### One-Click Reports\n\nKoji generates shareable insight reports from raw interviews automatically. The full research cycle — from question to stakeholder-ready report — completes in hours. Product managers can run a study on Monday and present findings on Tuesday.\n\n### Customizable AI Consultant Personas\n\nKoji allows teams to configure their AI interviewer's tone, focus area, and expertise domain. A B2B SaaS team can configure a consultative, technical moderator. A consumer brand can configure a warm, empathetic conversation style. This customization makes every study feel purpose-built rather than generic.\n\n## Where Typeform Wins\n\nTypeform remains the right tool when you need:\n- Quantitative data at high volume with low analysis overhead\n- Beautiful, branded forms with consistently high completion rates\n- Simple NPS, CSAT, or satisfaction data points\n- Lead capture and registration workflows\n- Logic-driven quizzes or assessments\n\nIf you already know the question and just need a number or a structured response, Typeform delivers it cleanly and efficiently.\n\n## When Koji Is the Right Call\n\nChoose Koji when your research question begins with *why* or *how*:\n- Why are users churning after the first month?\n- How do customers describe the job your product does for them?\n- What is blocking trial users from converting?\n- How does your product compare to competitors in users' own words?\n- Why did customers choose you over the alternatives?\n\nAny time you need qualitative depth — not just data — Koji is built for that work.\n\n## Pricing Comparison\n\n**Typeform pricing (2026):**\n- Free: 10 responses/month\n- Basic: ~$29/month (100 responses, billed annually)\n- Plus: ~$59/month (1,000 responses, billed annually)\n- Business: ~$99/month (10,000 responses, billed annually)\n- Enterprise: custom pricing\n\n**Koji pricing:** interviews start as low as €1 per qualified interview and €3 per qualified voice interview.\n- Free: 10 credits at signup, no card required, one active study at a time\n- Pay as you go: no subscription, no seats, no minimums\n- You pay for what worked: a conversation that scores below 3 out of 5 is free and never charged\n- Plans: volume pricing and plans are there when you want them\n\nYou are never tied to a subscription. There is no overage. You pay as you go, and you only ever pay for the conversations that bring insights.\n\nFor teams running real qualitative research — even a single study per month — Koji's model delivers substantially more value per dollar spent, because every interview includes automatic analysis and report generation. Typeform's equivalent tier includes none of that.\n\n## Real Teams Are Making the Switch\n\nProduct teams using Koji report a 10x reduction in time-to-insight compared to traditional qualitative methods. Founders are validating new product ideas in hours instead of weeks. Research programs that previously ran 2–3 studies per quarter are now running weekly discovery cycles.\n\nThe key insight: Koji does not replace your research program — it makes one possible for teams that previously could not afford or staff one.\n\n## The Verdict\n\nTypeform is excellent software. It is not research software.\n\nIf your goal is structured data collection at scale, Typeform is a proven, polished choice. But if your goal is genuine user understanding — rich, qualitative, in their own words — you need a tool built specifically for that job.\n\nKoji is the AI-native research platform that makes deep qualitative interviews accessible to any team, with no research expertise, no scheduling overhead, and no analysis backlog. It is the modern answer to a question Typeform was never designed to answer: *why* do users behave the way they do?\n\n## Start Your First Koji Study Today\n\nKoji is free to try. Set up your first AI-moderated interview study in minutes and have insights ready the same day — no research background required. [Start free at koji.so](https://koji.so)\n\n*Last verified: April 2026*\n\n## Frequently Asked Questions\n\n**Q: Can Typeform be used for qualitative research?**\nA: Typeform can collect open-ended text responses, but it has no probing, follow-up, or analysis capabilities. For genuine qualitative research at any meaningful depth, a dedicated platform like Koji is far more effective.\n\n**Q: Is Koji more expensive than Typeform?**\nA: It depends on volume. Koji interviews run as low as €1 per qualified interview and €3 per qualified voice interview, with no subscription needed. A 20-interview voice study starts from about €60 and includes automatic analysis. Typeform's Business plan is $99/month but includes no analysis tooling, so the effective cost for qualitative research is significantly higher with Typeform once you account for the analysis work required.\n\n**Q: Can Koji and Typeform be used together?**\nA: Yes. Many teams use Typeform for quantitative surveys and Koji for qualitative discovery interviews. They serve different research needs and work well in combination.\n\n**Q: Does Koji support text interviews as well as voice?**\nA: Yes. Koji supports both voice and text interview modalities, so participants can choose their preferred format based on context and availability.\n\n**Q: Which is better for NPS surveys?**\nA: Typeform is well-suited for NPS score collection. Koji is better suited for follow-up qualitative research to understand *why* NPS scores are changing and what would move them.\n\n**Q: How long does it take to set up a Koji study?**\nA: Most teams set up their first Koji study in under 15 minutes. The AI guides you through study design, question crafting, and configuration — no research expertise required.","category":"Research","lastModified":"2026-10-29T15:33:31.223155+00:00","metaTitle":"Koji vs Typeform (2026): AI Interviews vs Survey Forms — Which Is Right for Research?","metaDescription":"Typeform builds beautiful forms. Koji conducts real AI interviews. Compare features, pricing, qualitative depth, and use cases to find the right research tool for your team in 2026.","keywords":["koji vs typeform","typeform alternative","typeform for user research","ai interviews vs surveys","qualitative research tools 2026","typeform limitations"],"aiSummary":"Comprehensive comparison of Koji and Typeform for user research in 2026, explaining why AI-moderated interviews deliver deeper qualitative insights than static survey forms, with pricing and use case analysis.","aiKeywords":["Koji vs Typeform","AI interviews","user research tools","qualitative research","survey forms","Typeform alternative","AI moderated interviews"],"aiContentType":"comparison","faqItems":[{"answer":"Typeform can collect open-ended text responses, but it has no probing, follow-up, or analysis capabilities. For genuine qualitative research at any meaningful depth, a dedicated platform like Koji is far more effective.","question":"Can Typeform be used for qualitative research?"},{"answer":"It depends on volume. Koji interviews run as low as €1 per qualified interview and €3 per qualified voice interview, with no subscription needed. A 20-interview voice study starts from about €60 and includes automatic analysis. Typeform's Business plan is $99/month but includes no analysis tooling, so the effective cost for qualitative research is significantly higher with Typeform once you factor in the analysis work required.","question":"Is Koji more expensive than Typeform?"},{"answer":"Yes. Many teams use Typeform for quantitative surveys and Koji for qualitative discovery interviews. They serve different research needs and work well in combination.","question":"Can Koji and Typeform be used together?"},{"answer":"Yes. Koji supports both voice and text interview modalities, so participants can choose their preferred format.","question":"Does Koji support text interviews as well as voice?"},{"answer":"Typeform is well-suited for NPS score collection. Koji is better for follow-up qualitative research to understand why NPS scores are changing.","question":"Which is better for NPS surveys?"},{"answer":"Most teams set up their first Koji study in under 15 minutes. The AI guides you through study design and question crafting with no research expertise required.","question":"How long does it take to set up a Koji study?"}],"relatedTopics":["User Research Tools","AI Interviews","Survey Tools","Qualitative Research","Typeform Alternative","Product Discovery"]},{"type":"documentation","id":"3476cdc0-a926-4751-8895-b8aeded3ca8f","slug":"crossing-the-chasm-guide","title":"Crossing the Chasm: The Technology Adoption Lifecycle for Product Teams","url":"https://www.koji.so/docs/crossing-the-chasm-guide","summary":"Crossing the Chasm, from Geoffrey Moore's 1991 book, describes the make-or-break gap between a technology product's early adopters (13.5% of the market, visionaries who buy on vision) and the early majority (34%, pragmatists who buy on proof and references). Built on Everett Rogers's Diffusion of Innovations, the technology adoption lifecycle has five segments: innovators (2.5%), early adopters (13.5%), early majority (34%), late majority (34%), and laggards (16%). The chasm exists because pragmatists only trust references from other pragmatists, creating a circular dependency. Moore's solution is the beachhead strategy: dominate one narrow niche completely to earn the references needed to reach the mainstream. Every step depends on customer research, which AI-native platforms like Koji accelerate through AI-moderated interviews and automatic thematic analysis.","content":"## What Is \"Crossing the Chasm\"? (Answer First)\n\n**\"Crossing the Chasm\" describes the make-or-break gap between a technology product's early adopters and its mainstream market.** In his 1991 book, Geoffrey Moore argued that most disruptive products die in this gap — the *chasm* — because the tactics that win visionary early adopters completely fail with pragmatic mainstream buyers. Crossing it requires dominating one narrow niche so completely that pragmatists finally feel safe adopting.\n\nThe framework builds on Everett Rogers's **Diffusion of Innovations** theory, which segments any market into five groups based on how readily they adopt something new. Moore's insight was that these groups aren't a smooth curve — there are cracks between them, and one crack is a canyon.\n\n> \"The point of greatest peril in the development of a high-tech market lies in making the transition from an early market dominated by a few visionary customers to a mainstream market dominated by a large block of pragmatists.\"\n> — Geoffrey A. Moore, *Crossing the Chasm*\n\n---\n\n## The Technology Adoption Lifecycle: Five Segments\n\nRogers modeled adoption as a bell curve. Each segment buys for fundamentally different reasons:\n\n![Table of the adoption lifecycle: innovators 2.5%, early adopters 13.5%, early majority 34%, late majority 34% and laggards 16%, plus what each wants.](/visuals/technology-adoption-lifecycle-five-segments-table/technology-adoption-lifecycle-five-segments-table.png \"The chasm sits between the 13.5% who buy vision and the 34% who want proof. · Source: Everett Rogers, Diffusion of Innovations; Geoffrey Moore, Crossing the Chasm\")\n\n| Segment | % of Market | Psychology | What They Want |\n|---|---|---|---|\n| **Innovators** (Techies) | 2.5% | Love technology for its own sake | To try the new thing first |\n| **Early Adopters** (Visionaries) | 13.5% | Seek competitive advantage | A strategic breakthrough; will tolerate rough edges |\n| **Early Majority** (Pragmatists) | 34% | Risk-averse, want proof | A reliable, proven solution that fits existing systems |\n| **Late Majority** (Conservatives) | 34% | Skeptical, price-sensitive | A safe, cheap, standard commodity |\n| **Laggards** (Skeptics) | 16% | Resist change | To avoid the technology entirely |\n\n**The Chasm sits between Early Adopters (13.5%) and the Early Majority (34%)** — between the 16% who buy vision and the 84% who buy proof.\n\n---\n\n## Why the Chasm Exists\n\nThe chasm is a psychology problem, not a product problem. Early adopters and the early majority want opposite things:\n\n- **Early adopters are visionaries.** They buy an early, incomplete product *because* it gives them a first-mover advantage. They expect to do integration work themselves. A reference from another early adopter is irrelevant — they want to be first, not to follow.\n- **The early majority are pragmatists.** They will not buy on vision. They want references from *people like them*, proven ROI, a complete \"whole product,\" and an established ecosystem. They ask: \"Who else in my industry is using this successfully?\"\n\nHere is the trap: **the only references a pragmatist trusts are other pragmatists — but you can't win pragmatists without references.** The visionary customers you already have don't count, because pragmatists distrust visionaries. This circular dependency is what kills products in the chasm. Startups mistake early-adopter traction for product-market fit, scale their spend, and fall in.\n\nThis is a core reason new products fail at rates commonly cited as high as 95% (a figure associated with Clayton Christensen's work, though the exact number is debated, with peer-reviewed studies often placing failure between 40% and 90% depending on category). Many of those \"failures\" are products that won the early market and never crossed.\n\n---\n\n## How to Cross the Chasm: The Beachhead Strategy\n\nMoore's prescription is counterintuitive: **narrow your focus dramatically.** Instead of chasing the whole mainstream market, dominate one small, specific niche until you own it.\n\n![Process showing Moore's four beachhead steps for crossing the chasm: target one segment, build the whole product, give a reason to buy, win references.](/visuals/crossing-the-chasm-beachhead-strategy-steps/crossing-the-chasm-beachhead-strategy-steps.png \"Narrow beats broad: win one niche completely before you chase the mainstream. · Source: Geoffrey Moore, Crossing the Chasm\")\n\n1. **Target a single beachhead segment.** Pick one narrowly defined group of pragmatists with a painful, urgent problem your product solves completely. Moore's D-Day analogy: concentrate all forces on one beach, not the whole coastline.\n2. **Build the \"whole product.\"** Pragmatists buy the complete solution — product plus integrations, support, documentation, and proof — not the core technology alone. Fill every gap for that one segment.\n3. **Create a compelling reason to buy.** For pragmatists the pain must outweigh the risk of change. Quantify the ROI.\n4. **Win references and word of mouth.** Total dominance of a small niche produces the references and case studies that let you tip into it — and then expand to adjacent segments (\"the bowling pin\" strategy).\n\nThe strategic error is trying to be a little bit relevant to everyone. In the chasm, a big fish in a small pond beats a small fish in the ocean every time.\n\n---\n\n## Where Customer Research Decides Whether You Cross\n\nEvery step above is a research problem. Crossing the chasm is fundamentally about *understanding pragmatists well enough to earn their trust* — and that understanding comes from talking to customers.\n\n- **Choosing the beachhead** requires segmentation research to find the niche with the most acute, urgent pain and the clearest willingness to pay.\n- **Defining the whole product** requires jobs-to-be-done and discovery interviews to learn what \"complete\" means to that segment.\n- **Crafting the compelling reason to buy** requires understanding the pragmatist's risk calculus and quantifying their cost of the status quo.\n- **Timing the leap** requires distinguishing real early-majority pull from lingering early-adopter enthusiasm — the single most expensive misread in the framework.\n\n> \"Visionaries are the early market's deep pockets... pragmatists are the mainstream market's. The problem is, they don't talk to each other.\"\n> — Geoffrey A. Moore\n\n---\n\n## The Modern Approach: Chasm Research with AI\n\nThe classic obstacle to this research is bandwidth. Interviewing enough pragmatists across multiple candidate segments to pick the right beachhead — and doing it fast enough to act — used to take a dedicated research team months. **Koji** removes that constraint.\n\n- **Segment discovery at scale.** Run AI-moderated interviews across several candidate segments simultaneously. Koji's AI moderator probes each participant on their pain, current workarounds, and buying criteria, then automatically clusters the themes so you can see which niche has the sharpest, most urgent problem — in days, not quarters.\n- **Whole-product mapping.** Use Koji's [structured questions](/docs/structured-questions-guide) — combining ranking and multiple_choice with open_ended follow-ups — to quantify exactly which capabilities, integrations, and proof points pragmatists require before they'll switch. All 6 question types (open_ended, scale, single_choice, multiple_choice, ranking, yes_no) live inside one conversational interview.\n- **Reason-to-buy validation.** Configure a custom AI consultant on a JTBD or discovery brief to surface each segment's cost-of-status-quo and decision triggers, with quality-scored, real-time reporting.\n- **Reference harvesting.** As your beachhead customers succeed, run automated win/loss and testimonial interviews to generate the pragmatist-to-pragmatist references that actually move the mainstream market.\n\nYou don't need a PhD in market research or a six-month study to find your beachhead. AI-native research turns \"who exactly should we sell to next, and what will make them trust us?\" into a question you can answer this week.\n\n---\n\n## Signs You Are Approaching the Chasm\n\nWatch for these warning signs that early-market momentum is fading and the chasm is near:\n\n- **Deals stall on \"who else uses this?\"** Prospects stop asking about vision and start asking about references from their own industry — the pragmatist tell.\n- **Sales cycles lengthen and require more proof.** Buyers want pilots, security reviews, and ROI models the visionaries never asked for.\n- **Your early adopters can't vouch to the mainstream.** The enthusiastic logos you won don't move pragmatist prospects, because pragmatists distrust visionaries.\n- **Growth plateaus despite more spend.** You pour budget into a market that has quietly changed underneath you.\n\nThe fatal mistake is reading early-adopter enthusiasm as broad demand and scaling headcount and spend to match. When the pragmatists don't follow, the burn rate that felt justified becomes the thing that sinks the company mid-chasm.\n\n## A Beachhead in Practice\n\nConsider a workflow-automation startup with scattered early adopters across a dozen industries. Instead of marketing to all of them, they interview prospects in each and discover that *mid-market insurance claims teams* have the most acute, urgent, expensive pain. They go all-in: build the integrations, compliance, and templates that segment specifically needs, land three lighthouse accounts, and turn those into case studies. Now every other insurance claims team hears \"your peers cut processing time 40%\" — a pragmatist-to-pragmatist reference. From that beachhead they expand to adjacent claims-heavy verticals. That is crossing the chasm: not a bigger net, but a beachhead earned through research.\n\n## After the Chasm: The Bowling Pin and the Tornado\n\nCrossing the chasm is the beginning, not the end. Moore describes what follows with two more metaphors. The **bowling pin** strategy: once you own your beachhead, knock into adjacent segments — each conquered niche gives you references and a whole product that make the next one easier, like pins falling into pins. Then comes the **tornado**: the moment the broad early majority adopts en masse and demand explodes, rewarding whoever has become the safe, standard choice. The teams that win the tornado are the ones that used the beachhead phase to build the references, reliability, and category leadership pragmatists demand — all of it grounded in continuous customer research about what \"safe\" and \"standard\" mean to the next segment.\n\n## Related Resources\n\n- [How to Find Early Adopters for Your Startup](/docs/how-to-find-early-adopters)\n- [Ideal Customer Profile (ICP): How to Define Yours](/docs/ideal-customer-profile-icp)\n- [Customer Segmentation Research Interviews](/docs/customer-segmentation-research-interviews)\n- [Product-Market Fit Interviews](/docs/product-market-fit-interviews)\n- [Jobs to Be Done Framework](/docs/jobs-to-be-done-framework)\n- [Structured Questions: 6 Question Types for Richer Research](/docs/structured-questions-guide)","category":"frameworks","lastModified":"2026-10-29T15:32:50.671358+00:00","metaTitle":"Crossing the Chasm Explained: Technology Adoption Lifecycle (2026)","metaDescription":"Crossing the Chasm explained: Geoffrey Moore's technology adoption lifecycle, the five adopter segments and percentages, why the chasm exists, and how customer research and AI interviews help product teams cross it.","keywords":["crossing the chasm","technology adoption lifecycle","geoffrey moore","diffusion of innovations","early adopters vs early majority","technology adoption curve","beachhead strategy","adopter categories","chasm marketing"],"aiSummary":"Crossing the Chasm, from Geoffrey Moore's 1991 book, describes the make-or-break gap between a technology product's early adopters (13.5% of the market, visionaries who buy on vision) and the early majority (34%, pragmatists who buy on proof and references). Built on Everett Rogers's Diffusion of Innovations, the technology adoption lifecycle has five segments: innovators (2.5%), early adopters (13.5%), early majority (34%), late majority (34%), and laggards (16%). The chasm exists because pragmatists only trust references from other pragmatists, creating a circular dependency. Moore's solution is the beachhead strategy: dominate one narrow niche completely to earn the references needed to reach the mainstream. Every step depends on customer research, which AI-native platforms like Koji accelerate through AI-moderated interviews and automatic thematic analysis.","aiPrerequisites":["Basic familiarity with go-to-market and product strategy"],"aiLearningOutcomes":["Explain the technology adoption lifecycle and its five segments","Describe why the chasm exists between early adopters and the early majority","Apply the beachhead strategy to cross the chasm","Connect each chasm-crossing step to specific customer research","Use AI-moderated interviews to identify a beachhead segment and map the whole product"],"aiDifficulty":"intermediate","aiEstimatedTime":"15 minutes"},{"type":"documentation","id":"268df3eb-2fd5-429e-8d19-c3be250149bb","slug":"usability-metrics-guide","title":"Usability Metrics: Task Success Rate, Time on Task, and Error Rate Explained","url":"https://www.koji.so/docs/usability-metrics-guide","summary":"A definitive reference on the three core usability metrics — task success rate (effectiveness), time on task (efficiency), and error rate (accuracy) — with formulas, the 78% MeasuringU completion benchmark, sample-size guidance, and how Koji captures all three automatically via structured questions and an AI moderator.","content":"## What are the core usability metrics?\n\n**The three core usability metrics are task success rate (effectiveness), time on task (efficiency), and error rate (accuracy).** Together they answer the only questions that matter in a usability test: Can users finish the job? How long does it take them? And how many mistakes do they make along the way? Every other quantitative usability measure — completion confidence, lostness, task-level satisfaction — is a refinement of these three.\n\nThe Nielsen Norman Group calls success rate \"the simplest usability metric\" precisely because it is the bottom line of usability: if users cannot complete what they came to do, nothing else about the interface matters. The strength of these three metrics is that they are objective and behavioral. Unlike attitudinal scores such as NPS or [SUS](/docs/system-usability-scale-guide), they record what users *actually did*, not what they later said they felt.\n\nThis guide defines each metric, gives you the formulas and the published industry benchmarks, explains how many participants you need, and shows how an AI-native platform like Koji captures all three automatically — turning a multi-day analysis grind into a real-time dashboard.\n\n## Metric 1: Task success rate (effectiveness)\n\nTask success rate is the percentage of participants who complete a task successfully out of everyone who attempted it. High task success is not automatically good news: a manipulative flow can score well precisely because it steers people, which is why [dark patterns testing](/docs/dark-patterns-user-testing) measures what users believed alongside what they did.\n\n> **Task success rate = (number of successful attempts ÷ total attempts) × 100**\n\nIf 17 of 20 participants successfully add an item to their cart and reach checkout, your success rate is 85%.\n\n**The benchmark:** In an analysis of 1,189 tasks across 115 usability studies, MeasuringU founder Jeff Sauro found the average task completion rate is **78%**. Most teams treat roughly 78–80% as the dividing line between \"acceptable\" and \"needs work\" for an important task — though success rate is highly sensitive to task difficulty, so the right target is always relative to the task and to your own historical baseline.\n\n**Binary vs. levels of success.** The cleanest version is binary: a participant either completed the task or did not. But many teams record *levels of success* — full success, partial success (completed with significant struggle or workaround), and failure — because a binary view hides the difference between a user who breezed through and one who barely limped to the finish line. Partial successes are often where your richest design insights hide.\n\n**The trap:** success rate alone is misleading. A task can show a 90% success rate while users take three minutes and make two errors getting there. That is why success rate must always be read alongside time and errors.\n\n## Metric 2: Time on task (efficiency)\n\nTime on task measures how long it takes a participant to complete a task, usually reported as the mean or median time in seconds for successful attempts only. (Including failed attempts pollutes the number — a user who gave up after 10 seconds would otherwise look \"efficient.\")\n\nBecause time data is almost always skewed by a few very slow users, the **geometric mean or the median** is the statistically appropriate measure of center for small samples, not the arithmetic mean. Report a measure of spread too — the range or confidence interval — because an average of 45 seconds means something very different if the spread is 40–50 seconds versus 10–120 seconds.\n\n**How to use it:** time on task is most powerful as a *comparative* metric — old design vs. new, your product vs. a competitor, or release over release. An absolute \"good\" time rarely exists in isolation; a 30-second task time is excellent for a complex configuration flow and terrible for a one-click action.\n\n## Metric 3: Error rate (accuracy)\n\nAn error is any unintended action, slip, mistake, or omission a user makes while attempting a task. Error rate is typically expressed as errors per task (the average number of errors across all attempts) or as a defect rate (the percentage of attempts containing at least one error).\n\n> **Errors per task = total errors observed ÷ total attempts**\n\n**The benchmark:** across an analysis of 719 tasks using consumer and business software, Jeff Sauro found an average of **0.7 errors per task**, with roughly **two out of every three users making at least one error**. Errors are far more common than most teams assume — which is exactly why counting them surfaces friction that success rate alone would never reveal.\n\nNot all errors are equal. Classify them by severity (does the error block completion, or merely slow the user down?) and by type (slips, where the user knows the goal but executes the wrong action, vs. mistakes, where the user has the wrong mental model). The pattern in *where* errors cluster is usually more actionable than the raw count.\n\n## Putting the three together\n\nEffectiveness, efficiency, and accuracy form a triangle. A mature usability scorecard reads all three at once:\n\n![Table of the three core usability metrics with benchmarks: task success rate, about 78%; time on task, comparative; error rate, about 0.7 per task.](/visuals/core-usability-metrics-success-time-errors-table/core-usability-metrics-success-time-errors-table.png \"Read all three together: success alone can hide slow, error-ridden tasks. · Source: Jeff Sauro, MeasuringU, benchmarks of 1,189 and 719 tasks\")\n\n| Metric | What it measures | Typical benchmark |\n| --- | --- | --- |\n| Task success rate | Effectiveness — can they finish? | ~78% average (Sauro/MeasuringU) |\n| Time on task | Efficiency — how fast? | Comparative; no universal target |\n| Error rate | Accuracy — how clean? | ~0.7 errors/task; ~2 of 3 users err |\n\nLayer a task-level satisfaction question on top — a single [scale question](/docs/structured-questions-guide) such as \"How easy or difficult was that task?\" — and you capture the user's attitude alongside their behavior. The Nielsen Norman Group repeatedly finds that performance and satisfaction metrics correlate only moderately, so measuring both protects you from shipping something users *can* use but *hate* using.\n\n## How many participants do you need?\n\nFor **qualitative, formative** usability testing — finding problems to fix — five users per round uncovers roughly 85% of issues, the classic Nielsen-Landauer finding. But the moment you want *reliable quantitative metrics* like a stable success rate or time on task, five is far too few: the confidence interval on a metric from five users is enormous.\n\nA practical rule of thumb:\n\n- **5–8 users** — formative testing, finding usability problems (not for reporting precise numbers).\n- **15–20 users** — a reasonably tight success-rate estimate for a single design.\n- **30–50+ users** — benchmark-grade metrics you intend to track over time or quote externally. (See our [usability benchmarking guide](/docs/usability-benchmarking-guide) for the full methodology.)\n\nUse adjusted-Wald binomial confidence intervals for small-sample completion rates rather than naive percentages — a 4-of-5 success \"80%\" actually carries a confidence interval running from roughly 36% to 98%.\n\n## The modern approach: capturing usability metrics with AI\n\nTraditionally, collecting these three metrics meant scheduling moderated sessions, watching every recording, manually timing each task with a stopwatch, tallying errors by hand, and reconciling notes across a research team. A 20-participant benchmark could swallow a week of analyst time — which is why most teams measured usability once a quarter at best, if at all.\n\nThis is exactly the bottleneck AI-native research platforms remove. **Koji** captures all three core metrics automatically:\n\n- **Task success rate** is recorded directly through structured questions. Frame each task with a `yes_no` or `single_choice` outcome question, and Koji aggregates the completion rate across every respondent in real time — no manual tallying.\n- **Time on task** is timestamped automatically for every session, with the distribution (median, range, outliers) computed and charted as responses arrive.\n- **Error and friction signals** surface through Koji's AI moderator, which probes in the moment (\"What made that step confusing?\") and then clusters the open-ended answers into themed friction findings, so you see *where* and *why* users struggle, not just *that* they did.\n\nKoji supports all six [structured question types](/docs/structured-questions-guide) — `open_ended`, `scale`, `single_choice`, `multiple_choice`, `ranking`, and `yes_no` — which means you can capture a binary success flag, a 1–5 ease rating, and a rich open-ended \"what went wrong\" in a single automated study. Because Koji runs 24/7, you can recruit 30–50 participants for a true quantitative benchmark in days rather than weeks: import your own users so each gets a personalized link, embed the study in the product, or open Panel recruit and Koji finds the users you describe across 130+ countries at a price per completed interview you approve first. Re-run the identical study every release to track the trend line. Teams using AI-assisted research tools consistently report dramatically faster time-to-insight precisely because the counting, timing, and tagging — the slow part — is done the instant the last response lands.\n\nYou do not need a PhD in measurement theory to run a rigorous usability study. Define the tasks, attach the right structured questions, and let the platform handle the statistics.\n\n## Common mistakes to avoid\n\n![Checklist of five usability metrics mistakes to avoid, from timing failed attempts to changing task wording between benchmark rounds.](/visuals/usability-metrics-mistakes-checklist/usability-metrics-mistakes-checklist.png \"Five checks before you report a usability benchmark.\")\n\n1. **Reporting time on task for failed attempts.** Always separate successful and unsuccessful times.\n2. **Using the arithmetic mean on small samples.** Time data is skewed — use the median or geometric mean.\n3. **Quoting a success rate without a confidence interval.** \"80% from five users\" is not a precise number.\n4. **Measuring success but never satisfaction.** A usable-but-frustrating product still loses users.\n5. **Changing the task wording between benchmark rounds.** Consistency is what makes release-over-release comparison valid.\n\n## Related Resources\n\n- [Structured Questions Guide](/docs/structured-questions-guide) — the six question types for capturing success, ease, and friction\n- [Usability Testing: The Complete Guide](/docs/usability-testing-guide) — the end-to-end method these metrics live inside\n- [Usability Benchmarking Guide](/docs/usability-benchmarking-guide) — turning these metrics into a tracked program\n- [System Usability Scale (SUS) Guide](/docs/system-usability-scale-guide) — the standard attitudinal usability score\n- [Customer Effort Score Guide](/docs/customer-effort-score-guide) — measuring perceived ease at the task level\n- [Think-Aloud Protocol](/docs/think-aloud-protocol) — surfacing the *why* behind every error\n- [Just-Noticeable Difference](/docs/just-noticeable-difference-product-changes) — the smallest change a user can actually perceive\n- [Threshold Testing](/docs/threshold-testing-adaptive-staircase-research) — finding the level where a metric starts to register\n- [Slips vs Mistakes: Classifying User Errors](/docs/slips-vs-mistakes-usability-errors) — why the same wrong click needs opposite fixes depending on intention\n- [UMUX-Lite Guide](/docs/umux-lite-guide) — a two-item perceived usability score that tracks SUS\n- [Confidence Intervals for Small-Sample UX Research](/docs/confidence-intervals-small-sample-ux) — how to report success rates, ratings and task times as ranges\n","category":"Research Methods","lastModified":"2026-10-29T15:32:49.179036+00:00","metaTitle":"Usability Metrics: Task Success Rate, Time on Task & Error Rate (2026)","metaDescription":"Master the three core usability metrics — task success rate, time on task, and error rate. Formulas, industry benchmarks (78% avg completion), sample sizes, and how to capture them automatically with AI.","keywords":["usability metrics","task success rate","time on task","error rate","usability testing metrics","task completion rate","usability benchmark","UX metrics","effectiveness efficiency","quantitative usability"],"aiSummary":"A definitive reference on the three core usability metrics — task success rate (effectiveness), time on task (efficiency), and error rate (accuracy) — with formulas, the 78% MeasuringU completion benchmark, sample-size guidance, and how Koji captures all three automatically via structured questions and an AI moderator.","aiPrerequisites":["Familiarity with usability testing concepts","A product or prototype to evaluate","Basic understanding of percentages and averages"],"aiLearningOutcomes":["Define and calculate task success rate, time on task, and error rate","Interpret each metric against published industry benchmarks","Choose the right sample size for qualitative vs quantitative usability studies","Avoid the five most common usability-metric mistakes","Capture all three metrics automatically in Koji using structured questions"],"aiDifficulty":"beginner","aiEstimatedTime":"13 min read"},{"type":"documentation","id":"5a71a601-1235-46ec-9baf-63a143fa03bf","slug":"zendesk-research-integration","title":"Zendesk + Koji: Trigger AI Interviews from Support Tickets and Sync Insights Back to Every Conversation","url":"https://www.koji.so/docs/zendesk-research-integration","summary":"The Koji + Zendesk integration works in both directions and needs no native app. From Zendesk into Koji: a Zendesk trigger emails the requester your study link when a ticket is solved, CSAT is low, or a tag like churn-risk is applied; for ticket-level tracking, the trigger's webhook calls your own service, which starts the interview through Koji's Start Interview endpoint with the ticket ID as respondent.external_id and runs the conversation through the message endpoint. For batch follow-ups, Koji can also email the invitation and a reminder itself on paid plans, with sent, opened and started tracking. From Koji back into Zendesk: Koji has no outbound webhooks, so a scheduled worker polls the read endpoint and, once an interview has ended and its analysis is present, calls the Zendesk API to add an internal note with top themes, sentiment, quality score, and a response link, and applies insight and sentiment tags. Three setup paths exist: a trigger with the study link, batch CSV import with personalised links, or the headless API from your own middleware. Studies use Koji's six structured question types so a single_choice root-cause question maps to your ticket taxonomy. The API, API keys and MCP connector are included on the Interviews and Enterprise plans; outbound webhooks are not available yet.","content":"# Zendesk + Koji: Turn Support Tickets Into AI Interviews That Explain the Root Cause\n\n**Answer first:** A Zendesk ticket tells you a customer had a problem; it rarely tells you *why the problem mattered, what they expected, or whether they are now a flight risk*. Koji closes that gap. The integration works in both directions: (1) a Zendesk **trigger** invites the customer to a Koji AI-moderated interview when a ticket is solved, a CSAT rating comes in low, or a specific tag is applied, and (2) once the interview is analyzed, a small worker posts a concise insight summary back onto the ticket as an **internal note and tags**: sentiment, top themes, quality score, and a link to the response. No native Zendesk app is required. Zendesk's own triggers handle the invitation, Koji's headless API ties an interview to a ticket ID, and a scheduled worker polling Koji's read endpoint handles the way back, because Koji has no outbound webhooks yet. A first version is live in well under an hour.\n\nThe result is a support org that does not just resolve tickets. It systematically learns from them, at a scale no manual QA or callback program can match.\n\n## Why connect Zendesk and Koji\n\nSupport conversations are the largest, most honest stream of customer signal most companies own, and the least mined. The ticket text captures the *symptom* (\"export is broken\") but almost never the *stakes* (\"I nearly cancelled because I present this to my CEO every Monday\"). Traditional follow-up options are weak: a one-question CSAT survey with no depth, or an agent-initiated callback that does not scale past a handful a week.\n\nKoji's AI moderator interviews the customer conversationally, asks adaptive follow-up questions, and works in text or voice across 30+ languages, so you can follow up on hundreds of tickets a week without adding headcount. Every resolved ticket can become a two-minute \"what really happened\" interview, and every low CSAT score can trigger a structured root-cause conversation instead of a lost customer.\n\n## What flows in each direction\n\n### Zendesk → Koji (invite customers from ticket events)\n- A Zendesk **trigger** fires on an event you choose (ticket solved, CSAT = bad, tag `churn-risk` added, priority = urgent escalated).\n- **Simplest version:** the trigger's email notification to the requester includes your study's interview link. Koji runs and analyzes the interview, but it does not know which ticket it came from, so this suits aggregate learning rather than ticket-level write-back.\n- **Ticket-linked version:** the trigger's webhook calls your own small service, which calls Koji's [Start Interview endpoint](/docs/starting-interviews-via-api) with a `respondent` object: the ticket ID as `external_id`, the requester's name as `display_name`, and anything else useful (ticket tags, locale) in `metadata`. The endpoint returns an `interview_id` and a session token, and your service runs the conversation in your own help-center page or widget through the message endpoint. The study is resolved from the API key, so each key belongs to one study.\n- On the ticket-linked path, Zendesk or your own mailer delivers the invitation. For batch follow-ups, Koji can also send the invitation email itself (see path 2).\n\n### Koji → Zendesk (insights land back on the ticket)\n- Koji has no outbound webhooks, so a scheduled worker polls `GET /api/v1/interviews/{id}` and acts once the interview has ended and `analysis` is present.\n- The response carries the `respondent.external_id` you set at start, so the worker knows which ticket to update.\n- The worker calls the **Zendesk API** to add an internal note to the original ticket with the top themes, sentiment, quality score, and a link to the response in Koji, and to apply tags like `insight:pricing-confusion` or `sentiment:frustrated`.\n- Agents and CX leads see the \"why\" attached to the very ticket that prompted it.\n\n## Ticket events worth interviewing\n- **Solved tickets:** a lightweight \"did we actually fix the underlying job?\" check.\n- **Low CSAT / detractor ratings:** a structured root-cause interview while it is fresh (see [churn interview questions](/docs/churn-interview-questions)).\n- **Churn-risk or cancellation tags:** route straight to a [cancel-flow exit interview](/docs/cancel-flow-exit-interview).\n- **Feature-request tickets:** probe the underlying need, not just the asked-for feature.\n- **Repeated escalations:** understand the cumulative frustration a single ticket hides.\n\n## Setup: three paths\n\n**1. Zendesk trigger with your study link (no code).** Create a trigger whose conditions match the event and whose action emails the requester with your Koji interview link. You get themed, aggregated insight in the Koji report within minutes of setup. There is no per-ticket write-back on this path.\n\n**2. Personalised invite links in batches.** Export the requesters you want to hear from (for example, last week's detractors) and import them into Koji as a CSV. Koji creates a personal interview link for each person, and each response is tied to that person in Koji. Send the links from Zendesk or your mailer, or import the requesters into Participants, save them as an audience and let Koji email the invitation and a reminder from the Self-recruit tab of the study. Each person gets their own link, and you can see who received the email, opened it and started the interview. See [Connect Koji to Zapier](/docs/zapier-research-automation) if you want to automate the export.\n\n**3. Headless API from your own middleware (ticket-linked).** If you already run a support-automation service, have the Zendesk webhook call it, start interviews through Koji's [headless API](/docs/headless-api-overview) with the ticket ID as `external_id`, and poll the read endpoint from your own scheduler; see [webhook setup](/docs/webhook-setup) for the patterns that replace callbacks.\n\n## Design the interview so it maps to your taxonomy\n\nKoji's **structured questions** make the returned data align with how your CX team already thinks. A study can mix six types: `open_ended`, `scale`, `single_choice`, `multiple_choice`, `ranking`, and `yes_no`. A `single_choice` \"which best describes the root cause?\" maps directly to your ticket-reason taxonomy, while `open_ended` probes capture the story. See the [structured questions guide](/docs/structured-questions-guide). Every quantitative answer aggregates into a [real-time report](/docs/real-time-research-insights) you can watch fill in as interviews complete.\n\n## Example: turning CSAT detractors into a save motion\n1. Zendesk trigger: CSAT rating = *bad* → webhook to your service, which starts a Koji interview with the ticket ID as `external_id`.\n2. The customer gets a 3-minute interview: one `single_choice` root-cause question, two `open_ended` probes, one `scale` \"how likely are you to keep using us?\"\n3. Once the analysis lands, your worker reads it back, tags the ticket `sentiment:at-risk`, posts the summary as an internal note, and alerts the account owner in Slack.\n4. CX now has a ranked, themed view of *why* CSAT dips, and a triggered save conversation for every detractor.\n\n## Identity and privacy\nOn the headless path, pass the Zendesk ticket ID (or user ID) as `respondent.external_id`. The same value comes back when you read the interview, so notes land on the right ticket. Keep email addresses out of `metadata` unless you need them. For sensitive topics, use the plain study link instead: Koji does not know who the person is, and you skip the Zendesk write-back, keeping raw transcripts out of the helpdesk.\n\n## Plans and limits\nThe headless API, API keys and the MCP connector are included on the **Interviews** and **Enterprise** plans; on other plans the API answers with a 403 and asks for an Interviews plan or higher. CSV import with personalised links and voice interviews are also Interviews and Enterprise features. Sharing a study link from a Zendesk trigger works on every plan. Recruitment emails sent by Koji are available on the Insights, Interviews and Enterprise plans, each with a monthly email allowance, and you can buy extra emails when you need more. Outbound webhooks are not available yet. Credits cost €1 each: a qualified text interview uses 1 credit and a qualified voice interview uses 3.\n\n## The insight note that lands on the ticket\n\nWhen an interview is analyzed and your worker writes it back, the internal note it posts on the ticket is designed to be skimmable by a busy agent. A typical note includes the overall sentiment, the two or three ranked themes, the answer to each structured question, the interview quality score, and a link to the full response. Tags such as `insight:onboarding-gap` and `sentiment:at-risk` are applied to the ticket, so you can build Zendesk views and reports around them, for example a view of every \"at-risk\" customer surfaced by a follow-up interview this week.\n\n## Common pitfalls (and how to avoid them)\n- **Triggering on every ticket.** Volume kills goodwill. Gate interviews to the moments that carry real signal (bad CSAT, cancellation tags, complex solved issues) and cap frequency per requester.\n- **Asking what the ticket already answered.** Do not re-ask the reported issue. Use the interview to probe stakes, expectations, and alternatives considered, which the ticket never captures.\n- **Letting insights die in a note.** Route the highest-severity findings (negative sentiment plus a high quality score) to Slack or a weekly CX review so they drive action, not just documentation.\n- **Ignoring language.** Koji moderates in 30+ languages; pass the ticket's locale when you start an interview so international customers get a native-language conversation.\n\n## From tickets to a continuous voice-of-customer program\n\nHandled this way, Zendesk stops being only a cost center and becomes your largest research panel. Every week, solved tickets and CSAT detractors feed a themed, quantified stream of root-cause insight, the kind of signal product and CS leaders usually pay agencies for. Because Koji analyzes each interview automatically and aggregates structured answers into a live report, the CX team gets a standing \"why customers contact us\" view without running a single manual study, and on the ticket-linked path every new theme is traceable back to the exact ticket and response that produced it.\n\n## Related Resources\n- [Intercom + Koji](/docs/intercom-research-integration)\n- [Research Automation](/docs/research-automation-webhooks)\n- [Structured Questions in AI Interviews](/docs/structured-questions-guide)\n- [Churn Interview Questions](/docs/churn-interview-questions)\n- [Customer Success Interview Guide](/docs/customer-success-interview-guide)\n- [Understanding Themes & Patterns](/docs/understanding-themes-patterns)","category":"API Reference","lastModified":"2026-10-29T13:18:41.793863+00:00","metaTitle":"Zendesk + Koji Integration: AI Interviews From Support Tickets","metaDescription":"Trigger Koji AI interviews from Zendesk ticket events like solved tickets and low CSAT, then post themes, sentiment, and transcript links back onto the ticket automatically.","keywords":["zendesk koji integration","ai interviews from support tickets","zendesk webhook research","csat root cause interview","support ticket customer research","zendesk trigger interview","helpdesk voice of customer"],"aiSummary":"The Koji + Zendesk integration works in both directions and needs no native app. From Zendesk into Koji: a Zendesk trigger emails the requester your study link when a ticket is solved, CSAT is low, or a tag like churn-risk is applied; for ticket-level tracking, the trigger's webhook calls your own service, which starts the interview through Koji's Start Interview endpoint with the ticket ID as respondent.external_id and runs the conversation through the message endpoint. For batch follow-ups, Koji can also email the invitation and a reminder itself on paid plans, with sent, opened and started tracking. From Koji back into Zendesk: Koji has no outbound webhooks, so a scheduled worker polls the read endpoint and, once an interview has ended and its analysis is present, calls the Zendesk API to add an internal note with top themes, sentiment, quality score, and a response link, and applies insight and sentiment tags. Three setup paths exist: a trigger with the study link, batch CSV import with personalised links, or the headless API from your own middleware. Studies use Koji's six structured question types so a single_choice root-cause question maps to your ticket taxonomy. The API, API keys and MCP connector are included on the Interviews and Enterprise plans; outbound webhooks are not available yet.","aiPrerequisites":["A Koji account on the Interviews or Enterprise plan for the API path (a study link works on any plan)","A Zendesk account with admin access to create triggers and webhooks","Familiarity with webhooks or a small scheduled script"],"aiLearningOutcomes":["Trigger Koji AI interviews from Zendesk ticket events","Post interview insights back onto the originating ticket as notes and tags","Map structured interview answers to your ticket-reason taxonomy","Build a triggered save motion for CSAT detractors"],"aiDifficulty":"intermediate","aiEstimatedTime":"12 min read"},{"type":"documentation","id":"169b1233-92de-47c0-a725-0f77e5031ce2","slug":"translating-research-questions-trapd","title":"Translating Research Questions: Why Back-Translation Fails and What to Do Instead","url":"https://www.koji.so/docs/translating-research-questions-trapd","summary":"Back-translation is a weak check for translated research instruments: Behr (2017) found it produces false alarms and leaves many problems hidden. The established alternative is TRAPD (Translation, Review, Adjudication, Pretesting, Documentation), used by the European Social Survey. Write source questions without idioms, use one term per concept, adapt local examples, pretest in each language and document decisions. Koji runs AI-moderated interviews in 30+ languages from one brief, keeps answers to the same question together across languages, and lets you preview the interview in each language before launch.","content":"**Short answer: do not rely on back-translation to check a translated interview guide or survey. Write the source questions so they can be translated, have the translation reviewed by people who know both the language and the research goal, test it with a few real participants in each language, and write down every decision. Survey methodologists call this team approach TRAPD: Translation, Review, Adjudication, Pretesting and Documentation. Back-translation can still be one input, but on its own it misses real problems and flags false ones.**\n\nIf you run research in more than one language, translation is a measurement problem, not a copy-editing task. A question that means something slightly different in German than in English produces a German answer to a different question. The difference then shows up in your report as a \"market difference\" that is really a wording difference.\n\n## Why multilingual research is the normal case\n\nEven a single-country study is often multilingual. In the United States, the Census Bureau reported that more than 1 in 5 people (22%) aged 5 and older spoke a language other than English at home in 2017 to 2021. Among Spanish speakers in the 2018 to 2022 estimates, 61.0% said they spoke English \"very well\", which means a large minority did not. Interviewing everyone in English quietly drops or distorts the people who would answer more fully in their first language.\n\nCustomers also respond to language directly. CSA Research surveyed 8,709 consumers in 29 countries in 2020 and found that 76% of online shoppers prefer to buy products with information in their native language, and 40% will never buy from websites in other languages. If language shapes whether people buy, it also shapes how they explain why they bought, churned or complained.\n\nLarge cross-national surveys take this seriously. The European Social Survey requires a translation for each language used as a first language by 5% or more of a country's population. Few product teams need that rigour, but the principle transfers: decide in advance which languages your audience actually needs, rather than defaulting to the one your team speaks.\n\n## Why back-translation is not enough\n\nBack-translation means translating your questionnaire into the target language, having a second translator translate it back into the source language without seeing the original, and comparing the two source-language versions. If they match, the translation is assumed to be good.\n\nThe method is popular because it lets a team that does not speak the target language feel in control. The problem is that it checks the wrong thing. Dorothée Behr, a survey translation researcher at GESIS, studied back-translation documentation from the 2012 European Quality of Life Survey and compared it with independent reviews of the actual translations. Her conclusion, published in the _International Journal of Social Research Methodology_ in 2017: \"While back translation can uncover problems, it causes quite a number of false alarms, and even more importantly, many problems remain hidden.\"\n\nThere are three structural reasons:\n\n1. **Literal translations pass easily.** A word-for-word translation back-translates neatly, even when it reads awkwardly or shifts meaning for a native speaker. The translation agency cApStAn gives the example of a question about how successful the police are at catching people who commit \"house burglaries\". Translated literally into Slovak, the term covered houses but not flats, so the question measured something narrower than intended. The back-translation would still say \"house burglaries\".\n2. **Register and tone disappear.** Many languages distinguish a formal and informal \"you\". Both come back as \"you\" in English, so a translation that addresses participants too formally or too casually looks fine on the back-translation sheet.\n3. **It tells you there is a problem, not how to fix it.** Even when back-translation flags a mismatch, it gives the translator no guidance on what the right target-language wording is.\n\n## The TRAPD approach\n\nTRAPD was developed in cross-cultural survey research, associated with Janet Harkness and adopted by the European Social Survey, which has used a team approach to translation since its first round. Here is what each step means, scaled to a product research team.\n\n### T: Translation\n\nTwo people translate independently, or each translates half the instrument. Independent versions surface different interpretations early. In a small team, one professional translator plus one bilingual colleague who knows the product is a workable minimum.\n\n### R: Review\n\nTranslators and a reviewer meet to compare versions line by line and agree on wording. The reviewer should understand what each question is trying to measure, not just the words. This is where you catch questions that are grammatically right and conceptually wrong.\n\n### A: Adjudication\n\nSomeone makes the final call where reviewers disagree. In a product team that is usually the researcher who owns the study, ideally with a native speaker present. The point is that a named person signs off, so disagreements do not stay unresolved.\n\n### P: Pretesting\n\nRun the translated questions with a small number of real participants in each language and listen for hesitation, requests for clarification and answers that do not fit the question. A [cognitive interview](/docs/cognitive-interview-guide) (\"what does this question mean to you, in your own words?\") is the most direct way to check that a translated question is understood as intended.\n\n### D: Documentation\n\nRecord every non-obvious decision: why a term was adapted, which examples were swapped for local ones, what the pretest changed. Six months later, when someone asks why the French numbers differ, this record tells you whether the cause could be wording.\n\n## Write questions that can be translated\n\nMost translation problems are created in the source language. Before anything goes to a translator:\n\n- **Remove idioms and metaphors.** \"Is the tool a no-brainer?\" or \"What moves the needle for you?\" will not survive translation intact.\n- **Use one term per concept.** If the English version says \"workspace\" in one question and \"account\" in the next, translators will pick different words and participants will assume you mean different things.\n- **Avoid double meanings.** \"How often do you use reports?\" could mean reading or creating them. Pick one.\n- **Separate examples from the question.** Examples are often culture-specific (\"for example, at Thanksgiving\"). Mark them as replaceable so translators can choose local equivalents.\n- **Keep scale labels simple.** Labels such as \"somewhat\" and \"fairly\" have no exact equivalents in many languages. Fewer, clearer labels translate more reliably.\n- **Add translator notes.** For each question, write one line on what you are trying to learn. This lets translators prioritise meaning over form.\n\nThese are the same habits that make questions clearer for native speakers. See [How to Write Unbiased Survey Questions](/docs/survey-question-wording-guide) for wording principles that apply in any language.\n\n## Adapt, do not just translate\n\nSome questions need adaptation, not translation. Payment methods, job titles, school systems, public holidays and what counts as a \"small business\" vary between countries. A question can be translated perfectly and still not make sense locally. The review step should ask two questions for each item: is the translation accurate, and does the question still make sense here? If the answer to the second is no, change the question and document why. [Cross-Cultural User Research](/docs/cross-cultural-user-research) covers the cultural side of this in depth, including response style differences that affect rating scales.\n\n## Analysing multilingual results\n\n- **Compare like with like.** Before you report a difference between language groups, check the translations of the questions involved. A difference on one question only is often a wording issue; a consistent difference across related questions is more likely real.\n- **Code in one language, quote in the original.** Agree a shared codebook, then keep original-language quotes alongside translations in the report so native speakers can check the interpretation.\n- **Watch scale use.** Groups can differ in how they use rating scales regardless of what they think. Open follow-up answers help you check whether a lower score reflects a real problem.\n\n## How Koji helps\n\nKoji runs AI-moderated interviews in 30+ languages, by text and, where a voice is available for the language, by voice. You choose which languages a study offers, and participants pick theirs from the list, shown in their own language. The AI interviewer then asks your questions and its follow-ups in that language.\n\nThat changes where the translation work happens, and it is worth being clear about what it does and does not replace.\n\n- **One brief, many languages.** You write the study once. The interviewer delivers it in each enabled language, so you do not maintain separate translated copies of the guide.\n- **Follow-ups in the participant's language.** Translated surveys only translate the fixed questions. In an AI-moderated interview, clarifications and probes also happen in the participant's language, so a participant who misunderstands a question can be guided back to it.\n- **Answers that roll up together.** Every structured question keeps a stable identity across languages, so answers from German, Japanese and Spanish participants to the same question land in the same chart in the report, with thematic analysis across all of them.\n- **Pretest every language before launch.** The Preview tab on a study's Interviews page lets you take your own interview, by text or voice, and pick the language. Previews are free and never appear in the report. This is where you do the R and P of TRAPD: ask a native-speaking colleague to run a preview in their language and flag anything that sounds wrong or means something different.\n\nWhat Koji does not do is replace review and documentation. The interviewer translates your intent as you wrote it, so source questions with idioms or ambiguous terms will still cause problems. Treat the AI's rendering as the first translation in TRAPD, then review, pretest and document as you would for any translation. Large survey programmes are moving in a similar direction: the European Social Survey's draft specification for Round 13 describes a modified TRAPD that adds a machine translation step, with human review still in place.\n\nCompared with a traditional survey tool, where each language version is a separate form you have to keep in sync, you maintain one brief and test each language directly. Compared with hiring moderators for each market, you get consistent delivery in every language without scheduling across time zones. See [Multi-Language User Research](/docs/multilingual-research-guide) for the set-up steps.\n\n## A practical checklist\n\n1. List the languages your audience actually needs.\n2. Clean the source questions: no idioms, one term per concept, replaceable examples, translator notes.\n3. Produce the first translation (human or AI).\n4. Review with a native speaker who understands the research goal.\n5. Have one named person adjudicate disagreements.\n6. Pretest with two or three participants per language, ideally with cognitive interview probes.\n7. Document every adaptation and why it was made.\n8. In analysis, check wording before you report a difference between language groups.\n\n## Key takeaways\n\n- Translation is a measurement step. A wording difference becomes a fake market difference.\n- Back-translation misses real problems and creates false alarms; use it, if at all, as one input.\n- TRAPD (Translation, Review, Adjudication, Pretesting, Documentation) is the established alternative and scales down to small teams.\n- Most translation problems start in the source questions. Write for translation from the start.\n- AI-moderated interviews remove the work of keeping separate versions in sync, but review and pretesting by native speakers still matter.\n\n## Related Resources\n\n- [Structured Questions Guide](/docs/structured-questions-guide)\n- [Multi-Language User Research](/docs/multilingual-research-guide)\n- [Cross-Cultural User Research](/docs/cross-cultural-user-research)\n- [Cognitive Interview Guide](/docs/cognitive-interview-guide)\n- [How to Write Unbiased Survey Questions](/docs/survey-question-wording-guide)\n- [Pilot Study Guide](/docs/pilot-study-user-research-guide)\n\n## Sources\n\n- Behr, D. (2017). Assessing the use of back translation: the shortcomings of back translation as a quality testing method. _International Journal of Social Research Methodology_, 20(6), 573–584.\n- U.S. Census Bureau (2023). Language spoken at home, 2018–2022 American Community Survey 5-year estimates (press release, December 2023).\n- U.S. Census Bureau (2025). 2017–2021 ACS language use tables (press release, June 2025).\n- CSA Research (2020). Can't Read, Won't Buy – B2C (press release, July 2020).\n- European Social Survey. Translation methodology and Round 11 translation guidelines.\n- cApStAn. Why back translation is inadequate to assess quality in translated surveys.\n- Harkness, J. A. (2011). Translation. In _Guidelines for Best Practice in Cross-Cultural Surveys_. Survey Research Center, University of Michigan.\n","category":"Research Methods","lastModified":"2026-10-29T13:18:41.495297+00:00","metaTitle":"Translating Survey and Interview Questions: TRAPD vs Back-Translation","metaDescription":"Back-translation misses real translation problems. Learn the TRAPD method, how to write translatable questions, and how to pretest research in every language.","keywords":["translating survey questions","back translation survey","TRAPD translation","questionnaire translation","translate interview guide","multilingual research","cross-cultural survey translation"],"aiSummary":"Back-translation is a weak check for translated research instruments: Behr (2017) found it produces false alarms and leaves many problems hidden. The established alternative is TRAPD (Translation, Review, Adjudication, Pretesting, Documentation), used by the European Social Survey. Write source questions without idioms, use one term per concept, adapt local examples, pretest in each language and document decisions. Koji runs AI-moderated interviews in 30+ languages from one brief, keeps answers to the same question together across languages, and lets you preview the interview in each language before launch.","aiPrerequisites":["An interview guide or survey you plan to run in more than one language","Access to at least one native speaker per target language for review"],"aiLearningOutcomes":["Explain why back-translation alone is an unreliable quality check","Run a scaled-down TRAPD process in a product research team","Write source questions that translate cleanly","Decide when a question needs adaptation rather than translation","Analyse multilingual results without mistaking wording differences for real ones","Pretest each language of an AI-moderated interview study in Koji"],"aiDifficulty":"intermediate","aiEstimatedTime":"12 min read"},{"type":"documentation","id":"69fbabc9-50c5-46de-bade-b4dea90529bb","slug":"setting-up-voice-interviews","title":"How to Set Up AI Voice Interviews: A Researcher's Complete Guide","url":"https://www.koji.so/docs/setting-up-voice-interviews","summary":"A complete researcher-facing guide to setting up AI voice interview studies in Koji. Covers creating the research brief, enabling voice mode, selecting interview language, configuring the landing page, setting up intake forms, adding context documents, customizing the interview slug, testing before launch, and monitoring responses. Voice interviews produce 3–5x richer qualitative data than typed answers, and platforms like Koji automate the entire workflow from moderation to analysis.","content":"# How to Set Up AI Voice Interviews: A Researcher's Complete Guide\n\nAI-powered voice interviews are the fastest way to run deep qualitative research at scale — no scheduling, no moderator burnout, and no transcript cleanup required. Koji's voice interview mode lets participants speak naturally with an AI interviewer that listens, probes, and adapts in real time, while you review rich transcripts and auto-generated insights in your dashboard.\n\nThis guide walks through everything you need to configure before launching a voice interview study: from research brief setup to interview mode selection, language settings, landing page branding, and quality optimization.\n\n## Why Voice? The Case for AI-Powered Spoken Research\n\nTraditional user research methods require scheduling live sessions, training moderators, and manually transcribing recordings — a process that can take weeks and cost thousands of dollars per research cycle. Voice interviews with AI eliminate nearly all of that friction.\n\nWhen participants speak rather than type, they express themselves more naturally, use richer language, and tend to share more emotional and contextual detail. Research shows that spoken responses average 3–5x more words than typed answers to the same question. Combined with Koji's AI interviewer — which probes for depth, asks follow-up questions, and guides the conversation across all your research topics — voice mode produces the kind of nuanced qualitative data that typically requires a skilled human moderator.\n\nThe key difference: Koji's AI runs 24/7, handles unlimited concurrent interviews, and never gets tired or introduces moderator bias. You can launch a voice study on Monday and have 50 rich, analyzed interviews by Wednesday.\n\n## Step 1: Create Your Study and Define the Research Brief\n\nBefore configuring voice settings, you need a clear research brief. Koji generates this automatically when you describe your research goal in natural language, but you can also edit it manually in the canvas editor.\n\nA strong brief for a voice study includes:\n\n**Problem Context**\n- What question are you trying to answer?\n- What decision will this research inform?\n- What is your current hypothesis?\n- What does it cost your team if this remains unanswered?\n\n**Target Participant**\n- Who should participate? (role, behavior, relevant experience)\n- What is the screening question that qualifies them?\n\n**Methodology**\n\nKoji supports several built-in methodologies that shape how the AI probes follow-up questions:\n\n- **The Mom Test** — great for validating ideas without getting false positives; focuses on past behavior rather than hypotheticals\n- **Jobs to be Done (JTBD)** — ideal for understanding switching behavior and purchase triggers\n- **Customer Discovery** — broad, exploratory research for early-stage products\n- **Exploratory** — open-ended discovery when you don't know what you don't know\n- **Lead Magnet** — designed to produce chartable, shareable insights for public reports\n\nThe methodology shapes how deeply the AI probes and what kinds of follow-up questions it asks. See [Choosing a Methodology](/docs/choosing-a-methodology) for guidance on which to use for your research goals.\n\n## Step 2: Enable Voice Mode\n\nOnce your brief is ready, go to your study's **Customize** tab to configure the interview experience.\n\nUnder **Interaction Mode**, you will find three settings:\n\n- **Enable Voice** — toggle to allow participants to choose voice mode\n- **Enable Text** — toggle to allow text mode as an option or fallback\n- **Default Mode** — set whether the landing page defaults to voice or text\n\n**Best practices:**\n- If voice data quality is a priority, set Default Mode to **Voice** and keep text enabled as a fallback. Participants who cannot use a microphone will not be blocked.\n- If you want exclusively voice data, you can disable text mode entirely — but this will reduce completion rates for participants in noisy environments or on mobile devices.\n- For most studies, leaving both enabled with voice as default produces the best combination of data quality and completion rate.\n\n## Step 3: Configure Your Interview Language\n\nKoji supports interviews in 30+ languages, including English, Spanish, French, German, Dutch, Japanese, Hindi, Portuguese, Italian, Polish, Turkish, Korean, Mandarin Chinese, Arabic, and Swedish.\n\nSet your study's language in the **Interaction Mode** section under **Default Language**. This controls:\n- The language the AI interviewer speaks\n- The language of speech-to-text transcription\n- The language of the interview UI and on-screen instructions\n\nWhen you set a non-English language, Koji automatically configures the underlying conversational agent for that language and selects an appropriate voice profile. Transcripts are captured in the native language, and AI analysis (quality scores, themes, insights) is generated in the same language.\n\nFor multilingual research spanning multiple markets, create separate studies per language with individual interview links. See [Multi-Language Research](/docs/multilingual-research-guide) for a full guide.\n\n## Step 4: Configure Your Landing Page\n\nThe landing page is the first thing participants see. A well-designed landing page significantly increases completion rates — the difference between a generic link and a well-crafted landing page is often a 2–3x lift in participation.\n\nIn the **Customize** tab, configure:\n\n**Headline and Description**\n\nWrite a clear, welcoming headline that explains what the interview is about. Participants convert better when they understand the purpose upfront.\n\n- ❌ \"Research Study Q1 2025\"\n- ✓ \"Share your experience with our checkout process — takes about 10 minutes\"\n\n**Duration Badge**\n\nEnable the duration badge to show the estimated interview length. This is the single highest-impact trust element on the landing page. Participants who know the interview takes 10 minutes are far more likely to start than those who have no idea what they are committing to.\n\n**Anonymity Badge**\n\nAdd an anonymity or privacy badge if participants might hesitate to share candidly. Common messages: \"Your responses are anonymous\", \"This interview is confidential\", or custom text.\n\n**Accent Color**\n\nSet a brand color to match your organization's identity. The landing page, animated background orb, and UI accents use this color — making the interview feel native to your brand rather than a generic research tool.\n\n## Step 5: Set Up an Intake Form (Optional)\n\nAn intake form collects participant information before the interview begins. This is useful for:\n- Capturing email addresses for follow-up or incentive distribution\n- Screening participants based on role, company size, or behavior\n- Pre-populating participant records with CRM data\n\nConfigure up to 6 fields in the **Lead Collection Form** section. Supported field types include text, email, phone, select dropdown, textarea, and checkbox.\n\nIf you need to screen participants before they proceed, add a required Select field with your qualifying criteria. Participants who do not match can be shown a graceful redirect message.\n\nSee [Intake Forms and Consent](/docs/intake-forms-and-consent) for detailed configuration options.\n\n## Step 6: Add Context Documents (Optional)\n\nFor studies involving a specific product, workflow, or concept, you can upload context documents that the AI interviewer uses to inform its questions and follow-ups.\n\nContext documents are useful for:\n- Product specs or mockup descriptions (concept testing)\n- Customer journey maps (UX research)\n- Feature release notes (post-launch feedback research)\n- Competitive comparison data (win/loss research)\n\nUpload documents in your study's **Settings** tab. The AI uses these as background knowledge but does not read them aloud to participants — it references them only when relevant to probe deeper on a specific topic.\n\n## Step 7: Customize Your Interview Slug\n\nBy default, Koji generates a random URL slug for your interview. You can customize it to something memorable and professional:\n\n- `koji.so/i/checkout-research-q1`\n- `koji.so/i/enterprise-discovery`\n- `koji.so/i/post-launch-feedback`\n\nCustom slugs are available on paid plans and configured in the **Collect** tab under **Interview Link Settings**. A clean, recognizable URL improves click-through rates in email campaigns and feels more professional when sharing with customers.\n\n## Step 8: Test Before You Launch\n\nBefore sending links to real participants, run a test interview yourself:\n\n1. Open your interview link in a fresh browser window (or incognito mode)\n2. Grant microphone permission when prompted\n3. Complete the full interview, including all question types\n4. Check the transcript in your dashboard for accuracy\n5. Review the auto-generated quality score and AI insights\n\nListen and evaluate:\n- Does the greeting feel natural and appropriate for your audience?\n- Does the AI probe appropriately when you give short or vague answers?\n- Does the interview cover all your key research questions?\n- Is the transcript readable and accurate?\n\nIf the AI is not covering a particular topic, edit your research brief to be more explicit about that area. The brief is the source of truth for the AI's behavior — more specific briefs produce more targeted interviews.\n\nBetter still, run the test from the **Preview** tab on your study's Interviews page. Preview conversations are marked as tests, so they never count toward your real analysis and you have nothing to clean up afterwards.\n\n## Step 9: Launch and Monitor\n\nPick the route that fits the study, all from the study's Interviews page. If you already have the people, a customer list, users in your product, employees, a CRM segment, import them so each person gets a personalized link: you can see who responded, the interviewer greets them by name, and the report is already segmented by whatever you imported. If candor matters more than tracking, an exit interview, an engagement pulse, a sensitive topic, share the generic link and leave the lead form off, so responses stay anonymous. If people would rather talk than type, share a call link or have Koji call them. And if you do not have the people, or you need a market you have no list for, open [Panel recruit](/docs/panel-recruitment) and Koji finds them across 130+ countries at a price you approve first. All routes land in the same study and the same report.\n\nWays to get the link in front of your own people:\n- Email outreach (manual or via your CRM using personalized links)\n- In-product prompts (embed or trigger after key user actions)\n- Customer success sequences\n- If you prefer to run recruitment yourself, marketplaces like User Interviews, Respondent or Prolific can source a specialist audience; share your Koji link with the people they find and the interviews land in the same study\n\nMonitor incoming responses in your **Results** tab. Koji's quality gate automatically filters out low-quality or incomplete interviews before they affect your analysis.\n\nVoice interviews are automatically transcribed, analyzed, and scored within minutes of completion. You do not need to wait for all responses before reviewing early themes — the Insights Dashboard updates in real time as new interviews arrive.\n\n## Voice Interview Best Practices\n\n**For study design:**\n- Keep your research brief focused on one core problem area. A tightly scoped brief produces more relevant voice data than a broad one.\n- For qualitative saturation, aim for 15–40 participants depending on your topic and audience homogeneity.\n- Set the interview duration expectation correctly — voice interviews typically run 8–15 minutes.\n\n**Tips to share with participants:**\n- Use Chrome or a Chromium-based browser for the best microphone support\n- Wear headphones or earbuds to prevent echo\n- Find a quiet space — background noise can interfere with speech detection\n- Plan for about 10 minutes of uninterrupted time\n\n## What Happens After the Interview\n\nEvery completed voice interview automatically generates:\n\n- **Full transcript** — timestamped, speaker-labeled text of the complete conversation\n- **Quality score** — 1–5 rating based on relevance, depth, coverage, and engagement\n- **AI-generated insights** — themes, sentiment, pain points, feature requests, and notable quotes\n- **Question coverage map** — which research questions were covered and with what confidence\n\nThese feed directly into Koji's [Insights Dashboard](/docs/insights-dashboard) and [Research Reports](/docs/generating-research-reports), giving you a complete analysis without any manual tagging, coding, or synthesis.\n\nWith platforms like Koji, the heavy lifting of voice research — scheduling, moderation, transcription, and analysis — is handled automatically. What once took weeks and a dedicated research team now takes days and a single researcher.\n\n## Further reading on the blog\n\n- [The 11 Best AI Survey Tools in 2026 (Ranked & Reviewed)](/blog/best-ai-survey-tools-2026) — We tested 11 AI-powered survey platforms head-to-head — from AI question generation to automatic open-ended analysis to fully AI-moderated c\n- [IVR Surveys Are Dead: Why AI Voice Interviews Win in Every Metric](/blog/ivr-surveys-are-dead) — IVR surveys have completion rates below 15% and capture zero qualitative depth. AI voice interviews achieve 55-70% completion with 3x longer\n- [Koji vs Listen Labs: AI Interview Platforms Compared (2026)](/blog/koji-vs-listen-labs-2026) — Listen Labs raised $69M and powers enterprise research at brands across the Fortune 500. Koji is the accessible AI-native alternative starti\n\n<!-- further-reading:blog -->\n","category":"guides","lastModified":"2026-10-29T13:18:40.847066+00:00","metaTitle":"How to Set Up AI Voice Interviews: A Researcher's Complete Guide — Koji Docs","metaDescription":"Step-by-step guide to configuring and launching AI voice interview studies in Koji. Covers brief setup, voice mode configuration, language settings, landing pages, and testing.","keywords":["AI voice interviews","voice interview setup","automated voice research","how to run AI voice interviews","voice research platform","voice interview configuration"],"aiSummary":"A complete researcher-facing guide to setting up AI voice interview studies in Koji. Covers creating the research brief, enabling voice mode, selecting interview language, configuring the landing page, setting up intake forms, adding context documents, customizing the interview slug, testing before launch, and monitoring responses. Voice interviews produce 3–5x richer qualitative data than typed answers, and platforms like Koji automate the entire workflow from moderation to analysis.","aiDifficulty":"beginner","aiEstimatedTime":"15 minutes"},{"type":"documentation","id":"9d03cbac-952e-4842-b4f5-fe5e3b1a0e4b","slug":"plain-language-research-questions","title":"Plain Language for Research Questions: Writing Questions Every Participant Can Understand","url":"https://www.koji.so/docs/plain-language-research-questions","summary":"Research questions should use common words, short sentences, one idea per question and context stated first. In the 2023 PIAAC round, 28% of US adults scored at the lowest literacy levels, up from 19% in 2017. Lenzner and colleagues (2010) found at least six of seven tested text features made survey questions harder to process, and less comprehensible questions produced lower-quality answers. Experts prefer plain wording too: in Kimble's studies, at least 80% of nearly 1,200 judges and lawyers preferred plain versions. Test with paraphrase tests, cognitive interviews and pilots. Koji's free interview preview, follow-up probing, brief assistant and answer confidence flags help find and fix unclear questions.","content":"**Short answer: write research questions in plain language: common words, short sentences, one idea per question, and the key condition stated before the question rather than buried after it. Then test them with real participants before you field the study. Plain wording is not dumbing down. It reduces the share of answers that respond to a question you did not mean to ask, and it improves data from experts as well as from everyone else.**\n\nEvery research question passes through a participant's comprehension before it produces an answer. If people read the question differently, they are answering different questions, and the numbers you add up are not measuring one thing. That makes question readability a validity problem, not a style preference.\n\n## Why reading level matters more than you think\n\nIt is easy to write for the people in your team: highly educated, fluent in your product's vocabulary, comfortable with long sentences. Your participants are a wider group.\n\nThe OECD's Programme for the International Assessment of Adult Competencies (PIAAC) measures adult literacy directly. In the 2023 round, released by the US National Center for Education Statistics in December 2024, 28% of US adults scored at the lowest levels of literacy, up from 19% in 2017. Average literacy scores were lower in 2023 than in 2017, and the US was tied for 14th of the 31 countries and economies that took part. Adults at the lowest level can typically read short texts well enough to perform simple tasks, such as filling out a basic form. A dense, multi-clause question is not a simple task.\n\nEven in a B2B study of educated professionals, comprehension is fragile. People answer quickly, on a phone, between other tasks, and often in their second language. The question that seemed clear in a document review is read once, at speed.\n\n## What makes a question hard to understand\n\nTimo Lenzner, Lars Kaczmirek and Alwine Lenzner tested this directly in a 2010 experiment published in _Applied Cognitive Psychology_. They compared well-written survey questions with versions containing specific problem features and measured response times as a sign of cognitive burden. At least six of the seven text features they tested made questions harder to process. In later work, Lenzner reported that respondents who received less comprehensible questions gave lower-quality answers than those who received easier versions.\n\nThe features to watch for include:\n\n- **Low-frequency words.** \"Utilise\", \"facilitate\", \"remuneration\". Use \"use\", \"help\", \"pay\".\n- **Vague or abstract noun phrases.** \"Your experience with onboarding-related functionality\" instead of \"setting up your account\".\n- **Complex sentence structure.** Long clauses placed before the main point (\"Thinking about the last three months, and excluding any time spent on training, how...\") force people to hold information in memory before they know what it is for.\n- **Ambiguous terms.** \"Regularly\", \"team\", \"use\" all mean different things to different people unless you define them.\n- **Hidden double questions.** \"How easy and fast was it?\" asks two things.\n- **Unnecessary negatives.** \"Do you disagree that it is not difficult?\" defeats most readers.\n\nThe survey methodologist Floyd J. Fowler Jr. put the goal simply in _Improving Survey Questions_: \"Word questions so that every respondent is answering the same question.\" Each feature above works against that goal.\n\n## Plain language is not just for low literacy\n\nThe most common objection is that expert audiences will find plain wording patronising. The evidence runs the other way. Joseph Kimble, a long-standing researcher of legal drafting, summarised studies in which nearly 1,200 judges and lawyers in three US states were shown paired passages from legal documents, one traditional and one in plain language. In every state, at least 80% preferred the plain version. If people whose profession is reading complex text prefer plain text, your product managers and engineers will too.\n\nExperts are also the group most exposed to jargon traps. They may know your industry term, but use it differently from you. Plain wording forces you to say what you mean.\n\n## How to write plain research questions\n\n### Lead with the context, keep it short\n\nPut the time frame or condition first and keep it to a short phrase: \"In the past month, how often did you export a report?\" is easier than a question that ends with three qualifying clauses.\n\n### One question, one idea\n\nIf you find \"and\" or \"or\" joining two concepts, split the question. In interviews, a double question gets an answer to whichever half the participant remembers.\n\n### Use the participant's words, not yours\n\nYour team says \"workspace\", \"seat\" or \"integration\". Customers may say \"account\", \"login\" or \"connecting it to Slack\". Early interviews and support tickets show you the words people actually use. Ask in those words.\n\n### Define anything that could vary\n\nIf \"team\" matters, say what you mean: \"the people you work with day to day\". Put the definition before the question, not after it.\n\n### Prefer concrete over abstract\n\n\"What happened the last time you tried to share a report?\" is easier to answer accurately than \"How would you describe your report-sharing experience?\" Concrete questions are also less likely to invite generic answers; see [Why a Vague Question Gets a Short Answer](/docs/question-specificity-answer-length).\n\n### Keep answer options as plain as the question\n\nScale labels and choice options need the same care. \"Somewhat\" and \"moderately\" mean little in isolation. Short, distinct options are easier to tell apart.\n\n### Read it aloud\n\nIf you run interviews by voice, participants hear the question once and cannot re-read it. A question that needs a second reading on screen will fail out loud. Reading every question aloud before fielding is the quickest test there is.\n\n## Testing comprehension before you field\n\nReadability formulas give a rough signal but miss ambiguity and jargon. The reliable test is people.\n\n1. **Paraphrase test.** Ask a few people who match your audience to say the question back in their own words. Any paraphrase that changes the meaning is a rewrite.\n2. **Cognitive interviewing.** Ask participants to think aloud as they answer, then probe specific words: \"What did 'regularly' mean to you here?\" See the [Cognitive Interview Guide](/docs/cognitive-interview-guide).\n3. **Pilot the whole study.** A short pilot shows where people hesitate, ask for clarification or give answers that do not fit. See the [Pilot Study Guide](/docs/pilot-study-user-research-guide).\n4. **Check the answers, not just the questions.** After fielding, look for questions with unusually short, off-topic or \"it depends\" answers. These are often comprehension failures, not disengaged participants.\n\n## Before and after\n\n| Before                                                                                                                      | After                                                                                                            |\n| --------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------- |\n| To what extent has the implementation of the new permissions functionality facilitated your team's collaborative workflows? | Since we changed who can edit reports, has it become easier or harder for your team to work on reports together? |\n| How satisfied and confident are you with our onboarding?                                                                    | How easy was it to set up your account? (Then ask about confidence separately.)                                  |\n| Do you not find the export feature difficult to use?                                                                        | How easy or hard is it to export a report?                                                                       |\n| Please indicate the frequency with which you utilise dashboards.                                                            | How often do you look at a dashboard?                                                                            |\n\n## How Koji helps\n\nKoji runs AI-moderated interviews by text or voice. Plain wording matters more in this format, not less, because the question is delivered once, conversationally, often by voice. Koji supports this in several practical ways.\n\n- **Test before anyone else sees it.** The Preview tab on a study's Interviews page lets you take your own interview, by text or voice, exactly as a participant would. Previews are free, never use credits and never appear in the report. Run each question by voice to hear whether it holds up when spoken.\n- **Follow-up probing catches confusion.** Because the AI interviewer asks follow-up questions on the participant's answer, a reply that shows a misunderstanding can be explored instead of being recorded as a final answer. A static form records the wrong answer and moves on. You set how many follow-ups each question gets.\n- **Edit wording with the brief assistant.** The assistant that builds your study brief can change any question on request, and it tells you exactly what it changed. Ask it to shorten a question, remove jargon or split a double question.\n- **Signals when an answer is uncertain.** When Koji extracts answers to [structured questions](/docs/structured-questions-guide) from the conversation, each extracted answer carries a high, medium or low confidence flag. A question that collects many low-confidence answers is a candidate for rewording. See [Answer Confidence in Koji Reports](/docs/answer-extraction-confidence-flags).\n- **Quality scores that reflect what people said.** Every interview gets a quality score, capped by how much the participant actually said. A cluster of low-scoring interviews around one part of a study can point to a question participants could not engage with.\n- **Interviews in the participant's language.** Koji interviews in 30+ languages, so participants who would struggle with a second language can answer in their first.\n\nCompared with a traditional survey, where a confusing question produces silent, unusable data, an AI-moderated interview gives you a transcript of exactly where comprehension broke down. It does not fix a badly written question for you. The writing still matters most.\n\n## Key takeaways\n\n- Question comprehension is a validity issue: people who read a question differently answer different questions.\n- Over a quarter of US adults scored at the lowest literacy levels in the 2023 PIAAC round; write for your real audience, not your team.\n- Low-frequency words, abstract phrasing, complex structure, ambiguity, double questions and negatives all make questions harder and answers worse.\n- Experts prefer plain language too: in studies of nearly 1,200 judges and lawyers, at least 80% preferred plain versions.\n- Test comprehension with paraphrase tests, cognitive interviews and pilots before fielding.\n\n## Related Resources\n\n- [Structured Questions Guide](/docs/structured-questions-guide)\n- [How to Write Unbiased Survey Questions](/docs/survey-question-wording-guide)\n- [Why a Vague Question Gets a Short Answer](/docs/question-specificity-answer-length)\n- [Cognitive Interview Guide](/docs/cognitive-interview-guide)\n- [Pilot Study Guide](/docs/pilot-study-user-research-guide)\n- [Translating Research Questions](/docs/translating-research-questions-trapd)\n\n## Sources\n\n- National Center for Education Statistics (2024). Highlights of the 2023 U.S. PIAAC Results. Press release and report, December 10, 2024.\n- Lenzner, T., Kaczmirek, L., & Lenzner, A. (2010). Cognitive burden of survey questions and response times: A psycholinguistic experiment. _Applied Cognitive Psychology_, 24(7), 1003–1020.\n- Lenzner, T. (2011). _A psycholinguistic look at survey question design and response quality_. Doctoral dissertation, University of Mannheim.\n- Fowler, F. J., Jr. (1995). _Improving Survey Questions: Design and Evaluation_. Sage.\n- Kimble, J. (2012). _Writing for Dollars, Writing to Please: The Case for Plain Language in Business, Government, and Law_. Carolina Academic Press. See also Kimble's Plain Language columns in the _Michigan Bar Journal_.\n","category":"Research Methods","lastModified":"2026-10-29T13:18:40.564413+00:00","metaTitle":"Plain Language Survey and Interview Questions: How to Write Them","metaDescription":"Hard-to-read questions get worse answers. Learn which wording features hurt comprehension, what the literacy data says, and how to write and test plain research questions.","keywords":["plain language survey questions","survey reading level","readability of survey questions","how to write clear interview questions","question comprehension","survey question clarity","plain language research"],"aiSummary":"Research questions should use common words, short sentences, one idea per question and context stated first. In the 2023 PIAAC round, 28% of US adults scored at the lowest literacy levels, up from 19% in 2017. Lenzner and colleagues (2010) found at least six of seven tested text features made survey questions harder to process, and less comprehensible questions produced lower-quality answers. Experts prefer plain wording too: in Kimble's studies, at least 80% of nearly 1,200 judges and lawyers preferred plain versions. Test with paraphrase tests, cognitive interviews and pilots. Koji's free interview preview, follow-up probing, brief assistant and answer confidence flags help find and fix unclear questions.","aiPrerequisites":["A draft interview guide or survey","Basic familiarity with writing research questions"],"aiLearningOutcomes":["Explain why question comprehension affects the validity of research data","Identify wording features that make survey and interview questions hard to understand","Rewrite complex questions into plain language without losing meaning","Test question comprehension with paraphrase tests, cognitive interviews and pilots","Use interview previews, follow-up probing and answer confidence flags in Koji to catch unclear questions"],"aiDifficulty":"beginner","aiEstimatedTime":"11 min read"},{"type":"documentation","id":"9bfdd750-c496-473d-9c40-e83c31357f38","slug":"phenomenological-research-guide","title":"Phenomenological Research: How to Study Lived Experience in UX and Product Research (2026)","url":"https://www.koji.so/docs/phenomenological-research-guide","summary":"Phenomenological research studies how people make sense of significant lived experiences, asking not what users did but what the experience was like and what it meant. Two main traditions exist: descriptive phenomenology (Giorgi/Husserl) and Interpretative Phenomenological Analysis (Smith). IPA has an idiographic focus — small samples (3–10), deep within-case analysis first, then cross-case comparison. Best for studying transformative product moments: onboarding aha, trust formation, accessibility experience, identity-forming use.","content":"## Phenomenological Research: How to Study Lived Experience in UX and Product Research\n\n**Phenomenological research is a qualitative methodology that studies how people make sense of significant lived experiences — the texture, meaning, and personal interpretation of an event from the participant’s own perspective, rather than its external observable features.** Where most product research asks \"what did the user do?\", phenomenology asks \"what was that experience like for them, and what did it mean?\" The result is a depth of insight that survey instruments and behavioral analytics structurally cannot reach.\n\nPhenomenology originated as a philosophical movement (Edmund Husserl, Martin Heidegger, Maurice Merleau-Ponty) and was operationalized into research methodology by figures like Amedeo Giorgi (descriptive phenomenology) and Jonathan Smith (Interpretative Phenomenological Analysis, or IPA). It’s now used across psychology, healthcare, education, and — increasingly — UX research for studying transformative moments in a user’s relationship with a product: onboarding \"aha\" moments, breakdowns of trust, the experience of becoming a power user, the lived sense of an accessibility barrier. A 2017 IPA-in-UX paper notes that *\"IPA contributes to UX research by investigating both experience and meaning, and by providing holistic insights appropriate for service design and the fuzzy front-end of innovation.\"*\n\nFor product teams whose competitive edge depends on emotional resonance — habit-forming consumer apps, healthcare experiences, accessibility-critical products, premium-priced services — phenomenological research is one of the few methods rigorous enough to put words to what makes the product feel right (or wrong).\n\n---\n\n## What Makes Research \"Phenomenological\"?\n\nFour commitments define phenomenological research and distinguish it from adjacent qualitative methods:\n\n### 1. Idiographic Focus\n\nPhenomenology cares about the *particular*, not the general. You start with one person’s account, render it in detail, then carefully look for convergence and divergence with a small number of other accounts. This is the opposite of survey-style generalization. As IPA pioneers Smith, Flowers, and Larkin put it, *\"IPA has an idiographic focus, which means that instead of producing generalization findings, it aims to offer insights into how a given person, in a given context, makes sense of a given situation.\"*\n\n### 2. Lived Experience as the Subject\n\nThe data is the participant’s first-person account of how an experience felt and what they made of it — not their behavior, not their stated opinion, but the texture of the experience itself. \"What was it like for you when…?\" is the canonical phenomenological question.\n\n### 3. Bracketing (Epoché)\n\nBorrowed from Husserl: researchers must work to set aside (bracket) their own preconceptions, theoretical frameworks, and assumed meanings before encountering the data. Bracketing isn’t about achieving neutrality — that’s impossible — but about being explicit about your starting point so it doesn’t silently shape the analysis.\n\n### 4. Double Hermeneutic (in IPA)\n\nIPA explicitly acknowledges that the researcher is *interpreting the participant’s interpretation of their own experience.* This double layer of meaning-making is treated as a strength, not a flaw: insight comes from the researcher’s engaged interpretation, anchored carefully in the participant’s own words.\n\n---\n\n## When to Use Phenomenological Research\n\nPhenomenology is the right method when:\n\n- **You need to understand a significant, meaningful experience.** Routine, low-emotional-stakes interactions are better studied with usability methods or analytics. Phenomenology shines on experiences that matter: first time hitting a paywall, recovering from a billing error, the moment of cancellation, the first successful collaboration on a new tool.\n- **Behavioral data has run out.** Your analytics show *what* happened, but the team disagrees on *why*. Phenomenology supplies the meaning behind the behavior.\n- **The product is in an emotionally textured category.** Healthcare, mental health, fitness, finance, education, accessibility, grief tech — categories where the product touches identity and meaning. Phenomenology is the rigorous counterpart to vibe-based \"design empathy.\"\n- **You’re designing for transformation, not just task completion.** Habit-forming products, behavior change apps, learning platforms. The user’s sense of becoming someone different is the central design challenge — and the central thing phenomenology studies.\n- **You want depth on a small group, not surface across a large one.** IPA studies typically work with 3–10 participants, not 50.\n\nPhenomenology is *not* the right method for usability bug-finding, A/B test analysis, large-scale prioritization, or quick-turn discovery sprints. Use the [right method](/docs/ux-research-methods-guide) for the question. According to the [2025 State of User Research](https://www.userinterviews.com/state-of-user-research-report), researchers conduct ~3 qualitative studies for every 1 quantitative study — phenomenological designs anchor the deepest tier of that qualitative portfolio.\n\n---\n\n## The Two Main Traditions\n\n### Descriptive Phenomenology (Giorgi, after Husserl)\n\nFocuses on producing a faithful, structured description of the essential features of an experience. The researcher’s job is to *describe*, not interpret — bracketing is rigorous, and the output is a structured essence: \"Here is what this experience consists of, in its general structure.\"\n\n### Interpretative Phenomenological Analysis (IPA, Smith)\n\nThe more widely-used variant in applied research. Embraces interpretation as inevitable and productive. The researcher engages deeply with each transcript, develops emergent themes, and produces an interpretive account that balances participants’ own words with the researcher’s analytic insight. IPA is the dominant tradition in modern UX-adjacent phenomenology because it explicitly acknowledges the design researcher’s active role.\n\nA hybrid form — *hermeneutic phenomenology* (Heidegger, van Manen) — sits between the two and is common in healthcare research.\n\n---\n\n## How to Conduct a Phenomenological Study (IPA Workflow)\n\n### Step 1: Frame an Experiential Research Question\n\nGood phenomenological questions are open, experiential, and bounded. Examples:\n\n- \"What is it like for first-time users of our app to encounter our pricing for the first time?\"\n- \"How do power users experience the moment when a product changes a workflow they’d come to rely on?\"\n- \"What is the lived experience of a clinician using AI-suggested diagnoses?\"\n\nAvoid yes/no framings, behavioral framings, and framings that presuppose the answer.\n\n### Step 2: Recruit a Small, Purposive Sample\n\nIPA samples are small — typically 3–10 participants — and homogeneous in the sense that all participants have lived the experience under study. Quality of sampling matters more than quantity. Use a [careful screener](/docs/screener-questions-guide) to confirm the participant has actually had the experience you’re studying. For B2B contexts, see [B2B participant recruiting](/docs/recruiting-b2b-participants). In Koji, if the people who lived the experience are your own customers or users, import them so each person gets a personalized link, or share the generic link with the lead form off when anonymity will help them speak freely. If you need people outside your own base, open [Panel recruit](/docs/panel-recruitment), describe who you want to hear from, and Koji finds them across 130+ countries at a price per completed interview you approve first, with your screening questions running before the interview so only people who have actually had the experience get through.\n\n### Step 3: Conduct In-Depth Experiential Interviews\n\nPhenomenological interviews are unstructured-to-semi-structured, slow, and probing. Open with an invitation to narrative: *\"Take me back to the first time you tried to…\"* Follow the participant’s account; probe for the texture of their experience (*\"What did that feel like?\"*, *\"What was going through your mind?\"*, *\"What did that mean to you?\"*); avoid leading or imposing your framework.\n\nThis is where AI-moderated interview platforms like Koji genuinely change the methodology. Koji’s [AI follow-up probing](/docs/ai-probing-guide) is specifically trained on the kind of experiential follow-ups that phenomenological research depends on — staying with a moment, asking what it felt like, returning to a fragment the participant mentioned in passing. The AI moderator doesn’t tire, doesn’t fall back on its own assumptions, and doesn’t skip the slow questions. For methodology that historically required the most experienced human moderators, this changes who can run a phenomenological study and how many they can run.\n\nKoji’s [voice interview option](/docs/ai-voice-surveys-complete-guide) is particularly important for phenomenology: the texture of speech — pauses, emphasis, vocal warmth or hesitation — carries meaning that gets lost in text. Run voice for phenomenological work whenever you can.\n\n### Step 4: Transcribe Verbatim\n\nUse verbatim transcription with non-verbal markers (pauses, laughter, hesitations) where they’re analytically significant. Koji’s [AI transcription](/docs/ai-transcription-research-interviews) handles this automatically across long interviews.\n\n### Step 5: Bracket Your Assumptions\n\nBefore analysis, write a reflexive memo: what do you already think this experience is about? What theories or assumptions are you bringing? Naming them doesn’t neutralize them, but it makes them visible and auditable.\n\n### Step 6: Analyze Each Transcript Individually First\n\nIPA analysis is sequential, not aggregated. Take one transcript at a time. Read it through. Re-read it. Make exploratory annotations — descriptive (what the participant says), linguistic (how they say it), and conceptual (what it might mean). Develop emergent themes that stay close to the participant’s own words.\n\nOnly once you’ve fully analyzed Participant 1 do you move to Participant 2. Resist the temptation to look across participants too early — it dissolves the idiographic richness IPA is designed to preserve.\n\n### Step 7: Look for Patterns Across Participants\n\nOnce each transcript has been analyzed individually, look for super-ordinate themes that connect multiple participants. Where do accounts converge? Where do they meaningfully diverge? What does the divergence reveal about the conditions of the experience?\n\nKoji’s [insights dashboard](/docs/insights-dashboard) and the report, where every claim links to the quote behind it, accelerate this cross-participant comparison without bypassing the within-participant work. You can also ask questions about the study from Claude or any MCP assistant through the [Koji connector](/docs/mcp-overview), which reads the transcripts and study data. Use the AI surface as a *check* on your manual analysis (\"Does it surface the themes I identified? What did I miss?\"), not a replacement for it.\n\n### Step 8: Write Up With Quotes Anchoring Every Claim\n\nA phenomenological writeup interleaves the researcher’s analytic claims with the participants’ own words. Every interpretive move should be visibly anchored in a quote. The reader should be able to assess whether the interpretation is faithful to the data.\n\n---\n\n## Phenomenology vs. Other Qualitative Methods\n\n| Dimension | Phenomenology | [Grounded Theory](/docs/grounded-theory-qualitative-research) | [Thematic Analysis](/docs/thematic-analysis-guide) |\n|-----------|---------------|--------------|------------------|\n| Goal | Understand lived experience and meaning | Build theory inductively | Identify patterns |\n| Sample size | 3–10 (deep) | 20–30+ to saturation | 10–30 (flexible) |\n| Unit of analysis | The individual account | The emerging concept | The transcript dataset |\n| Output | Phenomenological essence or interpretive account | Theoretical framework | Themes with supporting evidence |\n| Depth | Deepest of the three | Deep | More accessible |\n\nIn practice, many UX studies combine methods: a phenomenological study of 6 power users, paired with a [thematic analysis](/docs/thematic-analysis-guide) of 30 broader users, and quantitative validation via [scale questions](/docs/scale-questions-guide). This is the [mixed-methods approach](/docs/mixed-methods-research-guide) the strongest research teams converge on.\n\n---\n\n## Phenomenology in Product and UX Research: Use Cases\n\n- **Onboarding \"aha\" moments.** What is it like, from the user’s point of view, to first experience the core value of the product? This is the prototype phenomenological question for early-stage product teams.\n- **Trust formation and breakdown.** What is the experience of starting to trust an AI-powered feature — and what is the experience of trust breaking down after a bad output?\n- **Accessibility experience.** The lived experience of using a product with a screen reader, with low vision, with motor impairments, or with cognitive accessibility needs. Phenomenology is the method that puts users’ own meaning at the center.\n- **Habit and identity formation.** Fitness apps, learning apps, habit trackers — what is it like to become \"the kind of person who uses this app daily\"?\n- **Significant negative experiences.** Cancel flow, billing dispute, account lockout, accidental data loss. Phenomenology captures what these feel like — input no usability test can produce. Related: [cancel-flow exit interviews](/docs/cancel-flow-exit-interview).\n\n---\n\n## Limitations and Common Pitfalls\n\n1. **Trying to generalize.** Phenomenology is idiographic. Don’t report \"78% of users experience…\"  — report on a small set of accounts in depth.\n2. **Imposing your framework.** The researcher’s framework is the enemy of phenomenology. Practice bracketing actively. See [research bias guide](/docs/research-bias-guide).\n3. **Skipping the within-case analysis.** Looking across participants too early destroys the idiographic depth. Stay with each account before comparing.\n4. **Surface-level interviewing.** Phenomenology needs slow, probing, experiential interviews. Use [active listening](/docs/active-listening-techniques) and the [Five Whys technique](/docs/five-whys-technique) tradition of returning to a moment until its texture is clear.\n5. **Treating phenomenology as a default method.** It’s the right method for some questions and the wrong method for many. Match method to question.\n\n---\n\n## How Koji Makes Phenomenological Research Accessible\n\nThe historic barrier to phenomenology is moderator skill: experiential interviewing takes years to develop, and very few teams have enough phenomenologically trained interviewers to do it at any scale. Koji changes that:\n\n- **Adaptive experiential probing.** Koji’s AI is trained to stay with significant moments, probe for the texture of experience, and avoid leading. Each interview goes deeper than scripted human interviews routinely manage.\n- **Voice as default.** Voice interviews preserve the speech texture phenomenology depends on. Koji supports voice as a first-class mode.\n- **Verbatim transcription with markers.** No information lost between conversation and analysis.\n- **Researcher in the loop, not replaced.** Koji is the interview engine, but interpretation stays with the human researcher. The platform accelerates the data collection and surfaces patterns; the researcher does the phenomenological work that the platform cannot.\n- **Multilingual support.** Phenomenological research is global. [Koji runs interviews in 30+ languages](/docs/multilingual-research-guide), picked by the participant on the welcome screen, without forcing translation that strips meaning.\n\nFor product teams who couldn’t previously afford the moderator hours phenomenology requires, this is the difference between *we know what users do* and *we know what it’s like to be our users*. According to industry benchmarks, teams using AI-assisted research tools achieve 60% faster time-to-insight — and for methods historically constrained by moderator skill, the time savings come with new accessibility.\n\n---\n\n## Related Resources\n\n- [Structured Questions in AI Interviews](/docs/structured-questions-guide) — Koji’s six question types let you layer scale and ranking inputs alongside experiential narrative when phenomenology pairs with quantitative validation\n- [Grounded Theory in Qualitative Research](/docs/grounded-theory-qualitative-research) — when your research goal is theory rather than lived experience\n- [The Complete Guide to Thematic Analysis](/docs/thematic-analysis-guide) — the broader analytic family phenomenology relates to\n- [Mental Models in UX Research](/docs/mental-models-research-guide) — adjacent method for understanding how users think\n- [Semi-Structured Interviews](/docs/semi-structured-interview-guide) — the interview format closest to phenomenological interviewing\n- [Critical Incident Technique](/docs/critical-incident-technique) — focuses experiential research on a specific incident\n\n## Further reading on the blog\n\n- [Why AI Interviewers Are the Future of Customer Research](/blog/why-ai-interviewers-are-the-future-of-customer-research)\n- [How to Conduct Remote User Interviews: The Complete Guide (2026)](/blog/how-to-conduct-remote-user-interviews-2026)\n\n<!-- further-reading:blog -->\n","category":"Research Methods","lastModified":"2026-10-29T13:18:40.292145+00:00","metaTitle":"Phenomenological Research: Lived Experience Research for UX & Product (2026)","metaDescription":"How to design and run phenomenological research and IPA studies — bracketing, idiographic analysis, and using AI-moderated voice interviews to capture lived experience at depth.","keywords":["phenomenological research","phenomenology","interpretative phenomenological analysis","IPA research","lived experience research","experiential research","qualitative methodology","descriptive phenomenology"],"aiSummary":"Phenomenological research studies how people make sense of significant lived experiences, asking not what users did but what the experience was like and what it meant. Two main traditions exist: descriptive phenomenology (Giorgi/Husserl) and Interpretative Phenomenological Analysis (Smith). IPA has an idiographic focus — small samples (3–10), deep within-case analysis first, then cross-case comparison. Best for studying transformative product moments: onboarding aha, trust formation, accessibility experience, identity-forming use.","aiPrerequisites":["Familiarity with qualitative research","Comfort with interview-based methods"],"aiLearningOutcomes":["Choose phenomenology vs. adjacent qualitative methods for the right research question","Bracket researcher assumptions before analysis","Conduct slow experiential interviews with proper probing","Run idiographic IPA analysis (within-case first, then cross-case)","Apply phenomenological insight to product and UX decisions"],"aiDifficulty":"advanced","aiEstimatedTime":"15 min read"},{"type":"documentation","id":"95912610-7d0d-4d8f-89d9-341d4ebb4fa9","slug":"multilingual-research-guide","title":"Multi-Language User Research: How to Interview Participants in Any Language","url":"https://www.koji.so/docs/multilingual-research-guide","summary":"Koji runs AI voice and text interviews in 30+ languages. One study can offer several languages: Koji translates every question, option and scale label on publish, participants pick their language on the welcome page, and answers are recorded against the original options so results stay comparable. Reports and analyses are written in the study language by default, or in any supported language chosen with the Report language picker. The guide also covers localizing briefs, formality registers, separate studies per market and cross-language synthesis.","content":"# Multi-Language User Research: How to Interview Participants in Any Language\n\nMost research teams run their user interviews in English, even when their customers are not. The result is systematic blind spots: non-English-speaking markets receive less research, their needs are underrepresented in product decisions, and entire customer segments become invisible in qualitative data.\n\nAI-powered interview platforms like Koji change the economics of multilingual research. Instead of hiring bilingual moderators, coordinating cross-timezone sessions, or outsourcing translation, teams can run voice and text interviews in 30+ languages with the same depth and automated analysis they get from English-language studies.\n\nThis guide covers how to set up multilingual studies in Koji, best practices for cross-language research, and how to synthesize findings across language groups.\n\n## Why Multilingual Research Matters\n\n- People describe problems most precisely in the language they think in. How users name a pain point, what emotional language they use and how they frame success all get flattened when they answer in a second language.\n- Non-English speakers are consistently underrepresented in qualitative research, even at global companies with international customer bases.\n- Market expansion decisions made on English-only interviews rest on the customers who were easiest to reach, not the ones you are trying to win.\n\nThe traditional barriers are real: bilingual moderators are scarce and expensive, translation adds days or weeks to a timeline, and many analysis tools struggle with non-English transcripts. Koji removes most of this by running the whole interview, from the questions to the analysis, in the participant's language.\n\n## Languages Supported\n\nKoji interviews in 30+ languages, in both voice and text mode:\n\nEnglish, Spanish, French, German, Portuguese, Portuguese (Brazil), Italian, Dutch, Japanese, Korean, Chinese, Hindi, Arabic, Russian, Turkish, Polish, Swedish, Danish, Norwegian, Finnish, Czech, Romanian, Hungarian, Greek, Indonesian, Vietnamese, Slovak, Ukrainian, Bulgarian, Croatian, Malay, Filipino and Tamil.\n\nWhen you enable a language on your study, Koji:\n\n1. Runs the AI interviewer in that language, with a native voice for voice interviews (you can pick a different voice per language)\n2. Translates your questions, including choice options, scale labels and ranking items, when you publish\n3. Shows the interview page, buttons and participant instructions in that language\n\n## Setting Up a Multilingual Study\n\n### Option A: One Study, Several Languages (recommended)\n\nFor most multi-market research, one study is enough. Turn on every language you need and let each participant choose.\n\nSteps:\n\n1. Create a new study in Koji and write your research brief\n2. Open the **Context** tab and find **Interview Languages**\n3. Tick each language you want to offer, and use **Make default** on the one most participants will need\n4. Publish. Koji translates every question into each enabled language, including answer options, scale labels and ranking items\n5. Use the **Viewing in** toggle on the brief to review the questions in each language before you share the link\n6. Run a test interview in each language to check that the AI sounds natural\n7. Share one interview link. Participants pick their language from the menu on the welcome page\n\n![Koji interview welcome page with the language menu open, showing English, Spanish, French, German and more](/email/shots/shot-language-picker.jpg \"With several languages enabled, each participant picks theirs on the welcome page\")\n\nParticipants read translated options, but their answers are recorded against your original option. A German participant who taps the German label for \"Too expensive\" is counted under \"Too expensive\", so charts and counts stay comparable across languages without any merging on your side.\n\nYou can also review and edit the landing page copy for each language in **Texts & Translations** (under Branding & Settings), where each text is marked as machine translated or reviewed.\n\n### Option B: Separate Studies per Market\n\nSeparate studies still make sense when markets need different questions, different screeners or different incentives, or when separate teams own each market. Duplicate your base study, adjust it for the market, and set that market's language in **Interview Languages**. You get a separate report per market and can compare them side by side.\n\nYou can import participants for each study via CSV, and personalized links take each participant to the right study. See [Importing Participants via CSV](/docs/importing-participants-csv) for the import workflow.\n\n### Option C: Setting the Language From Your Own Product\n\nIf you start interviews through the Koji API from your own product, you can pass the participant's locale when the interview starts, so they get the interview in their language without choosing it.\n\n## Writing Research Briefs for Multilingual Studies\n\nThe research brief is the most important document in any Koji study. For multilingual work, a few additional considerations apply:\n\n**Localize your brief, do not just translate it.**\nDirect translation of English research questions often sounds unnatural in other languages. Idioms, phrasing, and conceptual framing vary across languages and cultures. Review the translated questions in each language with the **Viewing in** toggle, and ask a native speaker to read them before launch where you can.\n\n**Be explicit about cultural context.**\nIf you are researching a behavior that differs across cultures, such as how different markets approach financial decisions, privacy expectations, or relationship-driven buying, add that context to your brief so the interviewer can probe for it.\n\n**Consider formality register.**\nSeveral languages have formal and informal registers: Japanese (keigo vs. casual), German (Sie vs. du), French (vous vs. tu), Korean (formal vs. informal speech levels). Specify the register in your brief. B2B research in Japanese typically requires formal keigo; consumer research for a younger audience may be more casual.\n\n**Localize screening questions.**\nScreening questions are translated along with the rest of your questions. Check them in each language, because a misread screener lets the wrong people in.\n\n## Analyzing Multilingual Results\n\n### Transcripts in the Participant's Language\n\nTranscripts are kept in the language the interview was held in. If a participant answered in Spanish, you see a Spanish transcript, and supporting quotes in the report stay in the participant's own words.\n\n### Choosing the Report Language\n\nReports and interview analyses (summaries, themes, key findings) are written in the study's own language by default. You can pick any supported language instead, not only the interview languages: a Dutch study run by an English-speaking team can have its report in English.\n\nSet it with the **Report language** picker on the Report page, beside Regenerate, or in the Languages section of the Context tab. If you change it after a report exists, the report is marked out of date so you can regenerate it in the new language.\n\nThis is the common workflow for global teams: interview people natively in their own languages for honest answers, then read and share the report in the team's working language.\n\n### Cross-Language Synthesis\n\nWith one multilingual study, every language feeds the same report, so themes are found across all participants and the quotes behind each one keep their original language. When you need to see markets apart, run separate studies per market (Option B) and compare their reports, or look at the transcripts for one language group.\n\nBefore comparing rating scores across markets, test [measurement invariance](/docs/measurement-invariance-comparing-groups): accurate translation alone does not guarantee that a rating means the same thing in every language.\n\n## Use Cases for Multilingual Research\n\n**International market expansion:**\nBefore entering a new market, run discovery interviews with potential customers in their native language. Understand their current alternatives, pain points, and buying criteria without the filter of a second language distorting your data.\n\n**Localization validation:**\nAfter translating your product or marketing materials, interview users in the local language to verify whether the translation reads naturally or sounds awkward and foreign.\n\n**Global employee research:**\nFor organizations with international workforces, run employee experience research in employees' native languages. Non-English speakers are often underrepresented in HR, culture, and engagement research.\n\n**Multilingual customer feedback programs:**\nBuild a continuous research pipeline that interviews customers in their language after key touchpoints (onboarding, support interactions, renewals), then read the findings in one report language.\n\n**Academic and institutional research:**\nFor research institutions studying cross-cultural questions, Koji lets you deploy one interview protocol across several language markets at once, with consistent methodology.\n\n## Audio Quality Tips for Non-English Voice Interviews\n\nVoice interview quality varies somewhat by language. A few practical notes:\n\n- **Test each language before launch.** Run a test interview in each enabled language to check the naturalness of speech and the voice.\n- **Some accents and dialects may reduce transcription accuracy.** For strongly regional accents, keep text mode available as a fallback.\n- **Encourage headphones.** Headphones improve audio quality in any language.\n- **Consider time zones when monitoring.** Participants in different markets may be active at very different hours.\n\n## Getting Started with Multilingual Research\n\nTo run your first multilingual study in Koji:\n\n1. Create a new study and write your research brief\n2. Turn on your languages in **Context → Interview Languages** and choose a default\n3. Publish, then review the translated questions with the **Viewing in** toggle\n4. Run a test interview in each language\n5. Share your link, and set the **Report language** your team reads in\n\nWith tools like Koji, multilingual research is no longer a specialized, resource-intensive effort reserved for large enterprise research teams. It is a standard capability available to any team that wants to understand customers wherever they are.\n\n## Further reading on the blog\n\n- [B2B Customer Research: The Complete Guide for Product Teams (2026)](/blog/b2b-customer-research-guide-2026) — B2B customer research is harder than B2C — you are navigating buying groups of 10+ stakeholders, gatekeepers, and enterprise procurement cyc\n- [B2C User Research: How to Understand Consumer Behavior at Scale (2026)](/blog/b2c-user-research-guide-2026) — B2C user research is systematically underinvested at most consumer companies. While B2B teams run structured customer discovery as a matter\n- [Getting Started with Customer Research: A Beginner's Guide](/blog/getting-started-with-customer-research-a-beginner-s-guide) — A practical, step-by-step guide for Product Managers, UX Researchers, and Founders who want to start doing customer research today and build\n\n<!-- further-reading:blog -->\n","category":"guides","lastModified":"2026-10-29T13:18:39.74753+00:00","metaTitle":"Multi-Language User Research: How to Run Interviews in Any Language — Koji Docs","metaDescription":"How to configure Koji for multilingual user research. Covers supported languages, brief localization, single vs. multi-market study setup, cross-language synthesis, and best practices for international research.","keywords":["multilingual user research","international user interviews","multi-language research","user research in Spanish French German","global user research","non-English research"],"aiSummary":"Koji runs AI voice and text interviews in 30+ languages. One study can offer several languages: Koji translates every question, option and scale label on publish, participants pick their language on the welcome page, and answers are recorded against the original options so results stay comparable. Reports and analyses are written in the study language by default, or in any supported language chosen with the Report language picker. The guide also covers localizing briefs, formality registers, separate studies per market and cross-language synthesis.","aiDifficulty":"intermediate","aiEstimatedTime":"12 minutes"},{"type":"documentation","id":"ba66cfe6-7daa-49a9-8676-1f7e93d13882","slug":"lead-user-method","title":"The Lead User Method: How to Find Customers Who Are Ahead of the Market","url":"https://www.koji.so/docs/lead-user-method","summary":"The lead user method, developed by Eric von Hippel at MIT, finds people who face a need months or years before the rest of the market and who benefit enough from solving it that many have built their own fixes. Teams pick a trend, search for the people furthest along it (often in analogous fields) by screening or pyramiding, interview them about what they have built, and develop concepts with them in a workshop. At 3M, ideas from the average lead user project were forecast to sell $146 million a year after five years, more than eight times traditional projects (Lilien et al., 2002).","content":"# The Lead User Method: How to Find Customers Who Are Ahead of the Market\n\n**The short answer:** the lead user method finds the small group of people who already face a need the rest of the market will face later, and who care about it enough to have built their own fixes. You identify a trend, search for the people furthest along it, often outside your own market, and learn from the solutions they have already made. Eric von Hippel at MIT developed the method in the 1980s. In a study at 3M, lead user projects produced ideas forecast to sell more than eight times as much as ideas from the company's usual methods.\n\nMost customer research asks typical customers what they need. That works for improving what exists. It works badly for breakthroughs, because typical customers can only describe needs they already feel. Lead users feel them first.\n\n## What is a lead user?\n\nVon Hippel defined lead users by two traits. In his words, lead users \"face needs that will be general in a marketplace, but face them months or years before the bulk of that marketplace encounters them\", and they \"are positioned to benefit significantly by obtaining a solution to their needs\".\n\nBoth traits matter:\n\n- **Ahead of the trend.** Their situation today looks like the market's situation tomorrow. A hospital treating a new class of infection, a bank running into a regulation before it spreads, or a creator already publishing in ten languages.\n- **High expected benefit.** The need costs them enough that they are willing to invest in a solution. That is why so many of them have already built one: a workaround, a script, a modified tool, sometimes a working prototype.\n\nThat second trait is the practical reason the method works. You are not asking people to imagine the future. You are studying solutions that already exist in rough form.\n\nUser innovation is more common than most teams assume. In a study of 153 buyers of outdoor sports products, Christian Lüthje (2004) found that 37% had come up with ideas for new or improved products, and more than 9% had built prototypes or even marketable products. Among 136 firms using printed circuit board design software, Urban and von Hippel (1988) found that 24.3% had developed or built their own version for their own use.\n\n## Lead users vs power users vs early adopters\n\nThese three groups get mixed up, and the difference decides who you recruit.\n\n![Comparison of lead users and power users, showing how they differ in where they come from, what they know and what research with them is good for.](/visuals/lead-users-vs-power-users-comparison/lead-users-vs-power-users-comparison.png \"Power users show how to get more from today's product; lead users show where the market is going\")\n\n- **Power users** are your heaviest current customers. They know your product inside out, and they are the right people for improving it. See [power user interviews](/docs/power-user-interviews).\n- **Early adopters** are the first to buy new products in a category. They adopt solutions; they do not usually create them.\n- **Lead users** are defined by their need, not by their relationship with you. Many have never used your product. The most useful ones often sit in an _analogous field_: a different market that faces an extreme version of the same problem.\n\nThe analogous field is the method's most distinctive idea. In the 3M project that von Hippel, Stefan Thomke and Mary Sonnack described in Harvard Business Review (1999), a team looking for cheaper ways to prevent surgical infections learned from specialists in veterinary hospitals, who keep infection rates very low, and from Hollywood make-up artists, who work with materials that are gentle on skin and easy to remove. Neither group was a typical buyer of surgical drapes, but both had already solved parts of the problem.\n\n## Why it works: the evidence\n\nThe best-known test of the method comes from 3M. Gary Lilien, Pamela Morrison and colleagues compared lead user projects with traditional idea-generation projects run at the same company during the same period (Management Science, 2002). They found that annual sales of ideas from the average lead user project were conservatively forecast at $146 million after five years, more than eight times the forecast for the average traditional project. A summary of the study published when it came out reported that the lead user ideas also had a higher market share in year five, 68% on average against 33% for other ideas.\n\nTwo caveats are worth stating. The 3M figures are sales forecasts, not realised sales, and lead user projects are more expensive and slower than a round of customer interviews. The method earns its cost when you are looking for a new product line or a new direction, not a feature tweak.\n\n## How to run a lead user study\n\nThe steps below are a practical sequence built from how lead user projects are usually run. Expect weeks, not days, for a full project.\n\n![Process showing five steps of a lead user study: set the scope, find the trends, search for lead users, interview them, and build concepts together.](/visuals/lead-user-method-five-step-process/lead-user-method-five-step-process.png \"Most of the time goes into steps 2 and 3: knowing the trend and finding the people furthest along it\")\n\n### 1. Set the scope\n\nAgree on the market or problem area, the kind of result you want (a new product line, a service, a new strategy) and how far ahead you want to look. Lead user work is generative. If the team has already chosen the solution, run concept testing instead.\n\n### 2. Find the trends\n\nList the trends that will change what people in your market need: new regulations, new technology, new behaviour, rising costs. Talk to experts and read widely. The aim is to pick one or two trends that matter, because lead users are only \"lead\" relative to a trend. Someone who is ahead on cost pressure may be behind on automation.\n\n### 3. Search for lead users\n\nThere are two ways to find them.\n\n**Screening** means surveying many people and picking out those who score high on both traits: how far along the trend they are, and how much they would gain from a solution. Screening works when lead users are reasonably common in a population you can reach.\n\n**Pyramiding** (also called networking) means asking each person you speak to who they know that is even further ahead, and following the chain. It relies on the fact that people deep in a field tend to know who is deeper still. In four experiments, von Hippel, Franke and Prügl (Research Policy, 2009) found that pyramiding located the single most expert person in a group after interviewing about 28.4% of the group on average, and one of the top three after about 15.9%. When lead users are rare, pyramiding is usually far cheaper than screening everyone.\n\nGood pyramiding questions:\n\n- \"Who do you know who has gone further with this than you have?\"\n- \"Who do people in your field go to when they hit this problem?\"\n- \"Is there another industry that deals with a more extreme version of this?\"\n\n### 4. Interviewing lead users\n\nLead user interviews are about what the person has done, not what they would like. Ask them to walk through the problem as they experience it today, what they have tried, and what they built or changed. Ask to see the workaround. Probe for why they made each choice, what it cost them, and what still does not work.\n\n![A maker at a workbench showing a homemade prototype, the kind of workaround a lead user interview sets out to find](https://images.unsplash.com/photo-1590880795696-20c7dfadacde?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHwxfHx3b3Jrc2hvcCUyMHdvcmtiZW5jaCUyMGhhbmRtYWRlJTIwcHJvdG90eXBlfGVufDF8MHx8fDE3OTE2OTc4MTF8MA&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Adam Patterson](https://unsplash.com/@adampatterson?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\nTreat their solution as evidence, not as the answer. A lead user's homemade tool shows you the need and some of the design constraints; your job is to work out the version the wider market will want in a few years.\n\n### 5. Build concepts together\n\nTraditional lead user projects end with a workshop where a handful of lead users, often from different fields, work with the company's team to develop concepts. This is where the ideas get combined and stress-tested. Agree on terms for any ideas they contribute before the workshop, not after.\n\n## Common mistakes\n\n- **Recruiting power users and calling them lead users.** Heavy use of your product is not the same as being ahead of a trend.\n- **Skipping the trend work.** Without a clear trend, \"lead\" has no direction and you end up with a list of enthusiasts.\n- **Looking only in your own market.** The most original solutions often come from analogous fields.\n- **Asking for ideas instead of evidence.** \"What should we build?\" gets wish lists. \"Show me what you built\" gets insight.\n- **Treating lead users as representative.** By definition they are not. Validate the resulting concepts with mainstream customers before you invest.\n\n## How Koji helps\n\nKoji runs AI-moderated interviews by voice, text or phone call, and it fits the parts of a lead user project that need many conversations run the same way: screening, pyramiding and the first round of lead user interviews. The concept workshop stays with your team.\n\n- **Screen for both traits at scale.** Add screening questions that check whether someone works in the area at all, then use scale questions inside the interview to measure how far along the trend they are and how much the problem costs them. Every answer lands in the report, broken down by option, so you can see who scores high on both. The [structured questions guide](/docs/structured-questions-guide) covers the question types.\n- **Pyramid in the same conversation.** End each interview with an open-ended question asking who is further ahead and which other fields face a harder version of the problem. Set the follow-up depth so the interviewer asks why they named that person or field.\n- **Interview the people you are referred to.** Import nominated contacts from a CSV and each person gets their own private interview link. Interviews run in 30+ languages, which helps when the best lead users are in another country or another industry.\n- **Dig into what they built.** Set follow-ups on the \"what have you tried or built\" question, with your own instructions, so the interviewer asks what the workaround cost, what it does and what still breaks.\n- **Separate lead users from everyone else.** The report's personas group participants by shared needs and behaviour, and every finding links to the interviews behind it, so the small group ahead of the trend does not get averaged away.\n\n![Koji question editor showing open ended, 1 to 10 scale and single choice questions with follow-up settings](/email/shots/shot-question-types.jpg \"Open questions and scales in one interview let you screen and learn at the same time\")\n\nThe trade-off is a fair one to name. A skilled human interviewer can follow an unexpected referral chain on the spot and judge expertise in a way a script cannot. Koji's strength is reach and consistency: you can screen hundreds of people, in their own language and their own time, and spend your team's hours on the dozen who turn out to be ahead.\n\n## Related Resources\n\n- [Power user interviews](/docs/power-user-interviews)\n- [Non-user research](/docs/non-user-research)\n- [Design thinking research](/docs/design-thinking-research)\n- [Screener questions guide](/docs/screener-questions-guide)\n- [Concept testing methodology](/docs/concept-testing-methodology)\n- [Structured questions guide](/docs/structured-questions-guide)\n\n## Sources\n\n- Eric von Hippel, \"Lead Users: A Source of Novel Product Concepts\", _Management Science_ 32(7), 1986.\n- Glen L. Urban and Eric von Hippel, \"Lead User Analyses for the Development of New Industrial Products\", _Management Science_ 34(5), 1988.\n- Eric von Hippel, Stefan Thomke and Mary Sonnack, \"Creating Breakthroughs at 3M\", *Harvard Business Review*, 1999.\n- Gary L. Lilien, Pamela D. Morrison, Kathleen Searls, Mary Sonnack and Eric von Hippel, \"Performance Assessment of the Lead User Idea-Generation Process for New Product Development\", _Management Science_ 48(8), 2002.\n- Christian Lüthje, \"Characteristics of innovating users in a consumer goods field\", _Technovation_ 24(9), 2004.\n- Eric von Hippel, Nikolaus Franke and Reinhard Prügl, \"Pyramiding: Efficient search for rare subjects\", _Research Policy_ 38(9), 2009.\n","category":"Research Methods","lastModified":"2026-10-29T13:18:39.478551+00:00","metaTitle":"Lead User Method: How to Find and Interview Lead Users | Koji","metaDescription":"The lead user method finds people who face tomorrow's needs today. Learn von Hippel's definition, pyramiding vs screening, and how to run a lead user study.","keywords":["lead user method","lead user research","lead users","von hippel lead user","lead user innovation","pyramiding","analogous field research"],"aiSummary":"The lead user method, developed by Eric von Hippel at MIT, finds people who face a need months or years before the rest of the market and who benefit enough from solving it that many have built their own fixes. Teams pick a trend, search for the people furthest along it (often in analogous fields) by screening or pyramiding, interview them about what they have built, and develop concepts with them in a workshop. At 3M, ideas from the average lead user project were forecast to sell $146 million a year after five years, more than eight times traditional projects (Lilien et al., 2002).","aiPrerequisites":["Experience planning customer interviews","A product or innovation question that looks beyond small improvements"],"aiLearningOutcomes":["Define a lead user using von Hippel's two traits","Tell lead users apart from power users and early adopters","Choose between screening and pyramiding to find lead users","Run lead user interviews that collect evidence of what people built","Know when the cost of a lead user project is worth it"],"aiDifficulty":"advanced","aiEstimatedTime":"10 min read"},{"type":"documentation","id":"8c4c8ca5-8587-4870-8c81-8ca079a483d3","slug":"koji-vs-voicepanel","title":"Koji vs Voicepanel: AI Voice Interview Platform Comparison (2026)","url":"https://www.koji.so/docs/koji-vs-voicepanel","summary":"A side-by-side comparison of Koji and Voicepanel — two AI-native voice interview platforms. Covers modalities, structured questions, MCP integration, panel access, compliance, pricing, and the specific use cases where each wins.","content":"## The short answer\n\n**Koji and Voicepanel are both AI-native customer research platforms — but they're built for different buyers.**\n\n- **Pick Koji** if you're a product team, founder, or in-house researcher who wants a faster path from research question to insight, structured + open-ended question mixing (six question types in one interview), and deep developer integrations (MCP for Claude/Cursor/VS Code, Headless API, REST API).\n- **Pick Voicepanel** if you're a market research team that wants the largest owned consumer panel (30M+ respondents), video interviews, and enterprise compliance certifications (SOC 2 Type II, HIPAA) as table stakes.\n\nBoth platforms run AI-moderated voice interviews. Both auto-transcribe, theme, and summarize. Both let you ask questions of your data (Koji through its MCP connector). The differences are in the workflows around the interview — who you're recruiting, how you're analyzing, and how the platform plugs into the rest of your stack.\n\nThis guide is for buyers actively evaluating both. We'll cover what each platform is built for, where they overlap, and the specific scenarios that should tip the decision.\n\n## What Voicepanel is built for\n\nVoicepanel is an AI-native feedback platform with a heavy market research lineage. Its positioning is \"AI user research & product feedback\" and its largest customer base is teams that previously used traditional MR panels — Toluna, Cint, Prodege — and are upgrading to AI-moderated interviews while keeping access to a panel.\n\n**Voicepanel's headline capabilities:**\n- AI-driven voice interviews with adaptive probing\n- Four modalities supported: chat, voice, video, phone\n- Built-in panel of 30M+ consumers and professionals across 150+ countries\n- Auto-translation across 35+ languages\n- Thematic analysis, highlight reels, and \"ask your data\" workflows\n- SOC 2 Type II, GDPR, and HIPAA compliance\n- Enterprise governance (spending caps, branding, reusable assets)\n- MCP integration for some workflows\n- Pricing starts around $99/month for the Pro plan, with free-tier capped at 50 responses; enterprise pricing is custom and scales with volume and seats\n\n**Where Voicepanel shines:** market research agencies, consumer research teams running ad-hoc concept tests, healthcare and regulated industries that need HIPAA, and anyone who needs to spin up a study without their own respondent list.\n\n## What Koji is built for\n\nKoji is an AI-native customer research platform built for **continuous discovery teams**: product managers, founders, designers, and in-house UX researchers who need to talk to customers weekly without standing up a research operations function.\n\n**Koji's headline capabilities:**\n- AI-moderated voice and text interviews\n- **Six structured question types in a single interview**: open_ended, scale, single_choice, multiple_choice, ranking, yes_no — letting one study capture both qualitative depth and quantitative comparability without splitting it into \"a survey + interviews\"\n- AI Consultant that turns a research goal into a complete interview guide in under 60 seconds\n- Real-time insights as interviews complete (themes, quotes, sentiment, quality scores)\n- MCP connector: ask questions about your study data from Claude or any MCP assistant; inside Koji, every claim in the report links to its quote\n- **MCP server** with first-class support for Claude, Cursor, and VS Code — letting engineers query research data directly from their IDE without leaving their workflow\n- Headless API for embedded interviews and white-label deployments\n- REST API and MCP for automation from your own code or Zapier, n8n and Make; share the public report link in Notion or Slack, or export CSV or JSON (no native connectors or outbound webhooks yet)\n- Built-in panel recruitment plus BYOR (bring your own respondents): recruit from Koji's panel with a live per-respondent quote in credits that you approve before launch, or use CRM-personalized interview links, anonymous interviews, in-product recruiting, and CSV import\n- Transparent pricing: interviews as low as €1 per qualified interview and €3 per qualified voice interview. Start free with 10 credits, no card, then pay as you go with no subscription\n\n**Where Koji shines:** product-led teams that run continuous discovery, founders doing customer interviews on their own audience, B2B teams researching their existing customer base, and engineering organizations that want research data accessible inside Claude / Cursor / their dev tools via MCP.\n\n## Side-by-side capability comparison\n\n| Capability | Koji | Voicepanel |\n|---|---|---|\n| AI voice interviews | Yes | Yes |\n| AI text interviews | Yes (1 credit each) | Yes (chat mode) |\n| Video interviews | No | Yes |\n| Phone interviews | Yes (0.6 credits per connected minute) | Yes |\n| Structured question types | **6 types in one interview** (open, scale, single, multi, ranking, yes/no) | Standard Q&A flow |\n| Multilingual interviewer | 30+ languages, picked by the respondent | 35+ languages |\n| AI guide generation | AI Consultant in <60s | Templates + AI assist |\n| Real-time analysis | Live theme/quote stream as interviews complete | Hours-scale auto-analysis |\n| \"Ask your data\" | Via the MCP connector (Claude, Cursor, any MCP assistant) | Yes |\n| Quality scoring per interview | Yes (auto) | Limited |\n| MCP integration | **First-class — Claude, Cursor, VS Code** | Available |\n| Headless API | Yes | Limited |\n| Webhooks | Not yet (poll the API or use embed events) | Available |\n| Built-in respondent panel | Yes: Koji's panel, quoted per respondent in credits | **30M+ consumers, 150+ countries** |\n| CRM-personalized links | Yes | Yes |\n| Anonymous interview mode | Yes | Yes |\n| Compliance | GDPR | GDPR + SOC 2 Type II + HIPAA |\n| Free tier | 10 credits on signup | 50 responses on Free plan |\n| Starting paid price | From €1 per qualified interview, pay as you go with no subscription | $99/month (Pro plan) |\n\n## Where the two platforms differ in philosophy\n\nThe capability table tells you what each platform does. The deeper difference is *how* they think about a research project.\n\n**Voicepanel treats the research project as the unit of work.** You design a study, you target a panel, you launch, you wait for analysis. The platform is built to be the start-to-finish workspace for a researcher running studies — and it's especially strong when you don't bring your own respondents.\n\n**Koji treats the research question as the unit of work.** A founder typing \"Why are users churning in week 2?\" into Koji gets back a complete interview guide in under a minute, a sharable link they can paste into a churn email, and live insights as responses come in. It's optimized for iteration speed, not project completeness.\n\nThis shows up everywhere:\n\n- Koji's **AI Consultant** can generate a research brief, methodology, screener, and interview guide from a single prompt. Voicepanel uses templates and an AI assist layered onto a traditional setup flow.\n- Koji's **structured questions** let one interview act as a survey *and* a depth interview — capturing NPS, ranked preferences, and open-ended stories in the same conversation. Voicepanel keeps quantitative and qualitative work more separated.\n- Koji's **MCP integration** lets a product engineer in Cursor query \"What are the top three onboarding complaints from users who signed up this month?\" without leaving their editor. The research becomes a first-class data source for AI coding tools.\n\n## When Voicepanel is the better choice\n\nPick Voicepanel if any of these describe you:\n\n- **You want the largest owned panel.** Voicepanel's 30M-respondent panel is a real asset for very large consumer studies. Koji also recruits participants now, from Koji's panel with a live per-respondent quote in credits, so not having a list no longer decides the choice on its own.\n- **You're a market research agency or consumer research team.** The workflows, governance, spending caps, and reusable assets are built for the agency model.\n- **You need HIPAA.** Healthcare research with PHI requires HIPAA-covered infrastructure that Voicepanel currently offers and Koji does not.\n- **You need video interviews.** Koji's AI moderates voice, text, and phone conversations, not video. If you need facial expressions or screen-share moderation, Voicepanel covers that case.\n- **You're running large-volume consumer concept tests.** Hundreds-of-responses ad/creative testing with panel recruitment is squarely Voicepanel's sweet spot.\n\n## When Koji is the better choice\n\nPick Koji if any of these describe you:\n\n- **You're a product team running continuous discovery.** Weekly customer interviews on your own users with insights ready by the next standup is the workflow Koji is built around.\n- **You need both qualitative depth and quantitative comparability.** Koji's six structured question types let one interview produce NPS scores, ranked feature preferences, *and* open-ended stories — without forcing you to also run a separate survey.\n- **You want research data accessible inside Claude or Cursor.** Koji's MCP server is the most complete in the category. Engineers can query research insights directly from their AI coding tools.\n- **You're a founder, solo PM, or small team.** Koji's onboarding is built for \"type a goal, get an interview in five minutes.\" No research-ops setup required.\n- **You want predictable, low-friction pricing.** Interviews are as low as €1 per qualified interview and €3 per qualified voice interview. Start free with 10 credits, then pay as you go. No subscription needed, no volume minimums, no sales process to start.\n- **You want real-time insights.** Themes, quotes, and quality scores appear as each interview completes — not after a batch-analysis delay.\n\n## Cost comparison: what each platform actually costs to run a 50-interview study\n\nLet's price a representative study: 50 voice interviews, US English, your own respondent list.\n\n**Koji:**\n- 50 voice interviews, and you pay only for the ones that qualify\n- As low as €3 per qualified voice interview, so €150 at the best rate\n- No subscription needed to run it, and volume pricing is there when you want it. If you do not have a list, Koji recruits the respondents for you on a live per-respondent quote in credits that you approve first\n- **Total: from €150 for the study, with nothing charged automatically**\n\n**Voicepanel (Pro plan, $99/month):**\n- Pro plan includes unlimited responses (per public pricing pages — exact limits vary)\n- **Total month one: $99 (~€90 at current rates)** before any panel recruitment\n- Add panel recruitment costs if you don't have your own list (priced per respondent)\n\nVoicepanel's flat-fee model wins on volume if you're fielding hundreds of interviews. Koji's pay as you go model wins on flexibility and on smaller, more frequent studies, and includes voice, text and reports in one bundle.\n\n## The honest take\n\nThese platforms aren't direct substitutes — they're solving overlapping problems for different buyers. Voicepanel is upgrading the market research workflow. Koji is collapsing the research-to-insight loop for product teams.\n\nIf you're evaluating both, ask yourself one question: **am I a researcher who needs a complete project workspace, or am I a builder who needs faster insight loops?** Most of the time the answer is obvious in retrospect.\n\nIf you're still unsure, both platforms have free tiers worth trying. Run the same research question through both, compare the interview experience and the resulting report, and the right tool will declare itself.\n\n## Related Resources\n\n- [Best AI Interview Software in 2026: 9 Platforms Compared](/docs/best-ai-interview-software-2026)\n- [Koji vs Listen Labs: AI Interview Platform Comparison (2026)](/docs/koji-vs-listen-labs)\n- [Koji vs Outset — Two AI Interview Platforms, Different Philosophies](/docs/koji-vs-outset)\n- [AI Voice Interviews: The Definitive Guide for 2026](/docs/ai-voice-interviews-definitive-guide)\n- [Structured Questions in AI Interviews](/docs/structured-questions-guide)\n- [Koji MCP Integration Overview](/docs/mcp-overview)\n- [AI Customer Insights Platform: The 2026 Buyer's Guide](/docs/ai-customer-insights-platform)","category":"Comparisons","lastModified":"2026-10-29T13:18:39.201743+00:00","metaTitle":"Koji vs Voicepanel: AI Voice Interview Platform Comparison (2026)","metaDescription":"Koji vs Voicepanel side-by-side: question design, modalities, MCP, panel access, pricing. Which AI voice interview platform fits your team in 2026.","keywords":["koji vs voicepanel","voicepanel alternative","AI voice interview platform","voicepanel pricing","voicepanel review","AI interview software","customer research platform","best AI interview tool 2026"],"aiSummary":"A side-by-side comparison of Koji and Voicepanel — two AI-native voice interview platforms. Covers modalities, structured questions, MCP integration, panel access, compliance, pricing, and the specific use cases where each wins.","aiPrerequisites":["Understanding of AI-moderated interviews","Familiarity with customer research workflows"],"aiLearningOutcomes":["Identify which AI interview platform fits your team","Compare capability and pricing differences","Understand built-in panel vs BYOR tradeoffs","Map MCP and developer integrations to your stack","Estimate monthly cost for a representative study"],"aiDifficulty":"intermediate","aiEstimatedTime":"10 min read"},{"type":"documentation","id":"d7975ad9-5fae-454e-881b-894a991bac96","slug":"koji-vs-marvin","title":"Koji vs Marvin (HeyMarvin): End-to-End AI Interviews vs. Analysis-Only Repository","url":"https://www.koji.so/docs/koji-vs-marvin","summary":"Koji and Marvin (HeyMarvin) sit at different points in the research workflow. Marvin is a qualitative analysis repository that auto-transcribes and codes recordings of interviews you have already conducted; it does not run the live interview. Koji is an end-to-end AI moderator that designs the brief, conducts the voice/text interview with adaptive AI probing, and analyzes the result in real time. Koji wins for teams that want to remove the human moderator bottleneck. Marvin wins for centralizing many feedback sources.","content":"## TL;DR — Koji vs Marvin\n\n**The headline difference:** Marvin (HeyMarvin) is a research repository and analysis layer. It transcribes, codes, and themes interview recordings *you have already conducted*. [Koji](https://koji.so) is an AI moderator — it actually *runs the interview*, then analyzes it. The two tools are not direct substitutes; they sit at different points in the research workflow.\n\n**Pick Koji if** you want to replace your moderator (and your survey tool, and your analysis pass) with one AI-native platform that can run an interview at 3 AM and have themed insights waiting in the morning.\n\n**Pick Marvin if** you already conduct moderated interviews on Zoom or Google Meet and want a repository that ingests recordings, transcribes them, codes them, and lets you query across many studies. You will still need a moderator.\n\n**The combination:** Some teams pair the two — run new interviews in Koji, archive everything in Marvin for cross-study repository search. But for most teams, Koji handles both jobs.\n\n## What Each Tool Actually Does\n\n### Marvin (HeyMarvin) at a Glance\n\nMarvin is positioned as the \"AI-native customer insights platform\" — but more precisely, it is a qualitative data analysis tool and research repository. It connects to 20+ data sources (Pendo, Qualtrics, Google Forms, Zendesk, Intercom, plus PPTs, PDFs, phone recordings, in-person notes) and centralizes them in a searchable, shareable repository.\n\n**Core capabilities:**\n- Auto-transcription of uploaded audio/video recordings\n- AI thematic analysis on transcripts (auto-tagging, theme extraction)\n- \"Deep Research\" prompts for hypothesis testing and report generation across the repository\n- Sentiment scoring on snippets\n- Repository-wide search to find quotes and patterns\n- 20+ integrations with feedback sources\n- Free trial, then tiered pricing (Free, Team, Enterprise)\n\n**What Marvin does NOT do:** Run the interview. There is no AI moderator that conducts a live conversation with a participant and asks adaptive follow-up questions. You still need a human researcher or another tool for the interview itself.\n\n**Best fit:** Mature research teams that already conduct dozens of moderated interviews per quarter and need a single home for transcripts, notes, and feedback from across the organization.\n\n### Koji at a Glance\n\nKoji is an end-to-end AI customer research platform. It designs the brief, runs the interview, analyzes the data, and writes the report — all with AI in the loop.\n\n**Core capabilities:**\n- AI-moderated [voice and text interviews](/docs/ai-voice-interviews-definitive-guide) with adaptive follow-up probing\n- All [6 structured question types](/docs/structured-questions-guide) (open_ended, scale, single_choice, multiple_choice, ranking, yes_no) inside one AI-moderated conversation\n- [AI Consultant](/docs/working-with-the-ai-consultant) that drafts the research brief and picks the methodology framework\n- Methodology-aware probing (Mom Test, JTBD, Customer Discovery, Open Exploration, Lead Magnet)\n- [Real-time insights](/docs/real-time-research-insights), themes, sentiment, and quality scoring\n- [MCP connector](/docs/mcp-overview): ask questions about your study data from Claude or any MCP assistant; inside Koji, every claim in the report links to its quote\n- [Quality gate](/docs/how-the-quality-gate-works) — only interviews scoring 3+ consume credits\n- [Multilingual research](/docs/multilingual-research-guide) in 30+ languages\n- Embed widget, headless API, [MCP integration](/docs/mcp-overview) with Claude, and a REST API for [your own automation](/docs/webhook-setup)\n- CRM-aware [personalized links](/docs/personalized-interview-links) and [CSV import](/docs/importing-participants-csv)\n\n**Best fit:** Founders, PMs, marketing, UX researchers, agencies, and any team that wants research to be a self-serve weekly habit rather than a quarterly project.\n\n**Pricing:** Interviews are as low as €1 per qualified interview and €3 per qualified voice interview. Free tier with 10 credits at signup, no card. After that it is pay as you go, with no subscription. You pay only for the interviews your study actually uses, and volume pricing and plans are there when you want them.\n\n## The Workflow Difference, Visualized\n\nA typical user research project has roughly 7 stages:\n\n1. Define the research question\n2. Design the brief and methodology\n3. Recruit participants\n4. Conduct interviews\n5. Transcribe and code\n6. Synthesize themes\n7. Write and share the report\n\n| Stage | Marvin | Koji |\n|---|---|---|\n| 1. Define research question | DIY | AI Consultant assists |\n| 2. Design brief + methodology | DIY | AI Consultant drafts; you edit |\n| 3. Recruit participants | DIY | Built-in panel recruitment, quoted in credits, or use your own list (CSV/CRM/link) |\n| 4. **Conduct interviews** | **Not supported** — DIY on Zoom/etc. | **AI moderator runs the interview** |\n| 5. Transcribe + code | Yes (auto) | Yes (auto) |\n| 6. Synthesize themes | Yes (auto) | Yes (auto, real-time) |\n| 7. Write + share report | Yes (templates) | Yes ([auto-generated reports](/docs/generating-research-reports), [publishable](/docs/publishing-sharing-reports)) |\n\nThe blank cell at row 4 for Marvin is the entire reason most teams that try Marvin still feel slow: the bottleneck in research is conducting the interview, not analyzing it. Marvin makes step 5–7 faster but does nothing for step 3–4. Koji collapses 3–7 into a single, mostly automated flow.\n\n## Side-by-Side Feature Matrix\n\n| Capability | Koji | Marvin |\n|---|---|---|\n| **Conducts AI-moderated interviews** | Yes (voice + text) | No |\n| **AI follow-up probing in real time** | Yes | N/A (no live interview) |\n| **Structured questions (NPS, ranking, etc.)** | Yes — all 6 types | No (analysis only) |\n| **AI brief designer** | Yes | No |\n| **Auto-transcription** | Yes | Yes |\n| **Auto theme detection** | Yes (real-time) | Yes (post-hoc) |\n| **Sentiment analysis** | Yes | Yes |\n| **Ask your data / repository querying** | Via the MCP connector (Claude or any MCP assistant) | Yes (Deep Research prompts) |\n| **Quality scoring per interview** | Yes (quality gate) | No |\n| **Cross-study repository** | Per study; MCP reads across studies | Org-wide repository (a strength) |\n| **Integrations with Pendo/Zendesk/etc.** | Limited (REST API and MCP) | Yes (20+ integrations) |\n| **MCP integration with Claude** | Yes | No |\n| **Webhooks to Slack/CRM** | Not yet (poll the API or use embed events) | Limited |\n| **Embed widget for in-product recruitment** | Yes | No |\n| **Pricing** | Public: free tier, then from €1 per qualified interview, pay as you go | Tiered, free trial |\n\n## Where Marvin Wins\n\n**1. Org-wide repository for many input sources.** If your customer feedback comes from 6 different tools — Pendo surveys, Zendesk tickets, sales call recordings, NPS responses, in-person notes — Marvin centralizes all of it in one searchable place. This is genuinely useful for large product orgs.\n\n**2. Analyzing existing recording libraries.** Got 200 historical Zoom recordings of customer interviews sitting in Google Drive? Marvin will ingest, transcribe, code, and theme them. It is purpose-built for this archive scenario.\n\n**3. Sales call ingestion.** If you want to mine sales calls from Gong/Chorus for product insights, Marvin's pipeline handles it well.\n\n**4. Mature integration ecosystem.** With 20+ integrations, Marvin slots into existing tooling stacks at large companies more cleanly than newer point solutions.\n\n## Where Koji Wins\n\n**1. Koji actually conducts the interview.** This is the foundational difference. With Marvin, you still need a researcher to schedule, moderate, and record the call. With Koji, you send a link and the AI moderator runs the conversation 24/7 in the participant's preferred language. The biggest cost in research — researcher time — disappears.\n\n**2. Structured + open-ended questions in the same conversation.** Koji is the only platform that supports all 6 [structured question types](/docs/structured-questions-guide) inside an AI-moderated interview. You get an NPS distribution and the qualitative reasons behind every score from a single session. With Marvin, structured measurement requires a separate survey tool.\n\n**3. Methodology-aware AI moderation.** Koji ships with built-in methodology frameworks ([Mom Test](/docs/mom-test-user-interviews), [JTBD](/docs/jobs-to-be-done-interviews), Customer Discovery). The interviewer agent applies framework-specific anti-patterns — for example, refusing to ask hypothetical \"Would you buy this?\" questions in a Mom Test interview. Marvin has no analog because it does not conduct interviews.\n\n**4. Real-time insights, not batch analysis.** As each Koji interview completes, the [Insights Dashboard](/docs/insights-dashboard) updates with new themes, quotes, and quality scores. You do not wait for a batch run. Marvin processes recordings on a delay.\n\n**5. Quality gate pricing.** Koji's [quality gate](/docs/how-the-quality-gate-works) means low-quality conversations (drop-offs, gibberish, ghosters) cost nothing. You only pay for interviews that scored 3+ on the rubric. Marvin charges per processed transcript regardless of quality.\n\n**6. End-to-end speed.** A typical workflow on Marvin: spend a day moderating, a day transcribing/uploading, a day coding/synthesizing. On Koji: spend 15 minutes designing the brief, send the link, and have a [report](/docs/generating-research-reports) draft ready by the next morning.\n\n**7. Self-serve, predictable pricing.** Koji has public pricing: interviews as low as €1 per qualified interview and €3 per qualified voice interview, with a free tier and pay as you go, no subscription needed. Marvin's pricing for paid tiers is request-based and varies by usage and seats — harder to estimate before talking to sales.\n\n## When You Might Use Both\n\nSome teams use Marvin and Koji together rather than choosing:\n\n- **Run new interviews in Koji** — automated voice/text moderation, real-time insights, structured questions\n- **Archive long-term in Marvin** — unify Koji exports with Pendo, Zendesk, Gong, and historical recordings in one repository\n- **Use Marvin for cross-source repository search** — when you need a single answer across years of feedback from many channels\n\nThis works, but it is overkill for most teams. If your only or primary research input is interview conversations, Koji on its own handles analysis and reporting (every claim in the report links to its quote, the [Insights Dashboard](/docs/insights-dashboard) tracks themes, and the [MCP connector](/docs/mcp-overview) lets Claude or any MCP assistant query the transcripts) without needing a separate analysis layer.\n\n## Decision Framework\n\n**Pick Koji if any of these are true:**\n- You want one tool that does the interview AND the analysis\n- You are a founder, PM, marketer, or small UX team\n- You want to ship a research study in 15 minutes without a sales call\n- You want structured + open-ended questions in the same conversation\n- You want AI follow-up probing that adapts in real time\n- Your research budget is under $20K/year\n- Your participants are existing customers (CRM list), or you want Koji to recruit them for you\n\n**Pick Marvin if any of these are true:**\n- You already have a robust moderated interview workflow you do not want to change\n- Your primary need is centralizing feedback from many sources (Pendo, Zendesk, Gong, etc.)\n- You have a backlog of 100+ historical recordings to mine for insights\n- You have a dedicated research ops function managing the repository\n- You need 20+ tool integrations more than you need a moderator\n\n**Pick both if** you have a mature research program with both heavy interview throughput AND many other feedback sources to integrate.\n\n## Common Misconception: \"Marvin is an AI Interview Platform\"\n\nYou will see Marvin sometimes listed alongside Koji, Listen Labs, Strella, and Outset in \"AI interview tools\" comparisons. This is misleading. Marvin's AI is applied to *post-hoc analysis* of interviews you have already recorded. The \"interview\" itself still happens between a human researcher and a human participant on Zoom.\n\nTrue AI interview platforms — like [Koji](https://koji.so), Listen Labs, Strella, and Outset — replace the human moderator with an AI agent that runs the live conversation. If you want to remove the bottleneck of researcher time and scheduling, you need a true moderator, not a repository.\n\n## Pricing Comparison\n\n| Tier | Koji | Marvin |\n|---|---|---|\n| **Free** | 10 credits at signup, no card required | Free tier (limited) |\n| **Entry paid** | Pay as you go, no subscription needed | Team — request pricing |\n| **Mid-tier** | Volume pricing and plans when you want them | Higher tiers — request pricing |\n| **Enterprise** | Custom | Custom |\n| **Per-interview cost** | As low as €1 per qualified interview, €3 per qualified voice interview; only if quality score ≥3 | Per-transcript via plan allowance |\n\nKoji's pricing is predictable: you can model exactly what 50 voice interviews per month costs, which is €150 at the best rate of €3 per qualified voice interview. Marvin's effective cost per interview depends on what plan you land on after the sales conversation.\n\n## Migration: From Marvin-Only to Koji\n\nIf you already use Marvin and want to add a true AI moderator:\n\n1. **Run new interviews in Koji.** Use the [AI Consultant](/docs/working-with-the-ai-consultant) to design the brief, then share the [interview link](/docs/sharing-your-interview-link).\n2. **Export Koji transcripts.** Use [export](/docs/exporting-research-data) to pull CSV or JSON.\n3. **Upload to Marvin** if you want repository-wide cross-source search.\n4. Or, **stay in Koji** for analysis. The report with its linked quotes, the [Insights Dashboard](/docs/insights-dashboard), and the [Themes & Patterns](/docs/understanding-themes-patterns) view cover most of what teams use Marvin for, and the [MCP connector](/docs/mcp-overview) lets Claude or any MCP assistant answer questions about the transcripts.\n\nMost teams that adopt Koji find they no longer need a separate analysis tool for interview data: every claim in the report links to the quote behind it, Themes & Patterns auto-clusters across all completed interviews, and the MCP connector lets Claude or any MCP assistant answer questions about the transcripts.\n\n## The Bottom Line\n\nKoji and Marvin both have AI in the name, but they solve different halves of the research problem. Marvin is excellent at *analyzing* the conversations a human team has already had. Koji is excellent at *having* the conversations in the first place — and it analyzes them too.\n\nIf you are spending most of your research budget on researcher time, scheduling, and moderating — which most teams are — then a true AI moderator like Koji is the higher-leverage bet. If your bottleneck is making sense of feedback from many channels you do not control, Marvin is the better single purchase.\n\nFor most teams, replacing the human moderator with an AI agent is the bigger unlock. That is what Koji does.\n\n[Try Koji free](https://koji.so) — design and ship your first AI-moderated study in 15 minutes.\n\n## Related Resources\n\n- [Best AI Interview Software in 2026](/docs/best-ai-interview-software-2026)\n- [Structured Questions in AI Interviews](/docs/structured-questions-guide)\n- [How AI Follow-Up Probing Works](/docs/ai-probing-guide)\n- [Koji vs Dovetail — End-to-End vs Analysis-Only](/docs/koji-vs-dovetail)\n- [Koji vs Listen Labs](/docs/koji-vs-listen-labs)\n- [AI Note-Taker for User Interviews](/docs/ai-note-taker-user-interviews)\n- [How to Build a UX Research Repository](/docs/research-repository-guide)\n\n\n## Further reading on the blog\n\n- [Koji vs Marvin: Full-Stack AI Research vs Analysis Repository (2026)](/blog/koji-vs-marvin-2026) — Marvin organizes research you've already collected. Koji runs new AI-moderated interviews from scratch. Here's how to choose — and when you \n- [Best UX Research Repository Tools in 2026: The Complete Buyer's Guide](/blog/best-ux-research-repository-tools-2026) — Dovetail, Marvin, Condens, Notably, Airtable — or something that replaces them all? Here's how every leading research repository tool compar\n- [8 Best Dovetail Alternatives in 2026: Why Teams Are Moving to AI-Native Research](/blog/dovetail-alternatives-2026) — The 8 best Dovetail alternatives for 2026 — including why AI-moderated platforms like Koji collapse the entire research lifecycle into one t\n\n<!-- further-reading:blog -->\n","category":"Comparisons","lastModified":"2026-10-29T13:18:38.923115+00:00","metaTitle":"Koji vs Marvin (HeyMarvin): End-to-End AI Interviews vs Analysis Repository","metaDescription":"Koji vs Marvin (HeyMarvin) head-to-head. Marvin analyzes interviews you have already done; Koji runs the interview AND analyzes it. Pricing, features, decision framework.","keywords":["koji vs marvin","marvin alternative","heymarvin alternative","marvin vs koji","marvin comparison","marvin vs dovetail","heymarvin pricing","ai research repository"],"aiSummary":"Koji and Marvin (HeyMarvin) sit at different points in the research workflow. Marvin is a qualitative analysis repository that auto-transcribes and codes recordings of interviews you have already conducted; it does not run the live interview. Koji is an end-to-end AI moderator that designs the brief, conducts the voice/text interview with adaptive AI probing, and analyzes the result in real time. Koji wins for teams that want to remove the human moderator bottleneck. Marvin wins for centralizing many feedback sources.","aiPrerequisites":["Familiarity with user research workflows"],"aiLearningOutcomes":["Understand the workflow difference between AI moderators and analysis repositories","Compare Koji and Marvin feature-by-feature","Apply a decision framework to choose one or both","Estimate per-interview cost on each platform"],"aiDifficulty":"beginner","aiEstimatedTime":"13 minutes"},{"type":"documentation","id":"11c4a330-60ba-4720-a831-9cf2c91acd01","slug":"koji-vs-listen-labs","title":"Koji vs Listen Labs: AI Interview Platform Comparison (2026)","url":"https://www.koji.so/docs/koji-vs-listen-labs","summary":"Side-by-side comparison of Koji and Listen Labs. Listen Labs is positioned as the enterprise AI research platform with managed panel and cross-study knowledge base; Koji is the self-serve AI-native option with public pricing, built-in panel recruitment quoted in credits, structured + open-ended questions in one interview, methodology-aware probing, and MCP integration. Koji wins for most teams under $25K annual research budget; Listen Labs wins for large enterprises with dedicated researchers.","content":"## TL;DR — Koji vs Listen Labs\n\n**Pick Koji if** you want a self-serve, AI-native research platform you can try in 5 minutes, with public pricing as low as €1 per qualified interview and €3 per qualified voice interview, voice + text + structured questions in one interview, and no sales call required. Best for founders, PMs, marketing teams, UX researchers, and any organization that wants research to happen weekly — not as a once-a-quarter event.\n\n**Pick Listen Labs if** you are an enterprise with a dedicated insights team, a $25K–$100K+ annual research budget, a need for a managed 30M+ participant panel, and a desire for a cross-study knowledge base that compounds findings over time across brand, product, and UX research.\n\nThe two tools belong to the same category — AI-moderated interview platforms — but solve different problems. Listen Labs is an enterprise insights platform with a sales motion. Koji is a self-serve AI research platform with public pricing. The right answer depends almost entirely on your company size, budget, and how often you actually want to talk to customers.\n\n## What Both Tools Do Well\n\nListen Labs and [Koji](https://koji.so) both belong to the AI moderator category — software that conducts an interview with a participant, asks intelligent follow-up questions in real time, transcribes the conversation, and synthesizes themes automatically. Both are credible alternatives to traditional methods like 1:1 moderated Zoom interviews or Typeform-style surveys.\n\nIf you have only ever run static surveys, either tool will be a 10x upgrade in insight quality. The real question is which fits your workflow, budget, and the cadence at which you want to run research.\n\n## Listen Labs at a Glance\n\nListen Labs is a Sequoia-backed AI research platform aimed at enterprise insights teams. It positions itself as an \"autonomous market researcher\" that runs the entire research cycle — recruit, conduct, analyze, report — at AI speed.\n\n**Core capabilities:**\n- AI-moderated interviews across voice, video, and text\n- A 30M+ global participant panel with 100+ language coverage\n- Multimodal emotional analysis using Ekman's universal emotions framework across 50+ languages\n- \"Mission Control\" — a cross-study knowledge base that compounds findings over time\n- Fraud detection on participant responses\n- Executive-ready reports including key themes, highlight reels, and slide decks\n\n**Best fit:** Insights teams at large companies running continuous research across brand, creative, product, and UX, with budget to support an enterprise contract.\n\n**Pricing model:** Sales-led. Public references suggest plans starting around $3,000 with separate participant incentive credits, scaling into custom enterprise contracts that typically run $25K–$100K+ annually depending on volume.\n\n## Koji at a Glance\n\nKoji is an AI-native customer research platform built for self-serve adoption. The core promise: design a study in 15 minutes, run interviews at 3 AM, get themed insights as conversations finish.\n\n**Core capabilities:**\n- AI-moderated voice and text interviews\n- All [6 structured question types](/docs/structured-questions-guide) inside an AI-moderated conversation: open_ended, scale (NPS/CSAT), single_choice, multiple_choice, ranking, yes_no\n- An [AI Consultant](/docs/working-with-the-ai-consultant) that designs the research brief and picks the methodology framework (Mom Test, JTBD, Customer Discovery, Open Exploration, Lead Magnet)\n- [Methodology-aware probing](/docs/ai-probing-guide) — the interviewer agent applies framework-specific anti-patterns and laddering rules\n- [Real-time insights](/docs/real-time-research-insights) — themes, quotes, sentiment, and quality scores update as interviews complete\n- [MCP connector](/docs/mcp-overview): ask questions about your study data from Claude or any MCP assistant; inside Koji, every claim in the report links to its quote\n- [Quality gate](/docs/how-the-quality-gate-works) — only interviews scoring 3+ on the rubric consume credits; low-quality conversations are free\n- [Multilingual research](/docs/multilingual-research-guide) in 30+ languages, picked by the respondent on the welcome screen\n- Embed widget, headless API, REST API, MCP integration with Claude (no outbound webhooks yet)\n- CRM-aware [personalized links](/docs/personalized-interview-links) and CSV [participant import](/docs/importing-participants-csv)\n\n**Best fit:** Founders, product managers, marketing teams, UX researchers, agencies, and any team that wants research to be a weekly habit rather than a quarterly project.\n\n**Pricing model:** Self-serve with public pricing. Interviews are as low as €1 per qualified interview and €3 per qualified voice interview. Start free with 10 credits at signup, no card. After that it is pay as you go, with no subscription. You pay only for the interviews your study actually uses, and volume pricing and plans are there when you want them. There is no overage. You pay as you go, and you only ever pay for the conversations that bring insights. If you run out, your study pauses until you top up. Nothing is ever charged automatically.\n\n## Side-by-Side Comparison\n\n| Capability | Koji | Listen Labs |\n|---|---|---|\n| **Pricing transparency** | Public, self-serve | Sales-led, custom |\n| **Free tier** | Yes (10 credits) | No |\n| **Time to first study** | ~15 minutes | Days to weeks (sales cycle) |\n| **Voice interviews** | Yes (3 credits each) | Yes |\n| **Text interviews** | Yes (1 credit each) | Yes |\n| **Video interviews** | No (audio + transcript) | Yes |\n| **Structured questions in interview** | Yes — all 6 types | Limited |\n| **AI follow-up probing** | Methodology-aware | Yes |\n| **AI brief designer** | Yes (AI Consultant) | Limited |\n| **Real-time theme detection** | Yes | Yes |\n| **Ask your data** | Via the MCP connector (Claude or any MCP assistant) | Limited |\n| **Quality gate (free low-quality interviews)** | Yes | No |\n| **Participant panel** | Built-in panel recruitment (quoted in credits) + bring your own (CSV/CRM/link) | 30M+ managed panel |\n| **Multilingual** | 30+ languages | 100+ languages |\n| **Emotional analysis** | Sentiment + quality scoring | Ekman 7-emotion framework |\n| **Cross-study knowledge base** | Per-study reports; MCP reads across studies | \"Mission Control\" enterprise repository |\n| **MCP integration with Claude** | Yes | No |\n| **Headless API + webhooks** | API and MCP yes; outbound webhooks not yet | Limited |\n| **Embed widget** | Yes | No |\n| **[Bring Your Own Key (BYOK)](/docs/bring-your-own-key)** | Yes | No |\n| **Starting price** | Free tier, then from €1 per qualified interview | ~$3,000+ entry |\n\n## Where Listen Labs Wins\n\n**1. Managed participant panel.** If you need 200 SF-based parents of toddlers next Tuesday, Listen Labs's 30M+ network is genuinely useful. Koji also recruits from trusted Koji's panel: you describe the audience, get a live per-respondent quote in credits, and approve the cost before launch. Listen Labs's owned panel and SLAs remain the deeper bench for very large B2C quotas.\n\n**2. Cross-study knowledge base.** Mission Control compounds insights across many studies — a real advantage if you run dozens of studies per quarter and need to surface patterns across brand, product, and UX research. Koji organizes insights per study; the MCP connector lets Claude or any MCP assistant read across your studies.\n\n**3. Multimodal video analysis.** Listen Labs analyzes facial micro-expressions on video using the Ekman universal emotions framework. Koji focuses on voice + text and sentiment scoring rather than video facial analysis.\n\n**4. Enterprise procurement-friendliness.** SOC 2 reports, custom contracts, dedicated success management, panel SLAs — Listen Labs is built for enterprise procurement processes. Koji is moving in this direction (enterprise plans available) but its DNA is self-serve.\n\n## Where Koji Wins\n\n**1. You can actually try it.** Sign up at [koji.so](https://koji.so), get 10 free credits, and ship a real study in 15 minutes. No demo, no NDA, no quote. Listen Labs requires a sales conversation before you can see the product live.\n\n**2. Structured questions inside the interview.** Koji is the only platform that natively supports all 6 [structured question types](/docs/structured-questions-guide) (open_ended, scale, single_choice, multiple_choice, ranking, yes_no) inside an AI-moderated conversation. You get an NPS distribution and themed quotes about why people scored that way — from the same interview. With Listen Labs, structured measurement typically lives in a separate tool.\n\n**3. Methodology-aware AI moderation.** Koji ships with built-in methodology frameworks ([Mom Test](/docs/mom-test-user-interviews), [Jobs to be Done](/docs/jobs-to-be-done-interviews), Customer Discovery, Open Exploration, Lead Magnet). The interviewer agent applies framework-specific anti-patterns — e.g., refusing to ask \"Would you buy this?\" in a Mom Test interview. Listen Labs is more general-purpose.\n\n**4. Quality gate pricing.** Koji's [quality gate](/docs/how-the-quality-gate-works) only consumes a credit when the interview scores 3+ on the rubric. If a participant ghosts after one answer, you pay nothing. This makes it safe to send your study to a wide list. Listen Labs charges per panel participant regardless of conversation quality.\n\n**5. MCP integration with Claude.** Run Koji from inside Claude. Pull a report, kick off a new study, ask questions about the transcripts, all from a chat interface. No competitor in this category ships an MCP server today. See the [MCP Tool Reference](/docs/mcp-tool-reference) for the full list of operations.\n\n**6. 10x or more cheaper for ongoing research.** A small team running monthly customer interviews on Koji pays for the interviews it runs and nothing else, as low as €1 per qualified interview. The same volume on Listen Labs typically requires a $25K+ annual contract. For continuous discovery cadence, Koji's economics are dramatically better.\n\n**7. Embed widget for in-product recruitment.** Drop Koji's [embed widget](/docs/using-the-embed-widget) into your product to recruit interview participants in-context — say, after a user completes onboarding or hits a paywall. Listen Labs does not offer in-product embeds.\n\n## Decision Framework: Which Should You Choose?\n\n**Choose Koji if any of these are true:**\n- You are a startup, founder, or PM running customer interviews yourself\n- Your annual research budget is under $20K\n- You want to run research weekly, not quarterly\n- You need to mix open-ended interviews with NPS/scale/ranked questions\n- Your participants are existing customers in your CRM\n- You want to read insights from Claude, or automate with the API and MCP\n- You care about per-interview unit economics (quality gate matters)\n- You want to start today, not next quarter\n\n**Choose Listen Labs if any of these are true:**\n- You are an enterprise insights team with a $50K+ annual budget\n- You need a managed B2C participant panel of millions\n- You run 50+ studies per quarter and need a cross-study knowledge layer\n- Video facial analysis is a requirement (Ekman emotions framework)\n- Procurement requires a sales-led, contract-based vendor\n- You have dedicated researchers operating the platform full-time\n\n## Real-World Workflows: How Each Tool Looks in Practice\n\n### Customer discovery sprint (50 interviews in 2 weeks)\n\n**On Koji:** [Import](/docs/importing-participants-csv) your customer list as a CSV, generate [personalized links](/docs/personalized-interview-links), email the link, and watch the [Insights Dashboard](/docs/insights-dashboard) populate as participants complete interviews. Total cost: 50 voice interviews, as low as €3 per qualified voice interview, so €150 at the best rate and no subscription required.\n\n**On Listen Labs:** Schedule a kickoff call with your account manager, define the screener, draw participants from the panel, run interviews. Total cost: typically several thousand dollars per study before incentives.\n\n### Continuous discovery (4 interviews/week, 50 weeks/year)\n\n**On Koji:** Set up an always-on study and paste the link in your weekly customer email. Four voice interviews a week is about 200 qualified voice interviews a year, which is €600 at the best rate. The API and MCP connector are included on every plan; outbound webhooks are not available yet, so poll the read endpoint or use the embed events.\n\n**On Listen Labs:** Annual enterprise contract, typically $25K–$100K depending on volume.\n\n### Brand tracker across 5 markets\n\n**On Koji:** Translate the brief into 5 languages with the [multilingual feature](/docs/multilingual-research-guide), launch in each market, view aggregated themes per segment. If you do not have a list in every market, Koji recruits respondents market by market on a live per-respondent quote in credits that you approve first. Affordable for a Series A startup.\n\n**On Listen Labs:** Strong fit if budget exceeds $50K. The managed panel handles recruitment in each market, and Mission Control compounds findings across waves. Likely the better choice for true global brand tracking at enterprise scale.\n\n## Migration Path: Coming from Listen Labs to Koji\n\nIf you are evaluating switching from Listen Labs to Koji, the typical migration is straightforward:\n\n1. **Export your past study briefs** from Listen Labs and feed them to the Koji [AI Consultant](/docs/working-with-the-ai-consultant). It will reproduce the methodology, question plan, and probing rules.\n2. **Import or recruit your panel.** If you used Listen Labs's panel, you can recruit through Koji's built-in panel instead: describe the audience and approve a live per-respondent quote in credits. If you brought your own panel to Listen Labs, just upload the CSV to Koji.\n3. **Reproduce historical reports** by exporting transcripts from Listen Labs and keeping them as context in Claude or any MCP assistant next to the [Koji connector](/docs/mcp-overview).\n4. **Set up automation.** Use the [Koji API](/docs/webhook-setup) from a server job or Zapier, n8n or Make (poll the read endpoint or use the embed events; there are no outbound webhooks yet) and share the public report link in Slack to replicate the cadence Mission Control was producing.\n\nFor most teams under enterprise size, the switch pays for itself in the first quarter.\n\n## The Bottom Line\n\nListen Labs and Koji are both serious AI interview platforms, but they target different ends of the market. Listen Labs is built for enterprise insights organizations with budget for sales-led contracts and a need for managed panels. Koji is built for the rest of us — founders, PMs, marketers, and lean research teams — who want a self-serve platform with public pricing, structured + open-ended questions in the same interview, methodology-aware AI moderation, and the ability to run research weekly without a procurement cycle.\n\nIf you are not sure which fits, the answer is usually Koji. The free tier removes the risk, you can ship a study before lunch, and you can always graduate to Listen Labs later if your team scales into the $50K+ procurement bracket. Most teams that try Koji never need to.\n\n[Try Koji free](https://koji.so) — no sales call required.\n\n## Related Resources\n\n- [Best AI Interview Software in 2026](/docs/best-ai-interview-software-2026)\n- [Structured Questions in AI Interviews](/docs/structured-questions-guide)\n- [How AI Follow-Up Probing Works](/docs/ai-probing-guide)\n- [AI Voice Interviews: The Definitive Guide](/docs/ai-voice-interviews-definitive-guide)\n- [Koji vs Marvin](/docs/koji-vs-marvin)\n- [Koji vs UserTesting](/docs/koji-vs-usertesting)\n- [Plan Comparison Guide](/docs/plan-comparison-guide)\n\n\n## Further reading on the blog\n\n- [Koji vs Listen Labs: AI Interview Platforms Compared (2026)](/blog/koji-vs-listen-labs-2026) — Listen Labs raised $69M and powers enterprise research at brands across the Fortune 500. Koji is the accessible AI-native alternative starti\n- [AI-Moderated Interview Platforms Compared: Which One Actually Works? (2026)](/blog/ai-moderated-interview-platforms-2026) — Not all AI interview platforms deliver real qualitative depth. This guide compares the top AI-moderated interview platforms in 2026 — Koji, \n- [Best AI Tools for UX Research in 2026: The Complete Buyer's Guide](/blog/best-ai-tools-ux-research-2026) — The UX research AI tool landscape has exploded. This guide maps the best AI tools for every phase of the research workflow in 2026 — from pl\n\n<!-- further-reading:blog -->\n","category":"Comparisons","lastModified":"2026-10-29T13:18:38.64082+00:00","metaTitle":"Koji vs Listen Labs: AI Interview Platform Comparison (2026)","metaDescription":"Koji vs Listen Labs side-by-side: pricing, features, recruitment, analysis, and which AI interview platform fits your team and budget.","keywords":["koji vs listen labs","listen labs alternative","listen labs vs koji","listen labs comparison","ai interview platform","listen labs review","listen labs pricing"],"aiSummary":"Side-by-side comparison of Koji and Listen Labs. Listen Labs is positioned as the enterprise AI research platform with managed panel and cross-study knowledge base; Koji is the self-serve AI-native option with public pricing, built-in panel recruitment quoted in credits, structured + open-ended questions in one interview, methodology-aware probing, and MCP integration. Koji wins for most teams under $25K annual research budget; Listen Labs wins for large enterprises with dedicated researchers.","aiPrerequisites":["Familiarity with user research workflows","Basic understanding of AI-moderated interviews"],"aiLearningOutcomes":["Compare Koji and Listen Labs feature-by-feature","Apply a decision framework to choose the right platform","Estimate total cost of ownership for each tool","Map a migration path from Listen Labs to Koji"],"aiDifficulty":"beginner","aiEstimatedTime":"12 minutes"},{"type":"documentation","id":"c1cd0b39-c03a-4acb-8546-df2bfa454d1a","slug":"in-app-ai-surveys-embedded-research","title":"In-App AI Surveys: Embedded Customer Research Inside Your Product","url":"https://www.koji.so/docs/in-app-ai-surveys-embedded-research","summary":"In-app AI surveys are short adaptive interviews embedded inside your product, triggered by user behavior, that capture customer feedback in the moment. Unlike static email surveys, an AI moderator generates contextual follow-up questions from each open-ended answer. Koji ships a drop-in embed widget that runs the full AI interview inside an iframe, supports six structured question types (open_ended, scale, single_choice, multiple_choice, ranking, yes_no), bills from €1 per qualified interview, on pay as you go with no subscription, and clusters responses into themes in real time. Best practices: one root question, AI-moderated probing, a structured anchor, a skip path, and behavior-based triggering — never fire on every page view.","content":"## What is an in-app AI survey?\n\nAn in-app AI survey is a short, conversational research prompt embedded directly inside your product — triggered by a specific user behavior (after onboarding, after a key action, on a pricing page) — that captures feedback in the exact moment of use. Unlike static intercept surveys, modern in-app AI surveys conduct adaptive interviews: the AI asks one question, listens, and generates intelligent follow-ups based on what the user just said.\n\nIf you ship product changes faster than you can run scheduled customer interviews, in-app AI surveys are the highest-leverage way to keep continuous discovery alive. Platforms like Koji make this turnkey — embed a widget on any route, let the AI moderate the conversation, and surface insights in real time without ever scheduling a call.\n\nThis guide covers when in-app AI surveys make sense, how to design one that respects users, the technical implementation, and how Koji's embedded interview widget compares to traditional in-app survey tools like Sprig, Pendo, and Survicate.\n\n## Why in-app AI surveys outperform email surveys\n\nThe classic \"send a survey link after the user does X\" workflow has three structural problems:\n\n- **Recall decay** — 72 hours after the event, users no longer remember the details that matter. By the time the survey lands in their inbox, the specific friction has been forgotten or rationalized.\n- **Self-selection bias** — email surveys mostly hear from two cohorts: the very happy and the very angry. The middle 80% never click through.\n- **Static questions** — traditional surveys ask the same questions to everyone, so you cannot follow up on the answer that actually matters.\n\nIn-app AI surveys fix all three. Triggered at the moment of action, they catch the experience while it is still vivid. Embedded inside the product, they don't need an email open. And because the conversation is AI-moderated, every follow-up question is generated from what the user actually said — not a pre-written branching tree.\n\nIndustry data backs this up. Hotjar's Trends in UX Research shows in-app feedback collects 4–6x more responses than email-based surveys, and Maze's 2025 State of Continuous Discovery report found teams using embedded research touchpoints ship 2.3x more validated features per quarter.\n\n## When in-app AI surveys are the right tool\n\nIn-app AI surveys are the right tool when you need:\n\n1. **Behavior-triggered feedback** — fire the survey after a specific event (first project created, sixth login, churn-intent click).\n2. **Continuous, low-effort discovery** — keep a passive listening channel open without scheduling moderated sessions.\n3. **High-volume qualitative data** — collect dozens or hundreds of short conversations a week, then let the AI synthesize them.\n4. **Mid-onboarding diagnostics** — figure out where users get stuck without re-watching session replays.\n5. **Pre-churn intervention** — catch the user before they cancel, ask them why, and route the response into the CS workflow.\n\nThey are not the right tool for deep 45-minute generative interviews (use scheduled voice calls instead), highly sensitive topics that require real trust-building, or purely statistical NPS/CSAT tracking (use a single-question intercept).\n\n## How Koji's embedded AI interviews work\n\nKoji's embed widget runs the full AI moderator inside an iframe or modal that you drop into your product with a single script tag. Once a user opens the widget, they get the same conversational interview experience you'd build in the Koji dashboard — except the entire session happens inside your app.\n\nUnder the hood, you get:\n\n- **Six structured question types** out of the box: `open_ended` (with AI probing), `scale` (NPS / CSAT), `single_choice`, `multiple_choice`, `ranking`, and `yes_no`. See the [structured questions guide](/docs/structured-questions-guide) for the full taxonomy.\n- **Adaptive AI follow-ups** — every open-ended answer can trigger up to three contextual follow-ups generated by the LLM, not by a hard-coded branch.\n- **Voice or text mode** — text by default for in-app use, with optional voice for desktop-class flows.\n- **Custom branding** — match your product's colors, logo, and tone via the [interview branding settings](/docs/customizing-branding).\n- **Personalized links** — pass `user_id`, `plan`, or any custom attribute as a URL parameter so the AI knows who it is talking to.\n- **Quality gate** — only conversations that score 3+ on Koji's substantive-answer scale consume credits. Drive-by clicks are free.\n\nA 5-question in-app interview starts as low as €1 per qualified interview. Even before any volume pricing, a hundred in-app conversations a month costs well under the €500+ per-seat minimums of dedicated in-app survey tools.\n\n## Designing a great in-app AI survey\n\nThe hardest part is restraint. Every extra question doubles the abandonment rate. A great in-app AI survey usually has:\n\n**1. One root question.** Open-ended, conversational, and tied to the trigger event. \"What were you trying to do here?\" beats \"Rate your experience 1–5.\"\n\n**2. AI-moderated probing.** Let Koji's AI handle the follow-ups. Two well-placed probes extract more insight than ten pre-written questions. See [AI probing guide](/docs/ai-probing-guide) for how it works.\n\n**3. A structured anchor.** Pair the open-ended question with a single scale or yes/no question so the data aggregates cleanly. \"Did this feel fast? Yes/No\" + \"What made it feel that way?\" — see [scale questions guide](/docs/scale-questions-guide) for the pattern.\n\n**4. A graceful skip path.** Always show an \"Ask me later\" button. Forced surveys destroy long-term trust.\n\n**5. A clear value exchange.** Explain why you are asking. \"We are improving onboarding for new teams — your 90 seconds will directly shape it\" outperforms \"Help us improve\" by roughly 3x completion rate.\n\n## Triggering rules that actually work\n\nThe biggest in-app survey mistake is firing on every page view. Calibrate triggers like a sniper:\n\n- **First-time feature use** — fire after the user completes a key action once.\n- **Repeated abandonment** — detect 2+ failed attempts at the same flow and offer the survey.\n- **Sentiment-driven** — route negative NPS scores into an immediate AI follow-up interview.\n- **Time-bounded** — never show the same survey to the same user twice within 30 days.\n- **Cohort-targeted** — use Koji's personalized link parameters to scope surveys to plan tier, signup date, or feature flag.\n\n## Implementation in 10 minutes\n\nKoji ships an embed widget you can drop into any React, Vue, or vanilla site:\n\n1. Build the study in the Koji dashboard with 2–4 questions (one open-ended plus one or two structured anchors).\n2. Publish and copy the embed snippet from [using the embed widget](/docs/using-the-embed-widget).\n3. Add it to the route where you want the survey to appear.\n4. Pass any user context as URL parameters: `?email=...&plan=pro&onboarded=true`.\n5. Set a trigger condition in your own code — modal opens after 30 seconds on the dashboard, or on click of a \"Give feedback\" button.\n\nFor deeper integrations, the [headless API](/docs/headless-api-overview) lets you start interviews server-side and post messages programmatically — useful when you want the conversation to live inside your own chat UI.\n\n## How Koji compares to traditional in-app survey tools\n\nThe legacy category (Sprig, Pendo Feedback, Survicate, Hotjar Surveys) was built around static, multi-question forms with simple branching logic. Koji is built around AI-moderated conversations:\n\n- **AI moderation with adaptive follow-ups** — Koji generates probes on the fly from every open-ended answer. Static tools only branch on pre-defined paths.\n- **Voice + text mode** — Koji supports both inside the same study. Legacy in-app tools are text-only.\n- **Automatic theme tagging and live reports** — Koji clusters responses into themes in real time; legacy tools export raw answers to a BI tool.\n- **Multilingual** — Koji runs interviews in 30+ languages, picked by the respondent on the welcome screen; legacy tools are English-first.\n- **Per-conversation pricing** — Koji bills per conversation, as low as €1 per qualified interview; legacy tools are per-seat at €500+ per month minimum.\n\nKoji's edge is that the same engine that runs your scheduled remote interviews also runs the 60-second in-app intercept, and insights from both flow into one repository (see [research repository guide](/docs/research-repository-guide)). The embed reaches people already in your product; when you need to hear from people who are not, open Panel recruit and Koji finds them across 130+ countries at a price you approve first, in the same study.\n\n## Privacy, consent, and quiet hours\n\nIn-app surveys still need consent. Koji handles this with a built-in intake form (see [intake forms and consent](/docs/intake-forms-and-consent)), fully customizable to match GDPR / CCPA requirements. Set quiet hours so the widget never interrupts active focus work, and never trigger more than one survey per session.\n\nFor enterprise teams operating in regulated environments, the [GDPR-compliant AI user research guide](/docs/gdpr-compliant-ai-user-research) covers data handling, retention, and DPA setup end to end.\n\n## Measuring success\n\nTrack three metrics on every in-app AI survey:\n\n1. **Open rate** — what percentage of triggered widgets get opened. Healthy: 15–30%.\n2. **Completion rate** — what percentage of opens reach the final question. Healthy: 50–70% for 3-question conversations.\n3. **Insight density** — average themes extracted per completed conversation. Healthy: 1.5–3 themes per response.\n\nKoji surfaces all three in the study dashboard automatically. If completion drops below 50%, shorten the survey or rephrase the root question.\n\n## Related Resources\n\n- [Structured questions guide](/docs/structured-questions-guide) — the six question types that power adaptive interviews\n- [Using the embed widget](/docs/using-the-embed-widget) — drop-in implementation for any site\n- [Embed widget reference](/docs/embed-widget-reference) — full configuration API\n- [Personalized interview links](/docs/personalized-interview-links) — pass user context into every conversation\n- [Intercept research guide](/docs/intercept-research-guide) — methodology background and best practices\n- [AI probing guide](/docs/ai-probing-guide) — how Koji generates adaptive follow-up questions","category":"Collecting Responses","lastModified":"2026-10-29T13:18:38.376468+00:00","metaTitle":"In-App AI Surveys: Embedded Customer Research Guide | Koji","metaDescription":"Embed adaptive AI interviews directly inside your product. A complete guide to in-app AI surveys, triggers, design patterns, and implementation with Koji.","keywords":["in-app surveys","embedded customer research","in-product feedback","in-app AI survey","AI feedback widget","continuous discovery","in-app NPS","customer feedback widget"],"aiSummary":"In-app AI surveys are short adaptive interviews embedded inside your product, triggered by user behavior, that capture customer feedback in the moment. Unlike static email surveys, an AI moderator generates contextual follow-up questions from each open-ended answer. Koji ships a drop-in embed widget that runs the full AI interview inside an iframe, supports six structured question types (open_ended, scale, single_choice, multiple_choice, ranking, yes_no), bills from €1 per qualified interview, on pay as you go with no subscription, and clusters responses into themes in real time. Best practices: one root question, AI-moderated probing, a structured anchor, a skip path, and behavior-based triggering — never fire on every page view.","aiPrerequisites":["Active Koji workspace","A study published with 2–4 questions","Ability to add a script tag to your product"],"aiLearningOutcomes":["Decide when an in-app AI survey is the right tool","Design a 2–4 question conversation that gets completed","Pick triggers that respect users and surface high-quality data","Install the Koji embed widget on any web product","Measure open rate, completion rate, and insight density"],"aiDifficulty":"intermediate","aiEstimatedTime":"12 min read"},{"type":"documentation","id":"5f27e9f7-4ec0-4fa6-86e2-b846d694a82f","slug":"how-ai-interviewers-work","title":"How AI Interviewers Work: A Step-by-Step Walkthrough","url":"https://www.koji.so/docs/how-ai-interviewers-work","summary":"An AI interviewer runs one-on-one customer interviews end-to-end by combining four parts: a research brief with methodology rules, a conversation state machine that tracks what is covered, an LLM that drafts each next question, and a post-interview analyser that scores transcripts. Koji's implementation supports voice and text, 30+ languages, configurable probing depth, six structured question types (open_ended, scale, single_choice, multiple_choice, ranking, yes_no), and automatic 1–5 quality scoring (only interviews scoring 3+ consume credits). The main loop is: ask conversationally, listen, decide whether to probe or move on, update state. Real-time aggregation across interviews surfaces themes within the first 5–10 completed conversations.","content":"## The Bottom Line\n\nAn **AI interviewer** is a conversational AI system that runs one-on-one customer interviews from start to finish — opening the conversation, asking questions in a coherent order, listening to each answer, deciding when to probe deeper, recognising when a topic is exhausted, moving on, and gracefully closing. It works by combining four moving parts: (1) a **research brief** that lists priorities and forbidden moves, (2) a **conversation state machine** that tracks what's been covered, (3) a **large language model** (LLM) that drafts each next question, and (4) a **post-interview analyser** that scores the transcript and extracts structured data.\n\nUnlike a survey form, the AI interviewer adapts in real time. If a participant says *\"I tried three workarounds before I gave up,\"* the AI doesn't move to the next question — it asks which workarounds, in what order, and what made each fail. That single behaviour (adaptive follow-up probing) is what makes an AI interviewer worth using over a Typeform survey.\n\nThis article walks through the full lifecycle of an AI-moderated interview, step by step, using Koji's implementation as the reference. By the end, you'll understand exactly what happens between the moment a participant clicks the interview link and the moment a report appears in your dashboard.\n\n## Step 1: The Brief Becomes Runtime Instructions\n\nBefore the AI ever talks to a participant, it loads a structured research brief. In Koji, the brief is generated by an **AI consultant** (a separate agent that interviews *you* about your goal) and contains:\n\n- A `problemStatement` and `decisionToInform` — what the research must help you decide.\n- A `targetParticipant` profile — who should answer.\n- A `methodology` framework — Customer Discovery, Jobs-to-be-Done, Mom Test, Concept Testing, or Custom. Each comes with embedded rules (e.g., Mom Test forbids hypothetical pricing questions; JTBD requires past-behaviour anchoring).\n- An ordered list of `StudyQuestion` objects, each with a type (`open_ended`, `scale`, `single_choice`, `multiple_choice`, `ranking`, or `yes_no`), a priority (`required` vs `optional`), and a probing depth (default 1 follow-up, up to 3).\n- Optional **context documents** — your product one-pager, a competitor matrix, or previous research — that the AI can reference to ask informed questions.\n\nThe brief is loaded as the system prompt for the interview session. Importantly, methodology principles are loaded as **runtime rules**, not just a label. If the brief is \"Mom Test\", the AI is instructed: *Never ask whether the participant would pay $X. Always anchor on what they've done before. Never lead with a feature description.* These rules constrain every LLM call during the conversation.\n\n## Step 2: Mode Selection — Voice or Text\n\nWhen the participant clicks the interview link, they're presented with a choice (configurable per study): voice or text.\n\n- **Voice mode** uses a real-time speech-to-text model (Deepgram or similar) to transcribe the participant's speech, and a text-to-speech engine to play the AI's replies. Voice interviews typically run 8–15 minutes and yield 3–4x longer answers than text in qualitative research.\n- **Text mode** is a chat interface — the participant types, the AI replies. Text interviews tend to run 4–8 minutes and have higher completion rates among participants who are at a computer.\n\nCrucially, the AI moderator behavior is identical across modes. The same brief, same methodology rules, same follow-up logic. The difference is the I/O channel.\n\nIn Koji, voice costs 3 credits per interview and text costs 1 credit. (Voice is more expensive because of speech-to-text and text-to-speech compute costs.)\n\n## Step 3: The Opening — Establishing Context and Consent\n\nThe AI opens with a short framing message:\n\n*\"Hi, thanks for taking 10 minutes to chat. I'm an AI interviewer working with [your company]. Your responses help us understand [the research topic]. We'll cover about [N] topics. There are no right or wrong answers — just curious to hear your experience. Ready when you are.\"*\n\nThis is not boilerplate. The opening pulls from your study config: your brand name, the topic from the brief, the question count, and any consent language you've added in the intake form. Koji also collects optional intake fields (name, email, role, custom screener questions) before the conversation starts so the AI can personalise from the first turn.\n\n## Step 4: The Main Loop — Ask, Listen, Decide, Probe or Move On\n\nThis is the core of the system. For each question in the brief:\n\n1. **Ask the question conversationally.** Even for structured types (scale, ranking), the AI phrases it naturally. A 1–5 scale becomes *\"On a scale of 1 to 5, where 5 is excellent — how would you rate the onboarding experience?\"* A ranking question becomes *\"You mentioned three tools. If you had to pick one to keep, which would it be? And the one you'd drop first?\"* No \"click a radio button\" prompts.\n\n2. **Listen.** Capture the response (full text in chat mode; transcribed text in voice mode).\n\n3. **Evaluate the answer against several criteria:**\n   - Did the participant *actually* answer the question? (Sometimes they pivot.)\n   - Did they reveal something the brief flagged as a probing opportunity? (e.g., a workaround, a \"I wish…\" statement, a number worth quantifying)\n   - Has the configured probing depth been reached?\n   - Is the participant clearly running out of energy?\n\n4. **Decide:**\n   - **Probe** — generate a follow-up that digs into the specific detail. Mom Test follow-ups anchor on past behaviour (\"When was the last time that happened?\"); JTBD follow-ups dig into the trigger event (\"What made you start looking for a solution?\").\n   - **Confirm + move on** — paraphrase the answer back (\"So the main blocker is the data import flow — got it\") and ask the next question.\n   - **Skip** — if a participant signals they have nothing to add, the AI moves on rather than forcing it.\n\n5. **Update conversation state.** Mark the question as covered, optionally with a confidence score. Note which themes have already emerged so subsequent probes don't repeat them.\n\nThis loop runs until either (a) all required questions are covered AND any optional questions worth asking have been asked, or (b) the participant signals they want to stop.\n\n## Step 5: Handling Edge Cases\n\nReal interviews have messy moments. A good AI interviewer handles them:\n\n- **Multi-language switching.** If the participant replies in Spanish while the brief is in English, the AI matches their language for the rest of the conversation. Koji supports 30+ languages.\n- **Repetition.** If the participant says the same thing in answer to question 4 as they did to question 2, the AI acknowledges it (*\"You mentioned earlier the import issue — anything else?\"*) rather than asking blindly.\n- **Refusal.** If the participant declines to answer (\"I can't share that\"), the AI accepts and moves on without pressing.\n- **Tangents.** In `exploratory` mode, the AI follows interesting tangents. In `structured` mode, it gently redirects. In `hybrid` mode (Koji's default), required questions are covered first, then tangents are explored.\n- **Time pressure.** If the participant says they only have 5 minutes left, the AI prioritises required questions and shortens its responses.\n- **Misunderstandings.** If the AI's question was ambiguous, it rephrases rather than re-asks the same words.\n\nThese behaviours come from the methodology framework + a set of generic conversation-handling rules. You don't configure them per question; they're part of the platform.\n\n## Step 6: Closing the Interview\n\nWhen all required (and worthy optional) questions are covered, the AI closes:\n\n*\"That's everything I wanted to cover. Before we wrap up — is there anything I didn't ask that I should have? Anything you want the team to hear?\"*\n\nThis \"open ending\" turn often yields the single most valuable answer of the interview, because participants surface concerns they didn't know fit any earlier question. Koji always includes it unless you explicitly disable it.\n\nThe AI then thanks the participant, optionally redirects to a thank-you page, and the conversation ends.\n\n## Step 7: Post-Interview Analysis\n\nThe moment the interview completes, an **AI analyst** agent runs (in Koji this happens automatically — no manual step):\n\n1. **Quality scoring.** Every transcript gets a 1–5 composite score across five dimensions: relevance (did the answers match the questions?), depth (were answers substantive?), coverage (were all required questions actually covered?), completion (did the participant finish?), and structured-answer quality (did scales/rankings parse cleanly?). Conversations scoring 1 or 2 are flagged as low-quality and **don't consume credits** in Koji.\n\n2. **Structured answer extraction.** For each `StudyQuestion`, the analyser produces a `StructuredAnswer` containing the `structuredValue` (the parsed numeric/categorical answer), the `qualitativeAnswer` (the participant's verbatim quote), and a `confidence` score. So a `scale` question yields a number from 1–5; a `multiple_choice` question yields one or more option IDs.\n\n3. **Theme tagging.** Common patterns across the transcript are tagged. As more interviews complete, themes are aggregated and ranked by frequency.\n\n4. **Quote extraction.** Memorable, on-point quotes are pulled and attached to the relevant theme.\n\nAll of this happens within seconds of the interview ending. You don't wait for a batch job.\n\n## Step 8: Real-Time Aggregation Across Interviews\n\nAs each interview completes, Koji updates the **dashboard** and **insights report** in real time. After 5 interviews, you can already see emerging themes. After 30, the dashboard shows distribution charts for every structured question, theme frequency, sentiment, and a stack of customer quotes per theme.\n\nThis matters because traditional research has a long synthesis tail — you finish interviews on Monday, get insights on Friday. With AI interviewers and real-time analysis, you see patterns while the study is still running and can decide whether to keep recruiting or stop.\n\n## What Makes a Good AI Interviewer (vs. a Bad One)\n\nNot all AI interviewers are built the same. A good one:\n\n- **Asks follow-ups based on what the participant said,** not on a pre-written script.\n- **Knows when to stop probing** — too many follow-ups feel like interrogation.\n- **Mixes structured and open-ended questions** without breaking conversational flow.\n- **Avoids leading questions** — *\"Would you say this is frustrating?\"* is a Mom Test violation.\n- **Respects participant constraints** — language, time, refusal to answer.\n- **Produces machine-readable outputs** so the rest of the stack can build on them.\n\nA bad one:\n\n- Reads questions verbatim regardless of context.\n- Treats every answer the same way (no quality differentiation).\n- Asks \"click a button\" prompts in voice mode.\n- Has no concept of methodology — just generic chat.\n- Outputs only free text, with no structured extraction.\n\nKoji is built around the \"good\" list. The probing-depth setting, methodology framework, conversation modes (structured/exploratory/hybrid), and post-interview structured extraction are all explicit design decisions.\n\n## Real-World Performance: What to Expect\n\nIn benchmark studies of AI vs human-moderated interviews:\n\n- **Completion rate.** AI interviews complete at ~70–85% rate (text) and ~60–75% (voice) — comparable to async human-moderated research.\n- **Average length.** Voice: 8–15 minutes. Text: 4–8 minutes.\n- **Answer richness.** Voice answers average 3–4x longer than text answers in qualitative research (academic studies of voice surveys).\n- **Insight surfacing.** With real-time analysis, teams typically see the first emerging theme within 5–10 completed interviews.\n- **Time to publishable report.** A 30-interview study can go from launch to shareable report in 3–7 days, vs. 3–6 weeks for traditional moderated research.\n\nThese numbers vary by study design, but the order-of-magnitude difference vs. traditional research is real.\n\n## Frequently Asked Questions\n\n**How does an AI interviewer know what to ask next?** It evaluates the participant's last answer against the research brief: which question is next, has the current question's probing depth been used, did the participant reveal something worth following up on, and does the methodology suggest a specific probe (e.g., past-behaviour anchor in Mom Test). The LLM then drafts the next utterance under those constraints.\n\n**Can the AI interviewer ask structured questions like rankings or scales?** Yes. Koji supports six structured question types — open_ended, scale, single_choice, multiple_choice, ranking, and yes_no — asked conversationally but stored as chartable structured values. So a 1–5 scale question is asked naturally and the answer is parsed into a number you can chart.\n\n**What if the participant goes off-topic?** The AI's behaviour depends on the interview mode. In `structured` mode it gently redirects back to required questions. In `exploratory` mode it follows interesting tangents. In `hybrid` mode (the default), required questions are covered first and tangents explored after. You can change the mode per study.\n\n**How does the AI handle a confused or disengaged participant?** It detects short, terse, or disengaged answers (e.g., \"idk\", repeated \"yes\" to open questions) and adapts — sometimes rephrasing the question, sometimes shortening the interview, sometimes ending gracefully. Low-quality conversations are flagged in the quality score and don't consume credits in Koji.\n\n**Is the AI interviewer biased?** Any LLM-based system carries some bias. Koji mitigates this by embedding methodology rules (Mom Test forbids leading questions; JTBD avoids feature questions), exposing every transcript transparently, and letting you flag transcripts for exclusion from the final report.\n\n**Can participants tell they're talking to an AI?** Yes — Koji explicitly identifies the moderator as an AI in the opening turn. Studies show participants are more candid with AI moderators on sensitive topics (compensation, dissatisfaction with employers, dropped subscriptions) precisely because they know it's not a human.\n\n## Related Resources\n\n- [AI Moderated Interviews: How Automated Research Works](/docs/ai-moderated-interviews) — broader category overview\n- [Understanding the AI Consultant](/docs/understanding-the-ai-consultant) — how the brief gets generated before the interview runs\n- [AI Interviewer Tuning Guide](/docs/ai-interviewer-tuning-guide) — getting research-grade voice interviews from Koji\n- [Structured Questions Guide](/docs/structured-questions-guide) — the 6 structured question types Koji supports natively\n- [AI Voice Interviews: The Definitive Guide](/docs/ai-voice-interviews-definitive-guide) — voice mode deep dive\n- [Understanding Quality Scores](/docs/understanding-quality-scores) — how the 1–5 transcript score is computed","category":"guides","lastModified":"2026-10-29T13:18:38.106087+00:00","metaTitle":"How AI Interviewers Work: Step-by-Step Walkthrough | Koji","metaDescription":"A clear, no-hype walkthrough of how an AI interviewer actually works — from the research brief to adaptive follow-ups to the quality score. Built on Koji's real implementation.","keywords":["how ai interviewers work","how does an ai interviewer work","ai interviewer explained","how ai moderates interviews","ai interview process","conversational ai user research","how ai conducts user interviews"],"aiSummary":"An AI interviewer runs one-on-one customer interviews end-to-end by combining four parts: a research brief with methodology rules, a conversation state machine that tracks what is covered, an LLM that drafts each next question, and a post-interview analyser that scores transcripts. Koji's implementation supports voice and text, 30+ languages, configurable probing depth, six structured question types (open_ended, scale, single_choice, multiple_choice, ranking, yes_no), and automatic 1–5 quality scoring (only interviews scoring 3+ consume credits). The main loop is: ask conversationally, listen, decide whether to probe or move on, update state. Real-time aggregation across interviews surfaces themes within the first 5–10 completed conversations.","aiPrerequisites":["Basic familiarity with user or customer research interviews","No technical background required"],"aiLearningOutcomes":["Understand the four moving parts of an AI interviewer","Walk through the eight-step interview lifecycle","Distinguish good AI interviewer design from bad","Know what to expect for completion rates, answer length, and time to insight","Decide when to use structured vs exploratory vs hybrid interview modes"],"aiDifficulty":"beginner","aiEstimatedTime":"12 min read"},{"type":"documentation","id":"56002254-e65f-4fba-b80a-c43d0c103677","slug":"hard-to-reach-participants-research","title":"How to Research Hard-to-Reach Audiences: Executives, B2B Buyers, and Niche Segments","url":"https://www.koji.so/docs/hard-to-reach-participants-research","summary":"Hard-to-reach audiences — executives, B2B buyers, niche specialists — refuse traditional 60-minute research calls but will complete a 10–15 minute async AI interview on their own schedule. Koji's voice and text interview modes, personalized links, and structured question types make it possible to run 50 executive interviews in the time it takes to schedule 5 live calls.","content":"# How to Research Hard-to-Reach Audiences: Executives, B2B Buyers, and Niche Segments\n\nThe biggest barrier to researching hard-to-reach audiences is not access — it is time. Busy executives, enterprise buyers, and niche specialists will not do 60-minute calls, but they will spend 10–15 minutes in a well-designed asynchronous AI interview that fits their schedule. Here is how to make it work.\n\n## The Hard-to-Reach Problem in Research\n\nEvery researcher knows the frustration: you need insights from the people who matter most — the C-suite buyer, the highly specialized engineer, the physician, the small-business owner working 14-hour days — and those are exactly the people who never respond to recruitment emails.\n\nTraditional user research is optimized for participants with flexible schedules and a willingness to sit on a video call. That is a structural bias: the people easiest to recruit are often not the people making purchase decisions or experiencing the problem most acutely.\n\nThe result? Teams end up interviewing proxies. They talk to lower-level users who \"can relay feedback\" from the executive buyer. They run surveys that busy people answer in 30 seconds with minimal thought. They make product decisions based on the people available, not the people who matter.\n\nAccording to research by the Nielsen Norman Group, recruiting hard-to-reach participants is consistently cited as the top challenge for UX researchers in enterprise and B2B contexts. The consequences go beyond inconvenience: products built on proxy research miss the actual decision criteria, resulting in features that look good in demos but fail to close deals.\n\n## Who Counts as \"Hard to Reach\"?\n\nHard-to-reach audiences share a common trait: high opportunity cost. Their time is genuinely scarce or expensive, and the research ask feels disproportionate relative to the perceived benefit.\n\n**C-suite and senior executives.** VPs and above are unlikely to respond to generic recruitment emails or participate in 60-minute moderated sessions. They are valuable informants for pricing research, enterprise positioning, and strategic roadmap validation.\n\n**B2B buying committee members.** Enterprise software purchases involve 6–10 stakeholders on average. Reaching all of them — especially finance approvers or legal reviewers who are not primary users — is notoriously difficult through standard channels.\n\n**Specialists with rare expertise.** Physicians, lawyers, engineers in niche domains, data scientists at specific companies — recruiting 10 people with the right background can take weeks using traditional methods.\n\n**Small business owners.** They are deeply engaged in their work with no administrative support and no structured time for \"meetings.\" Yet they represent a massive market segment whose unmet needs often drive category-defining product decisions.\n\n**Night-shift workers and international participants.** Synchronous scheduling across time zones or work patterns that do not overlap with the research team creates structural exclusion — you study the sample you can reach, not the sample that matters.\n\n## Why Traditional Methods Fail with These Audiences\n\n**Scheduled calls require calendar alignment.** Getting a mutual 45-minute slot with a busy executive can take 2–3 weeks of back-and-forth, and cancellation rates run 20–30%. A study that needs 15 participant calls can easily stretch across two months.\n\n**Video fatigue is real.** Many senior professionals have aggressively reduced their video call commitments. Asking for a Zoom call is a high-friction request that signals a misunderstanding of their constraints.\n\n**Surveys hit a depth ceiling.** Busy participants who do complete a survey rush through it. You get surface-level responses — no follow-up, no depth, no \"why\" behind the numbers.\n\n**Standard incentives do not scale to senior audiences.** The incentives that motivate general population participants ($25–50 gift cards) are meaningless to a VP of Engineering at a large company. And some incentive structures create compliance problems for professionals in regulated industries like healthcare or finance.\n\n## The Async AI Interview Advantage\n\nAsynchronous AI interviews solve the core constraint: they let participants engage on their own schedule, in 10–15 minutes, with a conversational experience that generates richer data than any survey.\n\n**No scheduling overhead.** With platforms like Koji, participants receive a link and complete the interview whenever they have a natural gap — during a commute, between calls, early morning. There is no calendar coordination, no cancellation problem, no timezone friction.\n\n**Natural conversation reduces friction.** Unlike surveys with radio buttons and text fields, AI interviews feel like a dialogue. The AI asks follow-up questions based on responses, just as a skilled human interviewer would. For busy professionals who care about being heard rather than clicking through options, this format is more engaging and generates more thoughtful responses.\n\n**Voice mode for the highest-value participants.** Koji's voice interview mode is particularly effective for executives and specialists who are comfortable speaking but resistant to typing lengthy responses. Koji can also reach them by phone: paste numbers and pick a language for outbound calls, or share a call link so they enter a number and Koji calls back, with the consent disclosure built in. They can complete an interview during a commute or a walk — a form of participation that costs nothing in terms of desk time.\n\n**Shorter perceived duration.** A 12-minute AI conversation feels shorter than a 30-minute survey because it is responsive. Questions adapt to what the participant actually says, so they never answer questions that do not apply to them.\n\n## Structured Questions for Efficient Data Collection\n\nFor hard-to-reach audiences, every question must earn its place. Koji's six structured question types let you mix quantitative efficiency with qualitative depth:\n\n- **Scale questions** (e.g., \"On a scale of 1–10, how critical is this problem for your team?\") give you comparable data across all participants without requiring lengthy written responses\n- **Single choice questions** (e.g., \"Which best describes your role in the buying decision?\") enable fast segmentation and chart-ready data in your report\n- **Open-ended questions** with AI follow-up probing surface the \"why\" behind the numbers — and the AI does the follow-up work automatically\n- **Yes/no questions** create efficient branching and signal depth without demanding explanatory responses\n- **Multiple choice and ranking questions** capture preference data without asking participants to write essays\n\nThis combination means a 10-minute interview with a busy executive yields data that would take a 45-minute moderated session to generate through manual methods — because the AI handles the probing work automatically, every time, without bias.\n\n## Recruitment Strategies That Work\n\nStart with the channel built into Koji. Open Panel recruit on your study's Interviews page and describe the audience you cannot reach: senior title, industry, company size, specialist credential. Koji maps that to panel profiling questions, shows feasibility and the price per completed interview in credits before you launch, and soft-launches about 10% first. Study-level screening questions catch what the panel profile cannot see, and the interviews land in the same study as the people you reach through the warm channels below. Quoting is free; launching is on paid plans. If you prefer to run recruitment yourself, marketplaces like User Interviews, Respondent or Prolific can source a specialist audience: share your Koji link with the people they find.\n\n### Warm Channel Recruitment\n\nThe highest response rate with hard-to-reach audiences comes through warm channels: your own customer list, CSM relationships, LinkedIn connections from a named researcher. A personalized outreach from someone they know beats cold recruitment by 5–10x.\n\nUse Koji's personalized link feature to pre-populate participant names and company details, making the interview invitation feel tailored rather than mass-distributed. Import the list from your CRM or a CSM export and each person gets their own link: you see who responded, the interviewer greets them by name, and the report is already segmented by account, role or seniority. Not every senior conversation should be tracked, though. If candor matters more than knowing who said what, a frank view of your vendor selection, a sensitive topic, share the generic link and leave the lead form off, so responses stay anonymous. If they would rather talk than type, share a call link or have Koji call them. And for the audience you have no list for, Panel recruit above finds them across 130+ countries at a price you approve first. All routes land in the same study and the same report. An invitation that says \"Hi Marcus, we wanted to understand how your team thinks about [specific problem]\" outperforms any generic research panel invitation.\n\n### The \"10 Minutes, No Call\" Pitch\n\nFrame the ask precisely. Do not say \"user interview\" — say \"a 10-minute AI-powered conversation, no scheduling required, complete whenever works for you.\" Executives who would ignore a \"60-minute research call\" invitation will often say yes to something that takes less time than reading a report.\n\n### Participant Advisory Networks\n\nIf you regularly need executive input, consider building a formal research advisory group: 15–20 participants who have agreed to quarterly research participation in exchange for early product access, industry benchmark reports, or a direct line to the product team. AI async interviews are ideal for these groups because the time commitment is minimal and the experience is substantively better than filling out forms.\n\n### Leverage Internal Champions\n\nIn B2B accounts, your internal champion can distribute the interview link to stakeholders you would never reach through cold outreach. Give the champion the generic link with the lead form off if those stakeholders will speak more freely without their name on the transcript, or ask for the names and import them so each person gets a personalized link and you know who said what. A CSM-enabled distribution to a single enterprise account can generate 5–8 participant completions in 48 hours — including the finance approver, the IT lead, and the department head who never appear on your contact list.\n\n### LinkedIn Voice Messages\n\nFor cold outreach to executives, LinkedIn voice messages from a named researcher often outperform text DMs by 30–40%. They signal authenticity and reduce the mass-campaign feeling that makes senior professionals delete most research invitations unread.\n\n## Timing and Incentives for Senior Audiences\n\n**Best send times.** For senior professionals, Tuesday through Thursday mornings before 9am local time consistently generate higher completion rates. Weekend morning sends also perform well — executives often have quieter, more reflective time then that they use for catching up on reading and professional development.\n\n**Charitable donations.** For executives and senior professionals, charitable donations made in their name often outperform cash incentives. A €100 donation to a charity of their choice feels more meaningful than a gift card and avoids the transactional quality that can subtly degrade response authenticity.\n\n**Research report access.** For specialists — physicians, engineers, lawyers — access to the anonymized final research report is a powerful incentive. They want to know what their peers are thinking. Offering a \"research summary shared with all participants\" creates goodwill and signals that you respect their expertise as a contributor, not just a data source.\n\n## Configuring Your Study for Busy Participants\n\nHard-to-reach participants are worth nothing if the data is low quality. Koji's quality gate automatically evaluates interview depth, conversation length, and response completeness — only consuming credits for conversations that score 3 or above on a 5-point scale.\n\nFor executive and specialist audiences specifically, configure your study with:\n\n- **Maximum 8–10 questions** (not 15–20). Respect their time and they will give you depth on the questions you do ask.\n- **Probing depth of 1 follow-up per question** for structured questions, 2 for open-ended questions. Deeper probing works but the interview should still complete in under 15 minutes.\n- **Estimated completion time displayed prominently** on the landing page. Participants who can see \"~12 minutes\" are more likely to start than those facing an uncertain time commitment.\n- **Voice mode enabled by default** for the most senior audiences. Many executives are more articulate verbally than they are in writing, and the voice format signals that this is a real research conversation, not a survey.\n\n## Quality at Scale\n\nOnce you have cracked the recruitment challenge with one hard-to-reach segment, the async AI model scales without degradation. You can run 50 executive interviews in the same time it would take to schedule 5 live calls. The AI moderates each conversation with identical consistency — no interviewer fatigue, no off-script moments, no variance in probing quality based on how interesting the participant seemed.\n\nThe combination of scale and consistency is what makes AI-moderated research particularly powerful for hard-to-reach audiences: you can actually get enough of them to build statistically meaningful patterns, not just anecdotes.\n\n## Related Resources\n\n- [AI Customer Research for Cybersecurity Companies](/docs/ai-research-for-cybersecurity)\n- [Screener Questions for User Research: A Complete Guide](/docs/screener-questions-guide)\n- [Personalized Interview Links: Send Targeted Research Invitations to Every Participant](/docs/personalized-interview-links)\n- [Importing Participants via CSV](/docs/importing-participants-csv)\n- [B2B User Research: How to Interview Enterprise Customers Without the Scheduling Nightmare](/docs/b2b-user-research-guide)\n- [Structured Questions in AI Interviews](/docs/structured-questions-guide)\n- [Research Incentive Strategies: What to Pay and How](/docs/incentive-strategies)\n- [The Network Scale-Up Method](/docs/network-scale-up-method-prevalence)\n\n## Further reading on the blog\n\n- [Koji vs Looppanel: End-to-End AI Research vs Analysis-Only (2026)](/blog/koji-vs-looppanel-2026) — Looppanel is a great AI analysis tool — but it only handles half the workflow. You still need a separate platform to actually run interviews\n\n<!-- further-reading:blog -->\n\n\n## When you need participants you don't have\n\nKoji can recruit them for you. Open the Panel recruit tab on your study's Interviews page and describe who you want to hear from. Koji finds matching people across 130+ countries and shows you the price per completed interview in credits before you launch.\n\n- **Targeting and screening.** Panel profile data filters people before they arrive, and anything the panel cannot see becomes a screening question asked before the interview starts. Anyone who does not qualify is returned to their panel and you are not charged.\n- **Quality.** Every interview passes a quality check. A conversation under a minute or with fewer than three answers is never billed, and interviews scoring below 3 on Koji's quality scale are free.\n- **Reporting.** Panel respondents land in the same study as your own customers. Koji codes every conversation as it comes in and writes the report with the themes, the evidence behind each one, and clear next steps.\n\nSee [Recruiting participants from Koji's panel](/docs/panel-recruitment).","category":"Participant Recruitment","lastModified":"2026-10-29T13:18:37.830594+00:00","metaTitle":"How to Research Hard-to-Reach Audiences: Executives, B2B Buyers & Niche Segments | Koji","metaDescription":"The people hardest to recruit are often the ones who matter most. Learn how async AI interviews from Koji unlock executives, B2B buyers, and niche specialists who won't take a 60-minute call.","keywords":["hard to reach research participants","executive user research","B2B research recruiting","niche audience research","async user interviews","user research recruiting"],"aiSummary":"Hard-to-reach audiences — executives, B2B buyers, niche specialists — refuse traditional 60-minute research calls but will complete a 10–15 minute async AI interview on their own schedule. Koji's voice and text interview modes, personalized links, and structured question types make it possible to run 50 executive interviews in the time it takes to schedule 5 live calls.","aiPrerequisites":["Basic familiarity with user research recruiting","An active Koji study with a defined participant profile"],"aiLearningOutcomes":["Identify the specific barriers that make different audiences hard to reach","Apply async AI interview techniques that work for executives and specialists","Configure Koji study settings to maximize completion from busy participants","Choose appropriate incentives and send-time strategies for senior audiences"],"aiDifficulty":"intermediate","aiEstimatedTime":"12 minutes"},{"type":"documentation","id":"f7c02b16-81a0-453c-aa50-9f3f257767b1","slug":"grand-tour-questions","title":"Grand Tour Questions: How to Open an Interview With the Big Picture","url":"https://www.koji.so/docs/grand-tour-questions","summary":"A grand tour question, a term from James Spradley's The Ethnographic Interview (1979), asks a participant to walk the interviewer through a whole experience, such as a typical day or an entire process, in their own words. Spradley's four types are typical, specific, guided and task-related, and mini tour questions then zoom in on one part. The guide links the approach to cognitive interview research: Geiselman et al. (1985) found about 41 correct facts recalled versus 29.4 in a standard interview, and Köhnken et al.'s 1999 meta-analysis of 42 studies found a large gain in correct details (d = 0.87) with similar accuracy (85% vs 82%). It gives a five-step grand tour to mini tour process, product research examples, mistakes, and how Koji's Opening Approach, per-question follow-ups and probing instructions run the same structure in AI-moderated interviews.","content":"A grand tour question asks a participant to walk you through a whole experience in their own words: \"Walk me through a typical Monday\", \"Tell me everything that happened from the moment you decided to switch providers\", \"Show me how you put the weekly report together.\" The term comes from the anthropologist James Spradley, who used it in _The Ethnographic Interview_ (1979) for the broad, descriptive question that opens an interview and maps the territory. You follow it with **mini tour** questions that zoom in on one part of what you heard. Together they get you the participant's version of events, in their order and their vocabulary, before your own questions have shaped anything.\n\nThis guide covers what grand tour questions are, Spradley's four types, why they work, how to follow up with mini tours, examples for product and UX research, mistakes to avoid and how Koji runs the same structure in AI-moderated interviews.\n\n## What a grand tour question is\n\nSpradley described three main kinds of question in ethnographic interviews. **Descriptive questions** ask people to describe their world. **Structural** and **contrast questions** come later and ask how things are organised and how they differ. Grand tour questions are the widest descriptive question: they ask for an overview of a space, a stretch of time, a process, a sequence of events, a group of people or a set of activities.\n\nA grand tour question has three traits:\n\n- **It is broad.** It covers a whole day, a whole process or a whole event, not one step of it.\n- **It asks for description, not opinion.** \"Walk me through\" and \"tell me about\", not \"what do you think of\".\n- **It lets the participant choose the route.** You learn what they consider the important landmarks, and what they skip.\n\nSpradley paired it with **mini tour** questions, which work the same way on a much smaller part of the experience. If the grand tour was \"Walk me through how you close the books each month\", a mini tour might be \"You mentioned chasing receipts. Walk me through how that part goes.\"\n\n## The four types of grand tour question\n\nSpradley distinguished four kinds. Each suits a different research goal.\n\n![Framework showing James Spradley's four types of grand tour question: typical, specific, guided and task-related, each with an example.](/visuals/grand-tour-question-types-spradley/grand-tour-question-types-spradley.png \"Pick the type that matches what you need to learn · Source: James Spradley, The Ethnographic Interview (1979)\")\n\n1. **Typical.** Ask the participant to generalise about how things usually go. \"Could you describe a typical day on the job?\" Good for a first map of routines and roles.\n2. **Specific.** Ask about one particular occasion. \"Tell me about yesterday, from when you started work.\" Good for real detail, because people remember a real day better than an average one.\n3. **Guided.** Ask the participant to show you around. \"Could you show me the dashboard you use each morning and talk me through it?\" Good for contextual inquiry, field visits and screen shares.\n4. **Task-related.** Ask the participant to do something and explain as they go. \"Could you set up a new project the way you normally would, and tell me what you're doing?\" Good for usability work and process mapping. It sits close to the [think-aloud protocol](/docs/think-aloud-protocol).\n\nIn product research, a useful habit is to start with a typical tour to get the shape of things, then switch to a specific tour (\"Now tell me about the last time that happened\") for the detail. A typical answer tells you what people believe they do. A specific one is closer to what they actually did.\n\n## Why grand tour questions work\n\n**They let memory work the way it wants to.** Asking someone to narrate an event from the start, in order and in their own words, is close to what forensic psychologists do in the cognitive interview, a technique designed to help witnesses remember more. Two of its core instructions are to mentally reinstate the context of the event and to report everything, however minor it seems. In an early lab test by Ronald Geiselman, Ronald Fisher and colleagues (1985), participants interviewed this way recalled about 41 correct facts on average, against 29.4 with a standard police interview.\n\n**The effect holds up across many studies.** A meta-analysis by Günter Köhnken and colleagues (1999) pooled 42 studies with 55 comparisons and nearly 2,500 interviewees. The cognitive interview produced a large increase in correctly recalled details (d = 0.87) and a much smaller increase in incorrect ones (d = 0.28), so accuracy stayed nearly the same: 85% for the cognitive interview against 82% for standard interviews. A later review by Amina Memon, Christian Meissner and Joanne Fraser (2010), covering 46 published articles, found the same pattern. A grand tour isn't a forensic interview, but it borrows the same idea: give people a familiar structure to walk through, and they remember more of it.\n\n**They reduce your influence.** Every narrow question you ask early carries your assumptions about what matters. A grand tour lets the participant raise the topics first. If they talk through a whole week of work and never mention your feature, that is a finding.\n\n**They give you the participant's vocabulary.** People describe their world in their own terms during a tour: \"the Monday scramble\", \"the approval chain\", \"the export dance\". Those phrases make better follow-up questions, survey options and theme labels than your internal names.\n\n**They give you landmarks for the rest of the interview.** Each step the participant mentions is a place you can come back to with a mini tour. Steve Portigal, author of _Interviewing Users_, lists sequence questions such as \"Walk me through a typical day…\" and \"…Then what do you do next?\" among the basic question types an interviewer needs.\n\n## From grand tour to mini tour: a step-by-step process\n\n![Process of five steps from asking a grand tour question to mini tours on the landmarks the participant mentions and a return to the big picture.](/visuals/grand-tour-to-mini-tour-interview-process/grand-tour-to-mini-tour-interview-process.png \"Go broad first, then zoom in on what they raised · Source: Based on James Spradley, The Ethnographic Interview (1979)\")\n\n1. **Ask the grand tour question, then stop.** Keep it open and give it a clear start and end: \"Walk me through what happened from the moment you first heard about the new expense policy until your first claim was paid.\" Then let them talk. Don't interrupt to ask about details yet. See [using silence in interviews](/docs/silence-in-user-interviews).\n2. **Listen for landmarks and note them.** Write down each step, person, tool and feeling they mention, in their words. These are your follow-up list. Mark anything they rushed past, sighed about or said \"that's a whole story\".\n3. **Pick one landmark and take a mini tour.** \"You said the approval chain takes a while. Walk me through how that part works.\" Choose the landmark that matters most to your research question, not the one that's easiest to ask about.\n4. **Ask for a specific instance.** \"Can you tell me about the last time that happened?\" A mini tour on a real occasion turns a general description into a story with details you can use.\n5. **Come back up and move on.** \"Thanks, that's really clear. Going back to the bigger picture, what happened after the claim was approved?\" Return to the overview before taking the next mini tour, so you cover the whole journey and not just the first step.\n\n## Examples for product and UX research\n\n| Research goal       | Grand tour question                                                                           | A mini tour to follow it                                                   |\n| ------------------- | --------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------- |\n| Understand a role   | \"Walk me through a typical week in your job.\"                                                 | \"You mentioned the Thursday planning call. Talk me through how that goes.\" |\n| Onboarding          | \"Tell me everything that happened from signing up to the first time it felt useful.\"          | \"You said setup took a while. Walk me through that part.\"                  |\n| Switching and churn | \"Take me back to when you started looking for something else. What happened, from the start?\" | \"You mentioned a demo with the other vendor. How did that go?\"             |\n| A workflow          | \"Could you show me how you put the monthly report together, start to finish?\"                 | \"When you copy the numbers across, what exactly are you doing?\"            |\n| Buying decisions    | \"Walk me through how your team chose the tool you use now.\"                                   | \"Who else had a say, and when did they come in?\"                           |\n\nNotice that the grand tour never mentions your product or a specific feature. If the participant brings it up, you can mini tour it. If they don't, you've learned where it sits in their world.\n\n## Running a grand tour in a live interview\n\n- **Set the frame.** \"I'd like to understand how this works for you from start to finish, so I'll mostly listen.\" People give longer answers when they know a long answer is wanted.\n- **Give it boundaries.** \"A typical day\" works. \"Your experience with software\" is too vague to walk through. Name a start and an end point.\n- **Use their words in your follow-ups.** If they call it \"the approval chain\", so do you.\n- **Resist the urge to fix things.** If they describe a workaround, don't explain the feature that would solve it. Ask how the workaround came about.\n- **Watch the time.** A good grand tour answer can run for several minutes. Plan your guide so that's welcome, not a problem. See [how long user interviews should be](/docs/how-long-should-user-interviews-be).\n\n![A researcher listening and taking notes while a participant talks through how a typical week at work goes](https://images.unsplash.com/photo-1758691737278-3af15b37af48?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHw5fHx0d28lMjBwZW9wbGUlMjB0YWxraW5nJTIwbm90ZWJvb2slMjBvZmZpY2V8ZW58MXwwfHx8MTc5MjQ3NTQxMXww&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Vitaly Gariev](https://unsplash.com/@silverkblack?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\n## Common mistakes\n\n- **Asking a grand tour and then a list of closed questions.** The tour gives you landmarks. If you ignore them and go back to your script, you lose the main benefit.\n- **Stacking several questions into one.** \"Walk me through your day, and what tools do you use, and what's frustrating?\" splits their attention. Ask one tour at a time.\n- **Starting too narrow.** \"Tell me about the export button\" is a mini tour. Without the big picture first, you can't tell whether the export matters.\n- **Accepting only the typical version.** \"Usually I just...\" is useful, but follow it with \"Tell me about the last time\" to get what actually happened.\n- **Interrupting the tour to probe.** Note the landmark and come back to it. Breaking in early teaches the participant to give short answers.\n\n## How Koji helps\n\nKoji runs AI-moderated interviews by voice, text and phone, in 30+ languages, and you can build the grand tour and mini tour structure straight into the study.\n\n**Set how the interview opens.** The research brief has an Opening Approach field where you describe, in plain instructions, how the interviewer should start. Write \"Open by asking them to walk through a typical week in their role\" and the interviewer turns it into a natural first question.\n\n**Make the grand tour your first question, with room to roam.** Add it as an open-ended question and set its follow-ups. For every question you can let the interviewer ask from none up to three follow-ups and tell it what to probe for, for example: \"Pick the step they spent longest on and ask them to walk through the last time it happened.\" That's a mini tour, asked consistently in every interview.\n\n![Koji voice interview in progress with an AI orb asking a follow-up question and a microphone button](/email/shots/shot-voice-interview.jpg \"The interviewer follows up on a step the participant raised, in their own words\")\n\n**It asks for real examples.** Koji's interviewer is set up to ask about past behaviour rather than hypotheticals and to look for specific examples and stories when it is allowed to probe, which is the same move as switching from a typical tour to a specific one.\n\n**Add topics without leading.** The brief's Topics to Explore field lists areas you want covered. The interviewer works them in naturally, so you don't have to name your feature in the opening question.\n\n**See the vocabulary in the analysis.** When Koji analyses an open-ended answer, it codes themes either with descriptive labels or in the participant's own framing, each tied to a supporting quote. The report then links every finding to the participants and quotes behind it, so the language from the grand tours carries through to what you share.\n\nA survey can't run a grand tour: a long free-text box usually gets short answers, and nobody follows up on the interesting step. A human interviewer can run a brilliant one but not 50 in a week. With an AI interviewer, every participant gets the same open start and the same follow-up rules, and you still read every story in full.\n\n## Related Resources\n\n- [Semi-structured interview guide](/docs/semi-structured-interview-guide)\n- [Ethnographic research](/docs/ethnographic-research)\n- [Contextual inquiry](/docs/contextual-inquiry)\n- [Funnel technique for interview questions](/docs/funnel-technique-interview-questions)\n- [Day in the life research](/docs/day-in-the-life-research)\n- [Structured questions guide](/docs/structured-questions-guide)\n\n## Sources\n\n- Geiselman, R. E., Fisher, R. P., MacKinnon, D. P., & Holland, H. L. (1985). Eyewitness memory enhancement in the police interview: Cognitive retrieval mnemonics versus hypnosis. _Journal of Applied Psychology_, 70(2), 401-412.\n- Köhnken, G., Milne, R., Memon, A., & Bull, R. (1999). The cognitive interview: A meta-analysis. _Psychology, Crime & Law_, 5(1-2), 3-27.\n- Memon, A., Meissner, C. A., & Fraser, J. (2010). The cognitive interview: A meta-analytic review and study space analysis of the past 25 years. _Psychology, Public Policy, and Law_, 16(4), 340-372.\n- Portigal, S. (2010). Deep-dive interviewing secrets (UIE seminar slides). See also Portigal, S. _Interviewing Users_. Rosenfeld Media.\n- Spradley, J. P. (1979). _The Ethnographic Interview_. Holt, Rinehart and Winston.\n","category":"Research Methods","lastModified":"2026-10-29T13:18:37.546767+00:00","metaTitle":"Grand Tour Questions: Types, Examples and Mini Tours (2026)","metaDescription":"What grand tour questions are, Spradley's four types, how to follow up with mini tours, examples for product and UX research, and common mistakes.","keywords":["grand tour questions","grand tour question examples","mini tour questions","spradley descriptive questions","ethnographic interview questions","walk me through questions","user interview opening questions"],"aiSummary":"A grand tour question, a term from James Spradley's The Ethnographic Interview (1979), asks a participant to walk the interviewer through a whole experience, such as a typical day or an entire process, in their own words. Spradley's four types are typical, specific, guided and task-related, and mini tour questions then zoom in on one part. The guide links the approach to cognitive interview research: Geiselman et al. (1985) found about 41 correct facts recalled versus 29.4 in a standard interview, and Köhnken et al.'s 1999 meta-analysis of 42 studies found a large gain in correct details (d = 0.87) with similar accuracy (85% vs 82%). It gives a five-step grand tour to mini tour process, product research examples, mistakes, and how Koji's Opening Approach, per-question follow-ups and probing instructions run the same structure in AI-moderated interviews.","aiPrerequisites":["Basic familiarity with user interviews or discussion guides"],"aiLearningOutcomes":["Define grand tour and mini tour questions","Choose between typical, specific, guided and task-related tours","Explain why open narrative questions improve recall","Follow a grand tour with mini tours on the landmarks a participant raises","Write grand tour questions for product and UX research"],"aiDifficulty":"beginner","aiEstimatedTime":"10 min read"},{"type":"documentation","id":"32b43b78-cdb6-4939-87fa-c178c7704397","slug":"custom-ai-interviewer-persona","title":"Custom AI Interviewer Persona: Brand Voice & Tone for AI-Moderated Research","url":"https://www.koji.so/docs/custom-ai-interviewer-persona","summary":"A custom AI interviewer persona is the configurable voice, tone, formality, and personality of the AI moderator running your interviews. Persona shapes completion rates, answer candour, and brand health. Koji lets you shape persona through three levers: brief-level tone instructions, uploaded company context documents, and branding settings on the interview landing page. The four dimensions worth configuring are formality, pace, warmth, and domain voice. Three common archetypes are the Curious Colleague (consumer/startup default), the Professional Researcher (B2B enterprise, regulated industries), and the Subject-Matter Peer (developer/designer/technical research). Persona is study-level so different research projects under the same account can use different voices.","content":"## The Bottom Line\n\nA **custom AI interviewer persona** is the configurable voice, tone, formality, and personality of the AI that moderates your customer interviews. When you're running research on behalf of a brand — Stripe, Notion, your own startup — the AI shouldn't sound like a generic chatbot. It should sound like a thoughtful interviewer from *your* team. That means matching the formality of your brand voice, using vocabulary your customers already hear from you, and projecting a personality that puts the right kind of participant at ease.\n\nIn Koji, you shape the AI interviewer's persona through three levers: the **brief-level instructions** (high-level tone, formality, must-mentions and must-avoids), the **company context documents** you upload (which the AI reads to understand your product and audience), and the **branding settings** on the interview landing page (logo, colours, intake messaging that primes the participant before the AI says a word). Together, these turn the default friendly-curious-researcher persona into a voice that feels native to your brand.\n\nThis guide explains the personality dimensions worth thinking about, which combinations work for which audiences, and the exact configuration knobs available in Koji.\n\n## Why Persona Matters for AI Interviews\n\nIt's tempting to dismiss persona as cosmetic. It isn't. Three concrete effects:\n\n1. **Completion rate.** When the AI's tone matches the brand the participant signed up with, they stay engaged longer. A formal AI interviewer on a casual consumer app feels off; a chatty AI on an enterprise compliance survey feels unserious. Either way, drop-off rises.\n2. **Answer candour.** Participants are 2x more candid with a moderator who feels like *one of the team* than with a generic third-party voice. Persona signals trust before the first answer.\n3. **Brand health.** Every customer interview is also a brand interaction. A clumsy AI moderator damages the relationship; a thoughtful one strengthens it.\n\nFor B2B research where the participants are your customers (not random panel respondents), persona is especially load-bearing. They're comparing the AI moderator to the human CSM or AE they've already met. The persona either lives up to that bar or undermines it.\n\n## The Four Persona Dimensions Worth Configuring\n\nA good persona is more than \"friendly vs formal\". Here are the four dimensions that actually shape how an AI interviewer feels:\n\n### 1. Formality\n\nOn a spectrum from *very casual* to *very formal*. A SaaS startup interviewing free-trial users wants conversational (\"Got it — what happened after you tried the import?\"). A regulated industry like healthcare or finance wants neutral-to-formal (\"Understood. Could you describe what occurred after the import attempt?\"). Pick the level your brand uses in customer-facing emails and match it.\n\n### 2. Pace\n\nHow quickly does the AI move through questions? *Brisk* personas confirm the answer and move on — good for power users with limited time. *Patient* personas linger on each answer, asking 2–3 follow-ups before progressing — good for empathy interviews or hard-to-reach customer segments. In Koji, pace is controlled by the **probing depth** setting (1 to 3 follow-ups per question) and the **interview mode** (structured = brisk, exploratory = patient, hybrid = balanced).\n\n### 3. Warmth\n\nHow much affect does the AI project? *Cool* personas are professional and neutral; *warm* personas acknowledge feelings (\"That sounds frustrating\") and use lightly conversational connectives (\"Totally — and after that…\"). Warmth is helpful for stay interviews, exit interviews, and any emotionally loaded topic. It's less appropriate for B2B procurement research where buyers expect efficiency.\n\n### 4. Domain Voice\n\nDoes the AI use industry vocabulary correctly? If you're researching developers, \"shipped\" and \"PR\" should land naturally. If you're researching CFOs, \"burn rate\" and \"MRR\" should be used precisely. Domain voice is mostly shaped by the **context documents** you upload to the study — your product one-pager, glossary, or sample interview transcripts.\n\n## Three Persona Archetypes (and When to Use Each)\n\nMost teams converge on one of three archetypes:\n\n<figure class=\"koji-figure\">\n  <img src=\"https://sybpuenocntpoywqhgkf.supabase.co/storage/v1/object/public/blog-images/custom-ai-interviewer-persona/inline-comparison.webp\" alt=\"Comparison table of three AI interviewer persona archetypes, the Curious Colleague, Professional Researcher and Subject-Matter Peer, across formality, pace, warmth, domain voice and best-fit audience.\" title=\"Three AI interviewer persona archetypes compared\" width=\"800\" height=\"690\" loading=\"lazy\" />\n  <figcaption>\n    <span class=\"caption-text\">Three persona archetypes and the audiences they fit best.</span>\n  </figcaption>\n</figure>\n\n**The Curious Colleague.** Casual formality, brisk pace, warm tone, light domain voice. Works for: consumer apps, prosumer SaaS, early-stage startup customer discovery, exit interviews where you want the participant to feel heard. Default starting point.\n\n**The Professional Researcher.** Neutral formality, patient pace, cool tone, precise domain voice. Works for: B2B enterprise research, healthcare or finance regulated industries, market sizing studies. Signals that the research is serious without being cold.\n\n**The Subject-Matter Peer.** Casual-to-neutral formality, brisk pace, cool-to-warm tone, heavy domain voice. Works for: developer research, designer research, technical buyer interviews — any audience where being seen as a peer matters more than being seen as a researcher. The AI uses the right jargon without explaining it.\n\nMost Koji users start with the Curious Colleague archetype, then adjust one or two dimensions as they see real transcripts.\n\n## How to Configure Persona in Koji\n\nThere are three places where persona is shaped:\n\n<figure class=\"koji-figure\">\n  <img src=\"https://sybpuenocntpoywqhgkf.supabase.co/storage/v1/object/public/blog-images/custom-ai-interviewer-persona/inline-framework.webp\" alt=\"Framework diagram of the three places an AI interviewer persona is shaped in Koji: brief-level tone instructions, company context documents, and branding and intake settings.\" title=\"Where AI interviewer persona is configured in Koji\" width=\"800\" height=\"798\" loading=\"lazy\" />\n  <figcaption>\n    <span class=\"caption-text\">Three configuration surfaces that shape the AI interviewer persona in Koji.</span>\n  </figcaption>\n</figure>\n\n### 1. Brief-Level Tone Instructions\n\nWhen you're generating the brief with the AI consultant (or editing it manually), you can add tone-and-style instructions directly:\n\n*\"Tone: casual but precise — match Stripe's developer-facing voice. Use 'shipped', 'rolled out', 'flag' naturally. Avoid corporate-speak ('synergy', 'leverage'). Acknowledge frustration when participants describe friction.\"*\n\nThese instructions are loaded into the AI interviewer's system prompt and apply across every conversation in the study. The more specific, the better. Generic instructions (\"be friendly\") produce generic personas. Specific instructions (\"warm but brisk, use 'ship', avoid 'utilise'\") produce distinctive personas.\n\n### 2. Company Context Documents\n\nUpload your product one-pager, brand voice guide, sample transcripts, or competitor analysis as **context documents** on the study. The AI references these to:\n\n- Use your product's terminology correctly (feature names, plan tiers, key metrics).\n- Match your brand's linguistic register.\n- Understand which customer concerns are already known vs. genuinely new.\n- Avoid pitching the product (research, not sales) while still sounding like part of the team.\n\nIn Koji, every study has a context-documents slot. The AI consultant reads these when drafting the brief; the AI interviewer references them mid-conversation when relevant.\n\n### 3. Branding & Intake Settings\n\nThe interview landing page is the participant's first impression. Configure:\n\n- **Logo and brand colours** — so the participant knows immediately whose research this is.\n- **Custom intake messaging** — a short welcome explaining the goal and what to expect. This primes the participant for the AI's tone.\n- **Consent and intro language** — if your brand has specific data-handling language, include it here.\n- **Voice mode default** — pick whether voice is the default or text is, depending on which feels more on-brand.\n\nWith brand consistency from landing page → intake → AI moderator, participants experience one coherent voice rather than a generic AI dropped into your domain.\n\n## Persona Across the Six Structured Question Types\n\nKoji's AI interviewer asks six native question types — `open_ended`, `scale`, `single_choice`, `multiple_choice`, `ranking`, and `yes_no`. Persona shapes how each is phrased:\n\n- **Open-ended** in a Curious Colleague voice: *\"Walk me through the last time you ran into that — what happened?\"*\n- **Scale (1–5)** in a Professional Researcher voice: *\"On a scale of 1 to 5, where 5 is excellent — how would you rate the experience overall?\"*\n- **Ranking** in a Subject-Matter Peer voice: *\"If you had to keep one tool and drop the rest — which stays, which goes first?\"*\n- **Yes/no** in a Curious Colleague voice: *\"Did you end up shipping that, or did it get parked?\"*\n\nThe structured value extracted is identical regardless of phrasing — but the conversational version is what shapes the participant's experience and (downstream) their willingness to give honest answers.\n\n## Common Mistakes\n\n**Trying to make the AI sound human.** Don't. Participants know it's an AI; pretending otherwise erodes trust. Studies show participants are *more* candid with AI moderators on sensitive topics (compensation, dissatisfaction with employers, dropped subscriptions) precisely because they know it's not a human. Lean into thoughtful AI, not fake human.\n\n**Over-engineering the persona.** Three lines of tone instructions usually beat thirty. Brief, specific, and concrete > a 500-word brand voice manifesto. The AI is good at applying a few constraints; less good at distilling a long abstract document into voice.\n\n**Mismatching persona to topic.** A breezy Curious Colleague voice in a stay-interview about why people are considering leaving the company feels tone-deaf. Match persona to emotional register of the topic, not just to your brand.\n\n**Forgetting branding on the landing page.** The AI's persona doesn't exist in a vacuum. If the participant arrives at an unbranded white page, the AI's personality has no context. Always brand the landing page.\n\n**Setting and forgetting.** Read 5–10 real transcripts after launching a study and adjust the persona based on what you see. Real participant interactions reveal where the AI's tone is misfiring far better than dry-run testing.\n\n## A Worked Example: Configuring Persona for a Developer Research Study\n\nLet's walk through configuring a persona for a study interviewing developers about their CI/CD tool experience.\n\n<figure class=\"koji-figure\">\n  <img src=\"https://sybpuenocntpoywqhgkf.supabase.co/storage/v1/object/public/blog-images/custom-ai-interviewer-persona/inline-numbered.webp\" alt=\"Numbered six-step breakdown for configuring a developer AI interviewer persona in Koji, from picking the Subject-Matter Peer archetype to running test interviews and reading transcripts.\" title=\"Configuring a developer AI interviewer persona in six steps\" width=\"800\" height=\"1113\" loading=\"lazy\" />\n  <figcaption>\n    <span class=\"caption-text\">A six-step loop for configuring a developer research persona.</span>\n  </figcaption>\n</figure>\n\n**Step 1: Pick an archetype.** Subject-Matter Peer — developers want to talk to someone who speaks their language, not a market researcher.\n\n**Step 2: Add tone instructions to the brief:**\n\n*\"Tone: casual peer voice. Use 'shipped', 'rolled out', 'flag', 'main' naturally. Don't explain technical concepts unless the participant asks. Acknowledge engineering frustrations ('yeah, flaky tests are the worst') but don't pile on. Keep questions short — engineers prefer crisp.\"*\n\n**Step 3: Upload context documents.** Add your product one-pager, your engineering blog's last three posts (for voice), and a glossary of CI/CD terminology you want used correctly.\n\n**Step 4: Configure interview mode.** Set to `hybrid` — cover required questions about current toolchain, then free-roam on what they wish existed. Probing depth: 2 follow-ups (engineers volunteer detail without prompting).\n\n**Step 5: Brand the landing page.** Use your engineering team's logo variant and a 1-sentence intro like: *\"Quick chat with our AI interviewer about your CI/CD setup. ~8 min, no sales pitch.\"*\n\n**Step 6: Run 3 test interviews and read the transcripts.** Adjust as needed. If the AI is using too much jargon, dial it back. If it's being too formal, add an explicit \"lean casual\" instruction.\n\nThis 30-minute setup loop produces a persona that feels like *your* engineering team interviewing customers — not a generic AI chatbot.\n\n## Frequently Asked Questions\n\n**Can I configure the AI interviewer's voice (literally — the voice it speaks in) for voice interviews?** Yes. Koji's voice mode supports configurable voice profiles — pick the gender, accent, and pace that match your brand. For most B2B research, a neutral, mid-pitch voice with a measured pace works best. For consumer research, you can lean warmer or more energetic.\n\n**Does persona affect interview quality scores?** Persona shapes *engagement* (completion rates, answer depth) more than it shapes the 1–5 quality score itself. Quality scoring measures whether questions were covered, whether answers were on-topic, and whether structured answers parsed cleanly — these are persona-agnostic. But because better personas drive longer, more candid answers, they indirectly raise quality scores.\n\n**Can I run different personas in different studies under the same account?** Yes. Persona is study-level, not account-level. You can run a Curious Colleague persona on your free-trial onboarding study and a Professional Researcher persona on your enterprise CFO interviews simultaneously.\n\n**What about multilingual studies — does persona translate across languages?** Tone instructions transfer, but specific vocabulary doesn't (industry terms differ across languages). For multilingual studies, upload language-specific context documents or note in the brief that the AI should use the local industry vocabulary in each language.\n\n**Can I clone a persona across studies?** Yes — duplicate a study and the brief (including tone instructions and context documents) copies over. Easiest way to maintain a consistent voice across multiple research projects.\n\n**Is there a default persona if I don't configure anything?** Yes — Koji's default is \"thoughtful, curious, professional researcher\" with neutral formality, balanced pace, mild warmth, and no specific domain voice. It's a sensible starting point but you'll get better engagement by tuning at least the formality and domain voice dimensions.\n\n## Related Resources\n\n- [Customizing Your Study](/docs/customizing-your-study) — study-level configuration including persona\n- [Customizing Branding](/docs/customizing-branding) — logo, colours, and landing page configuration\n- [Uploading Context Documents](/docs/uploading-context-documents) — how to feed the AI your product context\n- [Working with the AI Consultant](/docs/working-with-the-ai-consultant) — shape the brief (including tone instructions) at draft time\n- [AI Interviewer Tuning Guide](/docs/ai-interviewer-tuning-guide) — deeper tuning of conversational behaviour\n- [Structured Questions Guide](/docs/structured-questions-guide) — the 6 question types your persona shapes","category":"Interview Experience","lastModified":"2026-10-29T13:18:37.284895+00:00","metaTitle":"Custom AI Interviewer Persona: Brand Voice for AI Research | Koji","metaDescription":"Shape your AI interviewer's voice, tone, and personality so it sounds like your brand — not a generic chatbot. Personality dimensions, archetypes, and how to configure persona in Koji.","keywords":["custom ai interviewer persona","ai interviewer voice","ai moderator brand voice","custom ai persona research","ai interview tone","brand voice ai interviews","customize ai interviewer"],"aiSummary":"A custom AI interviewer persona is the configurable voice, tone, formality, and personality of the AI moderator running your interviews. Persona shapes completion rates, answer candour, and brand health. Koji lets you shape persona through three levers: brief-level tone instructions, uploaded company context documents, and branding settings on the interview landing page. The four dimensions worth configuring are formality, pace, warmth, and domain voice. Three common archetypes are the Curious Colleague (consumer/startup default), the Professional Researcher (B2B enterprise, regulated industries), and the Subject-Matter Peer (developer/designer/technical research). Persona is study-level so different research projects under the same account can use different voices.","aiPrerequisites":["A Koji study in draft or active state","Some idea of your brand's voice and audience"],"aiLearningOutcomes":["Understand the four persona dimensions: formality, pace, warmth, domain voice","Match persona archetypes to research audiences","Configure persona through tone instructions, context documents, and branding","Avoid common persona mistakes (trying to sound human, over-engineering)","Maintain consistent persona across multilingual or multi-study research"],"aiDifficulty":"intermediate","aiEstimatedTime":"10 min read"},{"type":"documentation","id":"73fb5c04-67ec-4c54-8750-b6f39c037c22","slug":"cross-cultural-user-research","title":"Cross-Cultural User Research: The Complete Guide for Global Product Teams","url":"https://www.koji.so/docs/cross-cultural-user-research","summary":"Cross-cultural user research studies users across countries and cultures so products feel native, not translated. CSA Research found 76% of consumers prefer buying in their native language and 40% never buy from foreign-language sites. This guide covers Hofstede, Hall, and Trompenaars frameworks, a 9-step workflow, common mistakes, and how AI-moderated platforms like Koji collapse the cost and timeline of multilingual studies.","content":"# Cross-Cultural User Research: The Complete Guide for Global Product Teams\n\n**Answer-first (BLUF):** Cross-cultural user research is the practice of studying users across different countries, languages, and cultural contexts to design products that feel native — not translated — in every market. The cost of skipping it is real: **76% of consumers prefer to buy products with information in their native language, and 40% will never purchase from sites in another language** (CSA Research, surveying 8,709 consumers in 29 countries). Modern teams now use AI-moderated platforms like Koji to run multilingual interviews and surveys at scale, replacing the months and tens of thousands of dollars that traditional global studies demand.\n\n## What is cross-cultural user research?\n\nCross-cultural user research is qualitative and quantitative inquiry conducted with participants from two or more distinct cultural contexts — countries, regions, languages, or ethnic groups — to understand how cultural factors shape attitudes, behaviors, and product expectations. It goes well beyond translation. It asks: *Does this product concept, message, flow, or interaction model carry the same meaning, value, and emotional weight across cultures?*\n\nMost teams discover the answer is \"no\" only after launch. Slack famously had to redesign onboarding flows for Japanese users who interpreted message bubbles differently. Airbnb learned that \"trust\" signals that worked in the U.S. (host photos, casual descriptions) read as suspiciously informal in formal-business cultures like Germany and Japan. Uber's early payment flows assumed credit cards in markets where cash and mobile money dominated.\n\nThe pattern is consistent: products that get cross-cultural research right outperform competitors who simply translate their interfaces.\n\n## Why cross-cultural research matters now more than ever\n\nThe business case is no longer abstract. According to research from Common Sense Advisory (now CSA Research), **localized marketing campaigns generate conversion rates up to 6× higher** than non-localized alternatives. Companies investing in language services see ROI increases of up to 120%, and 84% of marketing professionals report that translation and localization directly increased their income.\n\nThree forces are pushing cross-cultural research from \"nice to have\" to existential:\n\n1. **Geographic expansion is the default growth path.** SaaS companies routinely expand to LATAM, EMEA, and APAC within 18 months of finding U.S. product-market fit. The teams that win do so on the back of structured cultural research, not gut intuition.\n2. **AI-native global research is finally cheap.** Until 2024, running a 30-participant interview study across 5 countries required local research agencies, simultaneous interpreters, and budgets north of $50,000. AI-moderated interview platforms have collapsed that to days and a fraction of the cost.\n3. **Cultural missteps go viral.** A poorly-translated marketing campaign or a culturally tone-deaf feature can become a Reddit thread within hours. Pre-launch cross-cultural validation is now insurance.\n\n> \"Designers are required to integrate advanced technical proficiency, creative problem-solving, technological adaptability, and **cultural intelligence** to create inclusive, socially responsible, and market-relevant products.\" — *Research from Liao & Thomas (2025), published in Group & Organization Management*\n\n## The cultural dimensions framework: Hofstede, Hall, and Trompenaars\n\nBefore you can study culture, you need a vocabulary for it. Three frameworks dominate practitioner work:\n\n### 1. Hofstede's Six Dimensions\nGeert Hofstede's cultural dimensions remain the most widely-used reference for global research. The six dimensions to map your participants against:\n\n- **Power Distance Index (PDI):** Tolerance for hierarchy. High in Malaysia, Mexico, China. Low in Denmark, Israel, Austria. Implications: B2B software UX in high-PDI markets needs clear approval workflows; consumer apps need to respect status indicators.\n- **Individualism vs. Collectivism (IDV):** \"I\" vs. \"we\" cultures. The U.S. is the most individualistic country measured (91/100). Guatemala, Panama, and Venezuela are deeply collectivist. Implications: social proof, group plans, and \"join your team\" framing land very differently.\n- **Masculinity vs. Femininity (MAS):** Competition vs. cooperation orientation. Japan scores highest in MAS (95); Sweden lowest (5). Implications: tone, achievement framing, gamification all need calibration.\n- **Uncertainty Avoidance Index (UAI):** Tolerance for ambiguity. Greece, Portugal, Japan score high. Singapore, Denmark score low. Implications: high-UAI users want clearer error messages, more confirmation modals, more detailed help docs.\n- **Long-Term Orientation (LTO):** Time horizon. East Asian cultures score high; African and Latin American cultures often score low. Implications: pricing models, loyalty programs, retention messaging.\n- **Indulgence vs. Restraint (IVR):** Attitudes toward gratification. Implications: notification cadence, default settings, \"treat yourself\" framing.\n\n### 2. Edward T. Hall's High-Context vs. Low-Context Cultures\nHall divided cultures by how much meaning sits in *context* versus the *literal words*. High-context cultures (Japan, China, Korea, most of the Middle East, Latin America) communicate through subtext, relationships, and shared assumptions. Low-context cultures (U.S., Germany, Scandinavia, Netherlands) prefer explicit, direct communication. Practical implication: if your onboarding copy reads as \"blunt and friendly\" in the U.S., it may read as \"rude and incomplete\" in Japan.\n\n### 3. Trompenaars' Seven Dimensions\nParticularly useful for B2B research: universalism vs. particularism (rules vs. relationships), specific vs. diffuse (compartmentalized vs. holistic), achievement vs. ascription (earned vs. assigned status). Use these to interpret enterprise procurement and decision-making patterns.\n\n## The cross-cultural research workflow\n\nA defensible cross-cultural study follows this nine-step pipeline:\n\n### 1. Define your \"cultural sample frame\"\nBefore recruiting, decide *which* cultures matter and *why*. Don't pick countries — pick cultural contexts. \"Brazil + Mexico\" is not one Latin American sample; they have meaningfully different scores on every Hofstede dimension. A defensible frame names the dimensions you're testing for.\n\n### 2. Localize your research questions, not just translate them\nA question that works in English (\"How do you feel about this feature?\") can fail in cultures where direct opinion-sharing with strangers is uncomfortable. Indirect framings (\"Imagine you were showing this to a colleague — what would you tell them?\") often produce richer data in high-context cultures.\n\n### 3. Recruit through culturally-appropriate channels\nThe simplest route is Panel recruit: describe the audience per market and Koji finds native-language participants across 130+ countries at a price per completed interview in credits you approve first (launching is on paid plans), with quotas per country and your screener questions run before the interview. If you already have users in each market, import them so each person gets a personalized link and the report is segmented by market. If you prefer to run recruitment yourself, match the channel to the culture, not your home market: LinkedIn dominates B2B recruiting in the U.S. and Europe; in Japan, LinkedIn penetration is under 5%, and Wantedly and BizReach are the equivalents; in China, WeChat groups outperform any Western panel. Share your Koji link with the people you find and the interviews land in the same study.\n\n### 4. Choose moderation carefully\nCultural matching of moderator to participant has a measurable effect on disclosure. Research from Kvale & Brinkmann (2008) suggests participants share more authentically when the moderator shares their first language and basic cultural assumptions. This was historically expensive — AI-moderated interviews now neutralize the cost by letting the same AI moderator adapt language and tone per participant.\n\n### 5. Use a mix of structured and open-ended questions\nOpen-ended questions are essential for cultural depth — but they're also harder to compare across cultures. Pair them with [structured questions](/docs/structured-questions-guide) (scale, single_choice, multiple_choice, ranking, yes_no) that you can statistically compare across cohorts. Koji supports all six structured types alongside AI-probed open-ended questions, making cross-cultural comparison a native workflow.\n\n### 6. Watch for response style bias\nDifferent cultures have different \"scale use\" patterns. Japanese respondents tend to use middle-scale answers (\"central tendency bias\"); Latin American respondents use the extremes (\"extreme response style\"); Chinese respondents may avoid disagreement entirely (\"acquiescence bias\"). When comparing Likert results across cultures, standardize scores or use forced-choice methods like MaxDiff or [conjoint analysis](/docs/conjoint-analysis-guide).\n\n### 7. Triangulate with behavioral data\nSelf-reported research is least reliable cross-culturally. Pair interview and survey findings with actual product analytics, where culture-bound politeness norms don't distort the signal.\n\n### 8. Synthesize across, then within\nLook for patterns *across* cultures first — universal needs, shared frustrations. Then go *within* each culture for nuances. This avoids the mistake of treating one culture as the baseline and others as \"deviations.\"\n\n### 9. Pressure-test recommendations with native reviewers\nBefore shipping any cross-cultural research finding to product, have a culturally-native reviewer (ideally one who participated as a moderator or note-taker) sanity-check the interpretation. The single biggest failure mode of cross-cultural research is researchers from culture A interpreting data from culture B through culture A's lens.\n\n## Common cross-cultural research mistakes (and how to avoid them)\n\n| Mistake | Why it happens | Fix |\n|---------|----------------|-----|\n| Translating questions literally | Time pressure | Back-translate every question and run a 2-person cognitive interview |\n| Recruiting only English-speaking expats | They're easy to find | Insist on native-language participants who haven't lived abroad >5 years |\n| One moderator across all markets | Cost savings | Use AI moderators that adapt per language; or use local moderators |\n| Comparing raw Likert scores | Statistical naivety | Use within-respondent standardization (z-scores) or MaxDiff |\n| Treating \"Asia\" as one market | Convenience | Sample each country separately; Hong Kong ≠ Singapore ≠ Tokyo |\n| Skipping ethical review for sensitive topics | Move-fast culture | Localize consent forms; check local data protection laws (GDPR, PIPEDA, LGPD, PIPL) |\n\n## The modern AI-native approach with Koji\n\nTraditional cross-cultural studies were brutally expensive. A 5-country, 30-participant qualitative study with simultaneous interpretation and local moderators routinely ran $40,000–$80,000 and 6–10 weeks of calendar time. The mathematics of global research kept it locked inside enterprise budgets.\n\nAI-native research platforms have rewritten this economics. Here's how Koji specifically changes the workflow:\n\n- **Multilingual AI moderation.** Koji's AI moderator conducts interviews in 30+ languages, adapting tone and probing depth per cultural context. You design the interview once; the AI conducts it in each participant's native language.\n- **Built-in global recruitment.** Open Panel recruit, describe the audience per market, and Koji finds them across 130+ countries at a price you approve first. You can also distribute a single interview link across LinkedIn (US/EU), Wantedly (Japan) or WeChat (China), or, if you prefer to run recruitment yourself, marketplaces like Prolific, and Koji collects standardized results no matter the source.\n- **Automatic thematic analysis across languages.** Koji's thematic analysis layer identifies themes in source languages, then surfaces cross-cultural patterns to you in your reporting language. You don't need bilingual analysts to spot that \"trust\" means something different in Tokyo than in Berlin.\n- **Cultural calibration via [custom AI consultants](/docs/custom-ai-interviewer-persona).** Configure the AI moderator's persona, tone, and even acceptable probing style for each market.\n- **Built-in [structured questions](/docs/structured-questions-guide).** Use scale, ranking, MaxDiff-style choice, and yes/no questions to get statistically comparable data alongside open-ended depth.\n- **60% faster time-to-insight.** Teams using AI-assisted research tools report dramatically faster turnaround — making continuous cross-cultural research a practical reality, not a once-per-launch event.\n\nWhile traditional platforms like UserTesting and Qualtrics require you to assemble localized moderator pools and pay per session, AI-native platforms like Koji let a single PM run a 50-participant, 5-country study in the time it would take to schedule one traditional interview.\n\n## How to write cross-cultural research questions that work\n\nFive rules that consistently improve cross-cultural question quality:\n\n1. **Use behavioral, not hypothetical, prompts.** \"Tell me about the last time you…\" works in every culture. \"Would you ever consider…\" fails in cultures where speculation feels presumptuous.\n2. **Avoid idioms.** \"Move the needle,\" \"low-hanging fruit,\" \"elephant in the room\" — none translate. Use plain verbs.\n3. **Offer escape valves.** End sensitive questions with \"Or feel free to skip this one.\" This is critical in cultures where refusing a researcher's question feels rude.\n4. **Calibrate intensity words.** \"Excellent\" maps to different points on a scale across cultures. Where possible, use behavioral anchors (\"I would recommend this to my closest colleagues\") instead of intensity words.\n5. **Test your discussion guide with at least one native speaker** before deploying it. This is the single highest-ROI quality check in cross-cultural research.\n\n## Cross-cultural research and the future of product development\n\nThe teams that will lead the next decade of consumer and B2B software are the ones that figure out how to make cross-cultural research a *continuous practice* rather than a pre-launch checkpoint. AI is the lever. Platforms like Koji make it possible to run a 15-participant pulse across three markets every two weeks — a cadence that was inconceivable when each round required two months and a small fortune.\n\nThe mindset shift is from \"translate the product after we build it\" to \"design the research process so cultural insight informs every release.\" Companies that adopt the latter aren't spending more on global research — they're spending differently. Less on agencies, more on infrastructure.\n\n## Related Resources\n\n- [Structured questions guide](/docs/structured-questions-guide) — All 6 question types for statistically comparable cross-cultural data\n- [Ethnographic research](/docs/ethnographic-research) — Field methods that complement remote cross-cultural studies\n- [AI-moderated interviews](/docs/ai-moderated-interviews) — How multilingual AI moderators work\n- [Custom AI interviewer persona](/docs/custom-ai-interviewer-persona) — Calibrating tone and probing per culture\n- [Research bias guide](/docs/research-bias-guide) — Bias types beyond cultural bias\n- [Conjoint analysis guide](/docs/conjoint-analysis-guide) — Forced-choice methods that neutralize scale-use bias\n\n---\n\n*Sources: [CSA Research consumer language study (8,709 respondents, 29 countries)](https://csa-research.com/Blogs-Events/CSA-in-the-Media/Press-Releases/Consumers-Prefer-their-Own-Language); Hofstede Insights cultural dimensions framework; Liao & Thomas, \"The Emergence of Collective Cultural Intelligence in Teams in Multicultural Contexts,\" Group & Organization Management (2025); Kvale & Brinkmann, \"InterViews: Learning the Craft of Qualitative Research Interviewing\" (2008).*","category":"Research Methods","lastModified":"2026-10-29T13:18:36.726258+00:00","metaTitle":"Cross-Cultural User Research: Complete Guide (2026)","metaDescription":"Run user research across countries and cultures without losing nuance. Frameworks, recruiting, AI moderation, and bias controls for global product teams.","keywords":["cross-cultural user research","global user research","international ux research","multilingual user interviews","cultural dimensions research","localization research","global product research"],"aiSummary":"Cross-cultural user research studies users across countries and cultures so products feel native, not translated. CSA Research found 76% of consumers prefer buying in their native language and 40% never buy from foreign-language sites. This guide covers Hofstede, Hall, and Trompenaars frameworks, a 9-step workflow, common mistakes, and how AI-moderated platforms like Koji collapse the cost and timeline of multilingual studies.","aiPrerequisites":["Basic familiarity with qualitative research","Understanding of research planning"],"aiLearningOutcomes":["Apply cultural dimensions frameworks (Hofstede, Hall, Trompenaars) to research design","Build a defensible cross-cultural sample frame","Avoid the five most common cross-cultural research mistakes","Use AI-moderated tools to run multilingual studies at scale","Localize research questions beyond literal translation"],"aiDifficulty":"intermediate","aiEstimatedTime":"18 min read"},{"type":"documentation","id":"736e9a7d-c1bb-482d-a0f1-094f74d6280d","slug":"conversation-intelligence-customer-research","title":"Conversation Intelligence for Customer Research: A Practical Guide","url":"https://www.koji.so/docs/conversation-intelligence-customer-research","summary":"Conversation intelligence is software that records, transcribes, and analyses spoken or written conversations to surface patterns automatically. Originally a sales-call category (Gong, Chorus), it has crossed over into customer research where the AI also has to run the conversation. A complete research-grade CI stack has four layers: live transcription, structured extraction (Koji supports 6 question types: open_ended, scale, single_choice, multiple_choice, ranking, yes_no), theme aggregation, and insight surfacing. Research-grade CI differs from sales CI in five ways: it must be the moderator, uses a research question taxonomy, requires structured extraction, must be methodology-aware (Mom Test, JTBD), and has a different consent/candour profile. Koji combines all four CI layers in a single workflow, with real-time theme aggregation after as few as 5 interviews.","content":"## The Bottom Line\n\n**Conversation intelligence** is software that records, transcribes, and analyses spoken or written conversations to surface patterns — sentiment, topics, talk-time, keywords, and emerging themes — automatically. The category started in sales (Gong, Chorus, ZoomInfo Engage) where the goal is helping reps close deals. In 2026, it has crossed over into customer research, where the goal flips: instead of coaching reps, it's extracting insight from the people who matter most — your customers.\n\nFor customer research, conversation intelligence has to do something the sales-call variants don't: it has to also *run* the conversation. A sales-call CI tool listens to a human rep talking to a buyer; a research-grade conversation intelligence platform like Koji listens to **an AI interviewer talking to a customer** — and that AI is the same system that decides what to ask next, scores response quality, and synthesises themes across hundreds of conversations.\n\nThis guide explains what conversation intelligence looks like when you apply it to qualitative research, the four layers of analysis that matter (transcription, structured extraction, theme aggregation, insight surfacing), and how Koji combines all four into a single workflow. It's aimed at product managers, UX researchers, and market researchers who already know what conversation intelligence does for sales and want to bring the same magic to discovery research.\n\n## What \"Conversation Intelligence\" Actually Means in Research\n\nIn sales, CI is mostly post-call analytics: someone records a Zoom, the tool transcribes it, then surfaces talk-time ratios, filler words, competitor mentions. The conversation is over before the intelligence kicks in.\n\nIn research, conversation intelligence has to be live as well as post-hoc. Four layers compose a complete stack:\n\n1. **Live transcription.** Real-time speech-to-text during voice interviews; instant capture in text chats. This is table stakes; what matters is accuracy across accents and domain vocabulary.\n2. **Structured extraction.** Pulling chartable values out of conversational answers — a 1–5 satisfaction score from a casual answer, a ranking from a verbal preference, a yes/no from a hedged statement. This is what separates analytics-grade conversation intelligence from a transcript dump.\n3. **Theme aggregation.** Clustering common patterns across many interviews. Not just keyword counts — semantic grouping of \"the import broke\" and \"couldn't get my data in\" into one theme with quote evidence.\n4. **Insight surfacing.** Going from \"here are the themes\" to \"here's what to do\" — frequency-ranked themes, surprise findings (statements that contradict your prior beliefs), and structured-question distributions broken down by segment.\n\nA serious customer-research conversation intelligence platform delivers all four, integrated. Tools that only do layer 1 or 2 are conversation **capture**, not conversation intelligence.\n\n## How Conversation Intelligence Changes Customer Research\n\nThree concrete shifts when you bring CI into research:\n\n### 1. Synthesis Lag Collapses\n\nTraditional UX research: finish 30 interviews on Monday, schedule a synthesis workshop Friday, publish the report next Tuesday. With conversation intelligence applied research-grade, the synthesis is *already done* by the time the last interview ends. In Koji, themes start emerging after 5 completed interviews and the insight report regenerates on every new completion.\n\n### 2. Coverage Replaces Sampling\n\nWhen the marginal cost of one more interview drops near-zero (AI moderator, no scheduling, no manual coding), the right question shifts from *\"who are our 8 best participants?\"* to *\"who in our entire user base would benefit from telling us?\"* Conversation intelligence + AI moderation make it economically possible to interview hundreds where you previously interviewed dozens.\n\n### 3. Research Becomes Continuous\n\nThe sales analogy is useful here. No sales team uses Gong only for end-of-quarter reviews — they use it continuously, on every call. Conversation intelligence in research enables the same shift: continuous discovery, where every cancellation, every onboarding, every feature launch triggers a fresh round of customer conversations that auto-synthesise into the existing knowledge base. This is what tools like Koji enable through 24/7 always-on interview links.\n\n## The Conversation Intelligence Stack, Layer by Layer\n\n### Layer 1: Live Transcription\n\n**What it does.** Captures every word of voice interviews as searchable, timestamped text. For text interviews, captures the chat verbatim with turn-by-turn structure.\n\n**What to look for.** Multi-language support (Koji handles 30+ languages natively, picked by the respondent on the welcome screen), domain-vocabulary accuracy (technical terms, brand names, product features), speaker diarisation in mixed-mode calls.\n\n**Common failure modes.** Tools that strip out filler words can lose meaningful pauses. Tools that don't recognise your product's feature names produce useless transcripts. Tools that translate too aggressively flatten the participant's actual voice.\n\n### Layer 2: Structured Extraction\n\n**What it does.** Converts conversational answers into structured data — numbers, categories, rankings, yes/no — that can be charted, segmented, and exported.\n\n**Koji's implementation.** Six native structured question types: `open_ended`, `scale`, `single_choice`, `multiple_choice`, `ranking`, and `yes_no`. The AI interviewer asks each type conversationally (no \"click a radio button\" prompts) but the analyser extracts a `structuredValue` for every answer along with the participant's verbatim quote. So your dashboards have *both* the chart and the quotes that back it up.\n\n**Why this matters.** Sales-call CI tools don't do this — they were never designed for structured measurement. Research-grade CI has to, because the deliverable isn't a coaching note for a sales rep, it's a chart that drives a product decision.\n\n### Layer 3: Theme Aggregation\n\n**What it does.** Clusters patterns across many interviews into named themes, ranks them by frequency and emotional intensity, and attaches verbatim quotes as evidence.\n\n**Where AI changes the game.** Traditional thematic analysis is a 4–8 hour manual coding workshop per study. LLM-based theme extraction handles the first 80% in seconds, leaving researchers to override edge cases and add interpretation. In benchmark studies, LLM-based thematic extraction matches mid-level UX researcher accuracy.\n\n**What to look for.** Themes that are semantically grouped (not just keyword matched). Auditability — every theme should let you drill down to the underlying quotes. Real-time updates as new interviews complete. Koji provides all three.\n\n### Layer 4: Insight Surfacing\n\n**What it does.** Goes beyond \"here are the themes\" to \"here's what changed, what surprised us, and what to act on.\"\n\n**Key surfaces:** Theme frequency leaderboards, structured-question distribution charts (NPS, scale ratings), persona-segmented breakdowns, anomaly flags (findings that contradict your stated hypothesis), and shareable report URLs for stakeholders.\n\n**Koji specifics.** Insight reports regenerate on demand (5 credits per refresh) or automatically as enough new data accumulates. Reports are publishable as shareable links — no logins required — so non-Koji stakeholders can read the synthesis without you exporting slides.\n\n## Where Sales-Call CI and Research-Grade CI Diverge\n\nIt's worth being explicit about why dropping Gong or Chorus into your research workflow doesn't work. Five reasons:\n\n1. **No moderator.** Sales CI assumes a human is on the call running it. Research CI has to *be* the moderator (Koji's AI interviewer agent) or analyse an arbitrary recorded session (Dovetail, Marvin).\n2. **Different question taxonomy.** Sales CI tracks \"did the rep ask about budget? Did they mention competitor X?\" Research CI tracks \"what did the participant say about workflow Y? Did they rank A above B?\" Different ontologies entirely.\n3. **Structured extraction is non-optional.** Research needs numbers, not just quotes. Sales CI doesn't generally extract structured measurement.\n4. **Methodology awareness.** Research CI has to know what methodology you're running (Mom Test, JTBD, Concept Testing) and apply the methodology's rules to follow-ups and synthesis. Sales CI has no equivalent concept.\n5. **Participant trust.** Sales CI runs on calls where the buyer is selling-pitch-aware. Research CI runs on participants who agreed to research with no commercial pressure — different consent model, different candour profile.\n\nIf you're evaluating CI tools for research, evaluate them against this list. Most sales-derived tools score low.\n\n## A Practical Workflow: Conversation Intelligence in Action\n\nLet's walk through a real cancel-flow research scenario using a CI-enabled platform like Koji.\n\n**Setup (15 minutes).** You describe the goal to Koji's AI consultant: *\"Understand why trial users don't convert to paid.\"* The consultant drafts a brief with target participants (trial users, days 7–14, didn't convert), methodology (Jobs-to-be-Done switch interviews), and 8 questions including 2 structured ranking questions. You add a tone instruction: *\"Casual, peer voice — we're a developer tool.\"*\n\n**Trigger (continuous).** Your in-app cancel flow now routes anyone who cancels to the Koji interview link. They get a 6–10 minute conversation immediately, while the context is fresh.\n\n**Capture (per interview).** Each conversation auto-transcribes, gets a 1–5 quality score, and produces structured answers for every question. Low-quality interviews don't consume credits.\n\n**Aggregation (real time).** After 5 interviews, themes start emerging on the dashboard. After 15, the report shows: *\"Top cancellation reason: 'Import broke on CSV with non-ASCII characters' (60% of cancellers). Secondary: 'Couldn't figure out where to find feature X' (33%). NPS distribution: heavily bimodal — 8s/9s love it, 2s/3s churned.\"*\n\n**Action (after 30 interviews).** You ship a fix to the CSV importer. The conversation intelligence platform keeps running, so you'll see whether cancel reasons shift in the next batch.\n\nThat full loop — trigger → capture → aggregation → action — used to take six weeks. With research-grade conversation intelligence, it's 1–2 weeks and the loop runs continuously.\n\n## What to Look For When Buying\n\nIf you're evaluating conversation intelligence for customer research, here's the short checklist:\n\n- **Live transcription accuracy in your domain.** Test with your real product terminology and accents.\n- **Native structured question types.** Especially scale, ranking, multiple_choice — these are how charts get built.\n- **Adaptive follow-up probing.** Watch a live demo. Does the AI actually ask follow-ups based on what the participant said?\n- **Automatic quality scoring.** Look for 1–5 scoring at the transcript level (Koji uses this) — and ideally a way to exclude low-quality transcripts from final reports without consuming credits.\n- **Real-time theme aggregation.** Confirm themes update as interviews complete, not in nightly batch jobs.\n- **Auditability.** Every theme should drill into source quotes; every chart should drill into source answers.\n- **Agent composability.** In 2026, MCP support matters — can Claude or Cursor query the data as a tool? Koji exposes 15+ MCP tools for this exact use case.\n- **Pricing transparency.** Credit-based pricing on Koji (as low as €1 per qualified interview, no subscription needed) beats opaque enterprise contracts.\n\n## Common Mistakes\n\n**Treating CI as a transcription tool.** Transcription is layer 1 of 4. If your tool stops there, you'll still be coding transcripts manually — which means the time savings evaporate.\n\n**Buying sales CI for research.** Gong is brilliant at what it does. It's the wrong shape for customer research. Use research-native tools.\n\n**Ignoring structured extraction.** \"We'll just read the transcripts\" is fine for 5 interviews. For 50, you need structured data feeding charts. Without it, scaling stalls.\n\n**No methodology awareness.** A CI tool that treats every conversation generically misses the point. Methodology-aware extraction (Mom Test rules, JTBD switch interviews) produces dramatically better insights.\n\n**Skipping the proof-of-value step.** Don't buy on a sales pitch. Run a 5-interview pilot on your real participants before signing anything.\n\n## Frequently Asked Questions\n\n**What is conversation intelligence in customer research?** Software that captures conversations between an AI moderator (or human moderator) and customers, transcribes them, extracts structured measurement and themes, and surfaces insights — automatically and at scale. It's the research-grade descendant of sales-call CI tools like Gong, adapted for qualitative research workflows where the AI also runs the conversation.\n\n**How is research-grade CI different from sales CI like Gong?** Five differences: it has to *be* the moderator (or analyse moderator-free recordings), it tracks a research-question taxonomy (not a sales-question one), structured extraction is non-optional, it has to be methodology-aware (Mom Test, JTBD), and consent/candour profiles are different. Sales CI dropped into research workflows scores poorly.\n\n**Can I use conversation intelligence for both qualitative and quantitative research?** Yes — and that's a major reason to use a research-grade CI platform. Koji's six structured question types (open_ended, scale, single_choice, multiple_choice, ranking, yes_no) let one conversational interview produce both rich qualitative quotes and chartable quantitative data. No need to run separate surveys and interviews.\n\n**How accurate is automatic theme extraction?** In benchmark studies, LLM-based thematic extraction matches mid-level UX researcher accuracy. The key is auditability — every theme should drill into its source quotes, so you can override the AI when nuance matters.\n\n**Does conversation intelligence work for B2B research?** Especially well. B2B research lives or dies on getting in front of busy executives — and CI's big unlock is that an AI moderator can run an interview at 11pm with a CFO in another timezone. Personalised interview links, account context, and structured outputs that feed straight into client decks are why CI platforms have become popular with consultancies.\n\n**What about privacy and consent?** Look for GDPR compliance, automatic PII redaction options, clear data residency policies, and per-study consent capture. Koji captures intake-form consent before the AI starts asking research questions, and offers anonymisation options before sharing transcripts.\n\n## Related Resources\n\n- [AI Customer Insights Platform: 2026 Buyer's Guide](/docs/ai-customer-insights-platform) — broader category guide\n- [AI Transcript Analysis Guide](/docs/ai-transcript-analysis-guide) — deeper look at automatic transcript analysis\n- [Understanding Themes & Patterns](/docs/understanding-themes-patterns) — how Koji aggregates themes\n- [AI Generated Insights](/docs/ai-generated-insights) — what AI-generated insights look like in practice\n- [Real-Time Research Insights](/docs/real-time-research-insights) — continuous insight surfacing\n- [Structured Questions Guide](/docs/structured-questions-guide) — the 6 structured types that turn conversations into charts","category":"Reports & Analysis","lastModified":"2026-10-29T13:18:36.44188+00:00","metaTitle":"Conversation Intelligence for Customer Research: 2026 Guide | Koji","metaDescription":"Conversation intelligence moves from sales calls into customer research. The four layers (transcription, structured extraction, theme aggregation, insight surfacing) and how Koji applies them.","keywords":["conversation intelligence","conversation intelligence customer research","conversation intelligence for research","conversation analytics research","customer conversation intelligence","conversation intelligence platform 2026","gong for user research"],"aiSummary":"Conversation intelligence is software that records, transcribes, and analyses spoken or written conversations to surface patterns automatically. Originally a sales-call category (Gong, Chorus), it has crossed over into customer research where the AI also has to run the conversation. A complete research-grade CI stack has four layers: live transcription, structured extraction (Koji supports 6 question types: open_ended, scale, single_choice, multiple_choice, ranking, yes_no), theme aggregation, and insight surfacing. Research-grade CI differs from sales CI in five ways: it must be the moderator, uses a research question taxonomy, requires structured extraction, must be methodology-aware (Mom Test, JTBD), and has a different consent/candour profile. Koji combines all four CI layers in a single workflow, with real-time theme aggregation after as few as 5 interviews.","aiPrerequisites":["Familiarity with qualitative research workflows","Basic understanding of what conversation intelligence does for sales"],"aiLearningOutcomes":["Understand the 4-layer conversation intelligence stack for research","Distinguish research-grade CI from sales-call CI tools","Apply CI to continuous customer research workflows","Evaluate CI platforms against the right buying criteria","Avoid the common mistakes when buying CI for research"],"aiDifficulty":"intermediate","aiEstimatedTime":"11 min read"},{"type":"documentation","id":"8280b46e-793f-4333-8419-2342e6b5f551","slug":"continuous-discovery-tools-2026","title":"Continuous Discovery Tools 2026: The AI-Powered Stack for Weekly Customer Interviews","url":"https://www.koji.so/docs/continuous-discovery-tools-2026","summary":"Continuous discovery tools in 2026 fall into four categories: AI-native interview platforms (the linchpin), participant recruiting marketplaces, research repositories, and opportunity solution tree mapping software. Most continuous discovery programs stall not from lack of intent but from operational frictions — scheduling burden, synthesis lag, and the research-to-decision gap. AI-native platforms like Koji remove the scheduling and moderation bottlenecks entirely by running 24/7 always-on AI interviews (voice or text) against a single self-serve link, with real-time synthesis and API/MCP composability. The reference 2026 stack pairs Koji as the interview layer with optional repository (Dovetail/Marvin), recruiting marketplace (Respondent), opportunity tree mapping (Miro/Productboard), and Slack alerts from a job that polls the Koji API.","content":"## The Bottom Line\n\n**Continuous discovery** — Teresa Torres's framing of running at least one customer interview every week — is the gold-standard product practice for 2026. The problem has always been the same: weekly interviews don't sustain themselves. A PM running their tenth round of \"schedule the call, run the call, transcribe, code, share quotes\" usually burns out within two quarters.\n\nThe tools that actually make continuous discovery a habit fall into four categories: (1) AI-native interview platforms that remove the moderation bottleneck, (2) participant recruiting marketplaces, (3) research repositories that organize findings, and (4) opportunity solution tree mapping software. You need pieces from at least categories 1 and 3 to sustain the practice. Adding category 4 makes the connection between insights and product decisions explicit.\n\nThis guide is a 2026 evaluation. We'll show why traditional research stacks fail at continuous discovery, then walk through what to look for in each category — with practical comparisons including how Koji's AI-native model removes the largest bottleneck (the moderator).\n\n## Why Most Continuous Discovery Programs Stall\n\nTeresa Torres's *Continuous Discovery Habits* (and the community around it) makes the practice sound simple: a small product trio talks to customers weekly, captures opportunities in a tree, and tests assumptions through small experiments. In practice, three operational frictions kill the cadence:\n\n1. **The scheduling burden.** Recruiting, scheduling, rescheduling, and following up with even one interview/week consumes hours of PM or researcher time. Multiply by 50 weeks/year and you have an unsustainable load.\n2. **The synthesis lag.** A call ends Tuesday; transcription comes back Wednesday; tagging happens Friday; the PM forgets the most important quote by then. Insights cool down fast.\n3. **The \"research-to-decision\" gap.** Findings live in a Notion doc that nobody reads. Without a clear handoff from \"we heard X\" to \"we're changing the roadmap because of X,\" the team stops valuing the work.\n\nA continuous discovery stack works only if it removes all three. Adding more tools that solve only the third problem (a fancier repository) without addressing the first two is why most adoptions stall in month 3.\n\n## The Four Categories of Continuous Discovery Tools\n\n### Category 1: AI-Native Interview Platforms\n\nThis is the biggest 2026 shift. AI-moderated interview platforms remove the scheduling and moderation bottlenecks entirely — you publish one self-serve link, and the AI runs every interview 24/7. The category leader for continuous discovery is built around three primitives:\n\n- **Always-on conversation.** A standing study lives indefinitely; participants click the link and the AI moderates a full conversation (voice or text), asking adaptive follow-ups.\n- **Real-time synthesis.** Themes and structured answers aggregate as interviews complete. By interview #10 you have a working insights view, not a pile of recordings to process.\n- **Composable outputs.** The REST API, the public report link, CSV and JSON export, and Model Context Protocol integrations get findings into Slack, the repository, the roadmap tool, or another AI agent. Koji has no outbound webhooks yet, so a scheduled job polls the read endpoint.\n\n**Koji** is purpose-built for this pattern. The AI consultant drafts the brief from your research question (with methodology frameworks like Mom Test, Jobs-to-be-Done, and Customer Discovery embedded as runtime principles). The AI interviewer runs voice and text conversations in 30+ languages the respondent picks on the welcome screen, with 6 structured question types (open_ended, scale, single_choice, multiple_choice, ranking, yes_no) aggregated into real-time charts. Quality scoring (1–5 per interview) gates which transcripts count — interviews scoring 1 or 2 don't consume credits.\n\nPricing makes continuous discovery economically obvious: interviews start as low as €1 per qualified interview and €3 per qualified voice interview, and one credit is the only unit (text interview = 1 credit, voice = 3, report refresh = 5). Start with pay as you go. No subscription needed. You pay only for the interviews your study actually uses, and volume pricing and plans are there when you want them. New accounts get 10 free credits, no card. Compared with $300+/hour for a researcher to moderate one interview, that is the difference between a research budget and a rounding error. There is no overage. You pay as you go, and you only ever pay for the conversations that bring insights.\n\n**When you need this category:** if your team is committed to weekly interviews but the scheduling and moderation load is the reason it isn't happening. This is the bottleneck for 90% of continuous discovery programs.\n\n### Category 2: Participant Recruiting\n\nUnmoderated AI interviews still need participants. Three sub-categories:\n\n- **In-product recruiting.** Email or intercom users who match criteria (e.g., \"trialing for 7 days,\" \"downgraded last month\"). Koji's embed widget, CRM import, and personalized interview links make this turnkey — you can target named accounts with the AI agent already aware of the company.\n- **Panel recruitment.** For hard-to-reach audiences, panel recruitment is built into Koji: describe the audience, get a per-respondent quote in credits, and approve it before launch. External marketplaces (User Interviews, Respondent, Prolific) remain an option if you want to run recruitment yourself. Less needed for everyday continuous discovery if you have an active user base.\n- **Public/community recruiting.** Share interview links in newsletters, Discord, LinkedIn. Works well for early-stage products and for thought-leadership-driven companies.\n\nFor most product teams, in-product recruiting is sufficient for continuous discovery — your own users are the highest-signal participants. Koji simplifies this with intake forms, lead-collection fields, screener questions, and CSV import.\n\n### Category 3: Research Repositories\n\nA repository is where you keep the institutional knowledge. Once interviews are running automatically, the repository question becomes: *where do quotes, themes, and structured answers live so the team can find them six months from now?*\n\nOptions:\n\n- **Koji itself** stores all transcripts, themes, quality scores, structured answers, and reports, with every claim in a report linked to the quote behind it and an MCP connector your assistant can read them through. For many teams, this is enough — no separate repository needed.\n- **Dovetail, Marvin (HeyMarvin), EnjoyHQ** are dedicated repositories with rich tagging, search, and AI-assisted synthesis. Worth adding if you have research from multiple sources (interviews, support tickets, sales calls) you want centralized. Koji feeds these through its REST API and CSV or JSON export.\n- **Notion / Coda databases** are the lightweight option. Many teams pipe Koji's read endpoint into a Notion database on a schedule and tag by theme manually. Works fine for small teams.\n\n**Heuristic:** if Koji is your only data source, you don't need a separate repository for 6+ months. When you accumulate hundreds of interviews and want cross-study synthesis, that's when a dedicated repository starts paying off.\n\n### Category 4: Opportunity Solution Tree Mapping\n\nThis is the part Teresa Torres's methodology emphasizes most: visualizing opportunities, branches, and assumption tests in a single tree. Tools that map this:\n\n- **Mural, Miro, FigJam** — whiteboard-style trees. Most teams start here.\n- **Lucidchart, Whimsical** — dedicated diagramming.\n- **Notion + Linear** — linked databases (opportunity → solution → experiment).\n- **Productboard** — formal opportunity tracking with customer feedback inputs.\n\nKoji doesn't replace this category — but it makes feeding the tree dramatically faster. The MCP integration lets Claude or another LLM read Koji studies and draft an opportunity solution tree from raw interviews. A scheduled job reading the API can also feed Productboard or Linear with new opportunity entries.\n\n## A Reference Stack for 2026\n\nBased on what high-performing continuous discovery teams use:\n\n| Layer | Tool | Role |\n|---|---|---|\n| Interview platform | **Koji** | AI-moderated voice/text interviews, brief generation, real-time synthesis, report distribution |\n| Recruiting | Koji (+ marketplaces for niches) | In-product recruiting and built-in panel recruitment via Koji, quoted in credits; external marketplaces if you want to run recruitment yourself |\n| Repository (optional, year 2+) | Dovetail or Marvin | Cross-study synthesis once you have 200+ interviews |\n| Opportunity mapping | Miro / Notion / Productboard | Visual tree maintained weekly during the team trio's discovery review |\n| Routing & alerts | Slack, from a job polling the Koji API | Near real-time theme notifications |\n| AI workflow | Claude + Koji MCP | \"Read the latest 10 interviews and update the opportunity tree\" |\n\nThe critical insight: pick the interview platform first. Everything else is secondary. If interviews aren't happening weekly, no repository or tree tool matters.\n\n## Why the AI-Native Interview Platform Is the Linchpin\n\nLet's compare what a week of continuous discovery looks like with and without AI moderation.\n\n**Without AI moderation (traditional stack):**\n\n- Monday: PM emails 5 customers asking for time. 2 respond.\n- Tuesday-Wednesday: Calendar tetris. Two interviews booked for Friday.\n- Friday: Two 30-minute calls. PM takes rough notes.\n- Saturday-Sunday: PM debates whether to transcribe. Usually doesn't.\n- Following Wednesday: Maybe a Slack post with two quotes. Maybe not.\n\nNet output: 2 interviews, partially synthesized, ~3 hours of PM time, often dropped within a month.\n\n**With AI moderation (Koji-led stack):**\n\n- Monday: PM checks the always-on study. 6 new interviews completed over the weekend. Real-time report already shows two emerging themes.\n- Tuesday: PM reads the 6 transcripts (30 min total). Pulls 3 quotes into the opportunity tree.\n- Friday: 4 more interviews completed during the week. Theme #2 now has 8 supporting quotes — stable enough to act on.\n- Following Monday: PM presents the theme to engineering with verbatim quotes; team commits to test an assumption next sprint.\n\nNet output: 10 interviews, fully synthesized, ~1 hour of PM time, decisions made. The discovery practice sustains because the time cost dropped 75% and the data freshness improved 10×.\n\n## Evaluation Criteria\n\nWhen comparing continuous discovery tools, score each candidate on these dimensions:\n\n1. **Weekly cost of one extra interview.** Traditional vendor: $300–$1,000/interview. Koji: as low as €1 per qualified interview and €3 per qualified voice interview. Time cost matters more than dollar cost.\n2. **Time-to-insight.** From the moment an interview ends, how long until a usable theme exists? Koji: minutes (real-time synthesis). Traditional + repository: days.\n3. **Methodology rigor.** Does the platform enforce known frameworks (Mom Test, JTBD)? Koji embeds these as runtime principles. Most platforms leave it to the moderator.\n4. **Composability.** Can other tools (Slack, Linear, Notion, Claude) consume the data? Koji exposes everything via REST, MCP, CSV, and JSON.\n5. **Always-on capability.** Can the study run 24/7 without scheduling? AI-native platforms: yes. Everything else: no.\n6. **Quality control on AI interviews.** Look for per-interview quality scoring, source citations on every quote, and credit refunds on low-quality conversations. Koji does all three.\n\nWeight criteria 2, 4, and 5 most heavily — they're what determines whether the practice survives past month 3.\n\n## Frequently Asked Questions\n\n**Is \"continuous discovery\" the same as \"continuous research\"?** Mostly. Teresa Torres uses *continuous discovery* to mean a weekly cadence of customer touchpoints by the product trio (PM, designer, engineer). *Continuous research* is the broader UX research equivalent. The tools are the same.\n\n**Do we need a researcher to do continuous discovery?** No, and that's the whole point. Continuous discovery is owned by the product trio. AI-native interview platforms like Koji make it possible for PMs to run discovery without research training, because methodology principles are embedded in the AI's prompt — not assumed to live in the PM's head.\n\n**What about recording-based tools like UserTesting?** They're built for usability testing, not interviews. The session is recorded but unmoderated by default; if you want adaptive follow-up, you're back to scheduling a human moderator. They don't fit continuous discovery cadence well.\n\n**Can we just use ChatGPT?** ChatGPT can help with brief drafting and analysis, but it doesn't run interviews end-to-end, doesn't score quality, doesn't aggregate across participants, and doesn't expose findings as machine-readable feeds. A continuous discovery stack needs more than a chatbot.\n\n**How does this fit with usage analytics (Amplitude, Mixpanel)?** Beautifully — usage analytics shows *what* happens; continuous discovery interviews show *why*. Your analytics webhook can call Koji's API to start an interview when an event fires (e.g., user hits the upgrade modal three times without upgrading), getting the \"why\" alongside the \"what.\"\n\n**Where do I start if my team has never done continuous discovery?** Start with one weekly slot, an always-on interview link in your in-product onboarding, and a 30-minute Friday review of new transcripts. Pick a single product question per quarter (e.g., \"what blocks activation?\") and run it as a standing study. By week 4 you'll have 20+ interviews; by week 12 the practice has become institutional.\n\n## Related Resources\n\n- [Continuous Discovery User Research](/docs/continuous-discovery-user-research) — running weekly interviews as a sustainable team practice\n- [Customer Discovery Interviews at Scale](/docs/customer-discovery-interviews-at-scale) — talking to 100 customers in a week\n- [Always-On User Interviews](/docs/always-on-user-interviews-24-7-ai-moderator) — 24/7 AI moderator setup\n- [Koji for Product Managers](/docs/koji-for-product-managers) — PM-specific workflow guide\n- [Research-Driven Roadmap Prioritization](/docs/research-driven-roadmap-prioritization) — using interviews to build better roadmaps\n- [Best User Research Tools in 2026](/docs/best-user-research-tools-2026) — the complete category comparison\n- [Structured Questions Guide](/docs/structured-questions-guide) — the 6 question types Koji aggregates automatically\n- [Koji vs. Dovetail](/docs/koji-vs-dovetail) — interview platform vs analysis-only repository\n\n## Further reading on the blog\n\n- [Agile User Research: How to Run Continuous Research in Sprint Cycles (2026)](/blog/agile-user-research-2026) — Most teams know they should do user research every sprint. Almost none actually do. Here's the practical playbook for integrating continuous\n- [Best Product Discovery Tools in 2026: The Complete Buyer's Guide](/blog/best-product-discovery-tools-2026) — Product discovery is no longer a one-off pre-launch phase — it is a continuous loop. The best teams in 2026 are running weekly customer inte\n- [Koji vs Lookback: AI-Native Research vs Live Moderated Sessions (2026)](/blog/koji-vs-lookback-2026) — Koji and Lookback take opposite approaches to user research. One automates the entire interview, the other perfects the live observation exp\n\n<!-- further-reading:blog -->\n","category":"Comparisons","lastModified":"2026-10-29T13:18:36.13979+00:00","metaTitle":"Continuous Discovery Tools 2026: AI-Powered Stack for Weekly Interviews","metaDescription":"A 2026 buyer's guide to continuous discovery tools. Compare AI-native interview platforms, repositories, recruiting marketplaces, and opportunity tree mapping for product teams.","keywords":["continuous discovery tools","continuous discovery software","teresa torres tools","weekly customer interviews tool","opportunity solution tree software","continuous discovery habits stack","continuous discovery platform","product discovery tools 2026"],"aiSummary":"Continuous discovery tools in 2026 fall into four categories: AI-native interview platforms (the linchpin), participant recruiting marketplaces, research repositories, and opportunity solution tree mapping software. Most continuous discovery programs stall not from lack of intent but from operational frictions — scheduling burden, synthesis lag, and the research-to-decision gap. AI-native platforms like Koji remove the scheduling and moderation bottlenecks entirely by running 24/7 always-on AI interviews (voice or text) against a single self-serve link, with real-time synthesis and API/MCP composability. The reference 2026 stack pairs Koji as the interview layer with optional repository (Dovetail/Marvin), recruiting marketplace (Respondent), opportunity tree mapping (Miro/Productboard), and Slack alerts from a job that polls the Koji API.","aiPrerequisites":["Basic familiarity with product discovery practices","A Koji account (free tier works)"],"aiLearningOutcomes":["Identify the three operational frictions that kill continuous discovery programs","Map continuous discovery tools across four categories: interview, recruiting, repository, opportunity mapping","Compare AI-native interview platforms against traditional research stacks","Build a reference 2026 stack for weekly customer interviews","Apply six evaluation criteria when comparing continuous discovery tools"],"aiDifficulty":"intermediate","aiEstimatedTime":"12 min read"},{"type":"documentation","id":"02ddc7e9-d227-4a54-bc98-5381ec8afd13","slug":"chatgpt-for-user-research","title":"How to Use ChatGPT for User Research (And When You Need a Specialized Platform)","url":"https://www.koji.so/docs/chatgpt-for-user-research","summary":"ChatGPT excels at research-assistant tasks — drafting questions, summarizing single transcripts, brainstorming themes — but cannot moderate live interviews, recruit participants, aggregate across conversations, enforce quality, or produce defensible reports. For real research programs, pair ChatGPT for writing tasks with a purpose-built AI platform like Koji that handles moderation, recruitment, structured questions, and real-time aggregation end to end.","content":"# How to Use ChatGPT for User Research (And When You Need a Specialized Platform)\n\n**Short answer:** ChatGPT is excellent as a research *assistant* — it can help draft interview questions, summarize a single transcript, or brainstorm themes. It is not a research *platform*. It cannot moderate live interviews with hundreds of participants, run consistent follow-up probing across studies, recruit users, enforce quality, produce defensible aggregated reports, or store everything in a queryable repository. For one-off help, use ChatGPT. For an actual research program, use a purpose-built AI research platform like [Koji](/docs/quick-start-guide).\n\nThis guide shows you exactly which research jobs ChatGPT handles well, which jobs it quietly sabotages, and how to build a workflow that uses both tools where each shines.\n\n## What Counts as \"User Research\" Here\n\nWhen we say user research, we mean the full workflow that turns customer conversations into product, marketing, or design decisions:\n\n1. **Plan** — pick a research question, methodology, and target participants\n2. **Recruit** — find the right people and convince them to talk\n3. **Moderate** — ask the questions, probe deeper when you hear something interesting\n4. **Capture** — record, transcribe, and store the conversation\n5. **Analyze** — code themes, count patterns, surface quotes\n6. **Report** — present findings in a way that drives decisions\n7. **Repeat** — make this a continuous habit, not a one-off project\n\nChatGPT touches steps 1, 5, and 6. A platform like Koji handles all seven, end to end.\n\n## Where ChatGPT Genuinely Helps in Research\n\nLet us be fair to a very capable tool. There are real research jobs ChatGPT does well, especially for solo founders or product managers without a research budget.\n\n### 1. Drafting interview questions\n\nGive ChatGPT a research goal (\"I want to understand why trial users do not convert to paid\") and it will produce a reasonable opening list of 8–12 open-ended questions. Quality varies — questions tend to be generic and lean toward leading phrasing — but as a starting point it beats a blank page.\n\n### 2. Summarizing a single transcript\n\nPaste a 30-minute interview transcript and ChatGPT will return a tidy summary, list pain points, and pull a few quotes. Useful for individual debriefs.\n\n### 3. Brainstorming themes from a small set of quotes\n\nGive it 20–30 quotes you have already extracted and it will cluster them into themes. This is a shortcut for the manual affinity-mapping work you would otherwise do on a whiteboard.\n\n### 4. Reframing findings for different audiences\n\nOne strong use: take the same set of insights and ask ChatGPT to rewrite them for engineering, sales, and the executive team. It is a competent translator.\n\n### 5. Suggesting research methodologies\n\nAsk \"what is the right method to test pricing sensitivity for a B2B SaaS product?\" and ChatGPT will give you a respectable answer (Van Westendorp, Gabor-Granger, or conjoint analysis), often with execution notes.\n\nFor a list of 100+ open-ended question patterns you can mix into your own studies, see [open-ended interview questions](/docs/open-ended-interview-questions).\n\n## Where ChatGPT Falls Short for Real Research\n\nThe limitations show up the moment you scale beyond a single conversation, beyond yourself as the moderator, or beyond a one-time project.\n\n### 1. It cannot moderate live interviews\n\nChatGPT cannot have a real-time conversation with your customer over voice or text and decide on the fly which follow-up question to ask based on what they just said. You — a human — still have to run every interview. That is the bottleneck killing your research velocity.\n\nKoji turns this on its head. The AI moderator runs the interview directly with the participant: voice or text, in 30+ languages the respondent picks on the welcome screen, 24/7, with intelligent follow-up probing based on the [research brief](/docs/understanding-the-research-brief) and methodology you defined. You set the questions; Koji runs them with hundreds of participants in parallel.\n\n### 2. It has no memory across conversations\n\nEach ChatGPT conversation is an island. If you ran interview #1 on Tuesday and interview #15 on Friday, ChatGPT cannot tell you \"5 of the 15 participants mentioned the same onboarding friction at minute 3.\" It has no aggregation layer.\n\nKoji ties every interview into a study-level [insights dashboard](/docs/insights-dashboard) where themes, frequencies, sentiment, and structured-question distributions update automatically as new responses come in. See exactly how this works in [understanding themes and patterns](/docs/understanding-themes-patterns).\n\n### 3. It has no structured-question handling\n\nIf you want to ask \"On a scale of 1–10, how likely are you to recommend us?\" inside a longer conversation and then chart the distribution across 200 responses, ChatGPT cannot do that natively. It treats every answer as free text.\n\nKoji has six first-class question types — open_ended, scale, single_choice, multiple_choice, ranking, and yes_no — that the AI moderator asks conversationally and the analysis layer aggregates into proper charts. Read more in the [structured questions guide](/docs/structured-questions-guide).\n\n### 4. It does not recruit participants\n\nObvious but worth saying. ChatGPT cannot send invitation emails, host an interview link, or manage a participant panel. You have to glue together Calendly, your CRM, an email tool, and a moderation calendar. Most projects die in this glue.\n\nKoji does this part for you. If you do not have an audience, Koji recruits one: describe the market, demographics, role and screening you need, approve a live per-respondent quote in credits, and interviews start arriving. If you do have a list, a product, or a community, Koji interviews them through a hosted, branded interview link, [personalized links](/docs/personalized-interview-links) for individual participants, [CSV import](/docs/importing-participants-csv), or [CRM integration](/docs/crm-research-integration-guide). Every conversation runs 24/7 without you scheduling anything.\n\n### 5. Quality is unpredictable\n\nChatGPT will happily summarize a transcript that is 80% participant rambling and 20% useful signal — and you will not know which is which. There is no quality gate.\n\nKoji scores every conversation on a 0–5 quality scale, only counts conversations scoring 3+ against your credits, and flags low-quality responses for review. See [how the quality gate works](/docs/how-the-quality-gate-works) and [understanding quality scores](/docs/understanding-quality-scores).\n\n### 6. Privacy and compliance are your problem\n\nPasting customer interview transcripts into ChatGPT can violate GDPR, your DPA with customers, and any compliance program you have. Most enterprise legal teams will not allow it. Koji is built for research workflows from the ground up with proper data handling, role-based access, and an audited path from transcript to report.\n\n### 7. No defensible report you can share\n\nChatGPT outputs a chat message. Koji outputs a [structured research report](/docs/generating-research-reports) you can [publish and share](/docs/publishing-sharing-reports) with stakeholders, complete with quotes, themes, methodology, and traceability back to source transcripts.\n\n## The \"ChatGPT + Spreadsheet\" Anti-Pattern\n\nThe most common DIY workflow looks like this:\n\n1. Use ChatGPT to draft questions\n2. Run interviews manually over Zoom or Google Meet\n3. Use Otter or Fireflies to transcribe\n4. Paste each transcript into a fresh ChatGPT thread for synthesis\n5. Copy the bullets into a Google Doc\n6. Manually re-cluster across documents in a spreadsheet or Notion table\n\nThis works for 2–3 interviews. By interview 10, you are spending more time on logistics than learning. By interview 25, you have given up.\n\nThe AI-native workflow with Koji collapses steps 2–6:\n\n1. Tell Koji your research goal — [the AI consultant](/docs/working-with-the-ai-consultant) generates the brief and questions\n2. Share the link or [embed the widget](/docs/using-the-embed-widget) on your site\n3. Hundreds of participants take the AI-moderated interview asynchronously\n4. The dashboard updates in real time with themes, quotes, and structured data\n5. Export a report in one click\n\nResearch teams that switch report a 10x speed-up and the ability to run continuous discovery instead of one-off studies. Read the full migration playbook in [from survey to conversation](/docs/from-survey-to-conversation-guide).\n\n## When ChatGPT is Genuinely the Right Choice\n\nDo not overbuy. ChatGPT alone is fine if:\n\n- You will run **fewer than 3 interviews total** on this project\n- You are doing **secondary research** (analyzing reports, articles, public datasets) not primary research\n- You need to **draft a screener email** or polish a single insight statement\n- You want to **brainstorm methodology** before you commit to one\n- You are a student or learning the craft of qualitative research\n\nFor anything resembling a real research program — recurring studies, more than 5 participants, multiple stakeholders, regulated data — graduate to a platform.\n\n## A Hybrid Workflow That Actually Works\n\nMost teams end up using both tools intentionally. Here is the split that works:\n\n| Job                                | Best Tool        | Why                                                                 |\n|------------------------------------|------------------|---------------------------------------------------------------------|\n| Brainstorm research goal           | ChatGPT          | Fast, conversational, no setup                                      |\n| Generate interview brief           | Koji AI Consultant | Methodology-aware, produces a usable brief, links to your study     |\n| Recruit and moderate participants  | Koji             | Built-in recruitment, hosted link, AI moderator, 24/7                                |\n| Capture and transcribe responses   | Koji             | Built in, voice + text, multilingual                                |\n| Aggregate themes across interviews | Koji             | Real-time dashboard, structured-answer aggregation                  |\n| Pull a single quote for a deck     | ChatGPT or Koji  | Either works                                                        |\n| Reframe findings for a sales deck  | ChatGPT          | Tone shifting is its native skill                                   |\n| Build a defensible research report | Koji             | Structured, shareable, traceable                                    |\n\nPoint Claude, Cursor or any MCP assistant at your studies with the [Koji MCP connector](/docs/mcp-overview) to ask free-form questions about your research data. It is essentially \"ChatGPT for your interviews,\" but grounded in your real conversations rather than guessing.\n\n## How to Migrate from a ChatGPT-Only Workflow\n\nIf you have been running research with ChatGPT and a spreadsheet, here is how to move:\n\n1. **Export your existing transcripts and quotes** into a folder.\n2. **Sign up for Koji** and create your first study using the AI consultant. Tell it your research goal in plain English — it generates the brief, methodology, and questions.\n3. **Pick your interview mode**: structured for tight follow-ups, exploratory for open discovery, or hybrid. See the [interview mode guide](/docs/interview-mode-guide).\n4. **Configure structured questions** for anything you want to chart — NPS, satisfaction, feature preference. The [structured questions guide](/docs/structured-questions-guide) walks through all six types.\n5. **Publish the study** and share the link with your existing customer list. Use [personalized links](/docs/personalized-interview-links) for warm contacts.\n6. **Watch responses come in 24/7** while you focus on other work.\n7. **Generate the report** with one click. Compare it to what your old ChatGPT process would have produced. The difference is usually not subtle.\n\nKoji free tier comes with 10 credits — enough to test the full workflow on a real study. Text interviews cost 1 credit each, voice interviews cost 3 credits, and a report refresh costs 5 credits.\n\n## Bottom Line\n\nChatGPT is a Swiss Army knife for individual research tasks. It is not a research platform. The teams that win at customer research in 2026 use AI to *run* the interviews, not just to summarize them after a human did. Koji is built for exactly that.\n\nIf you are still using ChatGPT for research, you are not behind — but you are about to be. Move the moderation, recruitment, and aggregation to a purpose-built platform and keep ChatGPT for the writing tasks where it shines.\n\n## Related Resources\n\n- [AI-Assisted Answers in Open-Ends](/docs/ai-assisted-open-ended-responses) — what to do when participants let an LLM write their answers for them\n- [Structured Questions in AI Interviews](/docs/structured-questions-guide) — the six question types Koji aggregates automatically\n- [AI-Moderated Interviews: How Automated Research Works](/docs/ai-moderated-interviews) — the deep dive on AI moderation\n- [Best User Research Tools in 2026](/docs/best-user-research-tools-2026) — full comparison of the modern research stack\n- [From Survey to Conversation: The Complete Migration Guide](/docs/from-survey-to-conversation-guide) — switching from forms to AI interviews\n- [MCP Connector: Ask Any Question About Your Research Data](/docs/mcp-overview): Koji-grounded conversational analysis from Claude or any MCP assistant\n- [Customer Discovery Interviews at Scale](/docs/customer-discovery-interviews-at-scale) — running 100 interviews in a week\n\n## Further reading on the blog\n\n- [The Future of User Research in 2026: AI, Automation, and What's Next](/blog/future-of-user-research-2026) — User research is undergoing its biggest structural shift in decades. AI has moved from add-on feature to core infrastructure. Here is what t\n- [Koji vs Looppanel: End-to-End AI Research vs Analysis-Only (2026)](/blog/koji-vs-looppanel-2026) — Looppanel is a great AI analysis tool — but it only handles half the workflow. You still need a separate platform to actually run interviews\n- [Koji vs Qualtrics: AI-Native Research vs Enterprise Survey Platform (2026)](/blog/koji-vs-qualtrics-2026) — Qualtrics is the gold standard for enterprise survey programs. But when your team needs to understand *why* customers behave the way they do\n\n<!-- further-reading:blog -->\n","category":"Comparisons","lastModified":"2026-10-29T13:18:35.861382+00:00","metaTitle":"ChatGPT for User Research: Where It Helps and Where You Need a Real Platform","metaDescription":"ChatGPT is great for drafting interview questions and summarizing single transcripts, but it cannot moderate, recruit, or aggregate at scale. Here is the workflow that actually works in 2026.","keywords":["chatgpt for user research","chatgpt user research","ai user research","chatgpt customer research","using chatgpt for research","ai research assistant","llm research tool","chatgpt vs research platform","ai user interviews","koji vs chatgpt"],"aiSummary":"ChatGPT excels at research-assistant tasks — drafting questions, summarizing single transcripts, brainstorming themes — but cannot moderate live interviews, recruit participants, aggregate across conversations, enforce quality, or produce defensible reports. For real research programs, pair ChatGPT for writing tasks with a purpose-built AI platform like Koji that handles moderation, recruitment, structured questions, and real-time aggregation end to end.","aiPrerequisites":["Basic familiarity with user research","Access to ChatGPT or another LLM"],"aiLearningOutcomes":["When ChatGPT is the right tool for a research task","Where ChatGPT breaks down at research scale","How to combine ChatGPT and a research platform","How to migrate from a ChatGPT-only workflow to Koji","Which research jobs need purpose-built tooling"],"aiDifficulty":"intermediate","aiEstimatedTime":"11 min read"},{"type":"documentation","id":"cdb39113-ecff-4542-8eb9-366be40f1754","slug":"best-ai-interview-software-2026","title":"Best AI Interview Software in 2026: 9 Platforms Compared","url":"https://www.koji.so/docs/best-ai-interview-software-2026","summary":"A 2026 buyer-style comparison of nine AI interview platforms. Koji is positioned as the best overall self-serve AI-native option for most teams, with participant recruitment built in and quoted per respondent in credits, and with Listen Labs and Strella recommended for enterprise-only use cases. Each platform is rated on modality, AI probing depth, structured question support, recruitment, analysis, and pricing transparency.","content":"## TL;DR — The Best AI Interview Software in 2026\n\n**The short answer:** [Koji](https://koji.so) is the best AI interview software for teams that want a fully self-serve, AI-native research platform with voice and text interviews, automatic analysis, and pricing that does not require a sales call. Enterprise teams with $25K+ annual research budgets may also evaluate Listen Labs (best for cross-study knowledge bases) or Strella (best for long-form video interviews). Analysis-only repositories like Marvin centralize transcripts but do not actually run interviews — you still need a moderator.\n\n**Quick verdict by buyer type:**\n- **Founders, PMs, and small teams** → Koji (10 free credits, self-serve, interviews as low as €1 per qualified interview, no subscription needed)\n- **Enterprises with dedicated researchers** → Listen Labs or Strella (custom pricing, $20K+)\n- **UX teams doing prototype walkthroughs** → Outset (Figma integration)\n- **Teams that want a research repository, not a moderator** → Marvin (analysis only)\n- **Anyone who wants to actually try the product before buying** → Koji (others are sales-led)\n\n## Why \"AI Interview Software\" Is the Fastest-Growing Research Category in 2026\n\nUser research has hit an inflection point. According to Maze's 2026 Future of User Research Report, 69% of research teams now use AI in at least some of their projects — a 19-point jump in a single year. Of those teams, 63% report faster turnaround and 60% report better team efficiency. The market research industry is on track to hit roughly $150 billion in 2026, and AI-moderated interviews are the fastest-growing segment inside it.\n\nThe shift is driven by three forces:\n\n1. **Survey response rates have collapsed.** Average response rates for traditional online surveys now hover at 5–10% — and most respondents speed-click through. Closed-ended forms can no longer surface the \"why\" behind a number.\n2. **Recruiting and scheduling moderated interviews is the bottleneck.** A typical 1:1 interview cycle takes 5–7 days from scheduling to transcript. AI moderators run 24/7 and remove the calendar entirely.\n3. **LLMs finally understand what a participant means.** Voice interviews with smart probing now extract laddering and root-cause data that surveys never could.\n\nIf you are evaluating AI interview platforms, the core question is no longer \"does AI work for research?\" — it is \"which AI interview platform fits my team's budget, workflow, and research depth?\"\n\n## How to Choose: 6 Criteria That Actually Matter\n\nBefore the comparison, here is the evaluation framework most buyers use in 2026:\n\n1. **Modality.** Voice, text, video, or all three? Voice produces 3–5x more words per response than text. Text is better when participants are at work or in noisy environments.\n2. **AI follow-up depth.** Does the AI actually probe (\"Why?\", \"What would change that?\", \"Tell me more\"), or does it just read scripted questions? This is the single biggest quality differentiator.\n3. **Structured + unstructured questions.** Can you mix open-ended exploration with NPS scales, ranked choice, and yes/no validations in the same interview?\n4. **Recruitment.** Does the platform supply participants, or do you bring your own list?\n5. **Analysis automation.** Themes, quotes, sentiment, and a publishable report — automatic, or DIY?\n6. **Pricing transparency.** Self-serve with public pricing, or \"talk to sales\" enterprise contracts?\n\nKeep these in mind as you read the platform breakdowns below.\n\n## The 9 Best AI Interview Platforms in 2026\n\n### 1. Koji — Best Overall AI Interview Software\n\n**Best for:** Founders, PMs, marketing teams, UX researchers, and anyone who wants an AI-native research platform without a sales process.\n\n**Pricing:** Interviews as low as €1 per qualified interview and €3 per qualified voice interview. Free account with 10 credits at signup, no card. Start with pay as you go. No subscription needed. You pay only for the interviews your study actually uses, and volume pricing and plans are there when you want them. There is no overage. You pay as you go, and you only ever pay for the conversations that bring insights. Voice interviews cost 3 credits, text interviews cost 1 credit, and report refreshes cost 5 credits. A \"credit\" only counts if the interview passes Koji's [quality gate](/docs/how-the-quality-gate-works) — low-quality conversations are free.\n\n**What makes Koji different:**\n- **Voice + text + structured questions in one interview.** Koji is the only platform that supports all 6 [structured question types](/docs/structured-questions-guide) (open_ended, scale, single_choice, multiple_choice, ranking, yes_no) inside an AI-moderated conversation. You get NPS distributions and themed quotes from the same interview.\n- **AI follow-up probing built in.** Koji's interviewer agent decides in real time whether to probe deeper, based on per-question follow-up depth and methodology-specific anti-patterns (Mom Test, JTBD, Customer Discovery).\n- **AI Consultant designs the brief for you.** Tell Koji what you are trying to learn; the [AI Consultant](/docs/working-with-the-ai-consultant) writes the research brief, picks the methodology, and drafts the question plan.\n- **Real-time insights.** Themes, quotes, sentiment, and quality scores update [live](/docs/real-time-research-insights) as interviews complete — no batch processing.\n- **Evidence you can follow.** Every claim in the report links to the quote behind it, so \"What did Pro tier users say about pricing?\" is one click from the transcript.\n- **MCP integration with Claude.** Run Koji from inside Claude: kick off a study, pull a report, and ask questions about your transcripts and study data. Paste the public report link into Slack when you want to share it.\n- **Self-serve, public pricing.** No sales call required. Sign up and ship a study in 15 minutes.\n\n**Trade-offs:** Koji does not sell a flat annual panel contract. Recruitment is built in and quoted per study instead: you describe the audience you want (market, demographics, role, screening questions), Koji returns a live per-respondent price in credits, and you approve it before anything launches. Multiple markets are supported, and you can soft launch a small batch before committing the full sample. If you already have an audience, Koji interviews it instead through CSV import, [personalized links](/docs/personalized-interview-links), or an open recruitment URL.\n\n### 2. Listen Labs — Best for Enterprise with Cross-Study Knowledge Bases\n\n**Best for:** Large enterprises running continuous research across brand, product, and UX with a dedicated research team and a $25K+ annual budget.\n\n**Pricing:** Enterprise sales-led. Public references suggest plans start around $3,000 with separate participant incentive credits, scaling to large enterprise contracts.\n\n**Strengths:**\n- 30M+ participant panel with global coverage and 100+ languages\n- \"Mission Control\" — a cross-study knowledge base that compounds insights over time\n- Multimodal emotional analysis (Ekman framework) on voice and video\n- Sequoia-backed; established enterprise brand trust\n\n**Trade-offs vs Koji:** Sales-led pricing with no self-serve option, no public free tier, longer time-to-first-study, and per-seat licensing that adds up fast for distributed teams. Most teams under 50 employees are over-served.\n\n### 3. Strella — Best for Long-Form Conversational Video\n\n**Best for:** Consumer research teams running 30–45 minute video interviews where natural conversational flow matters more than throughput.\n\n**Pricing:** Enterprise sales-led. Estimated $10K–$25K+ per study based on third-party reviews.\n\n**Strengths:**\n- Continuous voice/video interviews that average ~45 minutes — closer to a human researcher\n- 90% NPS from participants on session experience\n- Theme synthesis after each session\n\n**Trade-offs vs Koji:** Per-study pricing makes ongoing weekly research expensive. No structured question types in the interview itself, so quantitative validations live in a separate tool.\n\n### 4. Outset — Best for UX Prototype Walkthroughs\n\n**Best for:** UX teams testing Figma prototypes or guided product walkthroughs with screen-share.\n\n**Pricing:** Per-seat, approximately $20K per seat annually based on third-party reports.\n\n**Strengths:**\n- Figma integration and screen-share capture\n- Strong fit for evaluative UX research\n\n**Trade-offs vs Koji:** Format leans toward short recorded video clips with text-based AI questions, which limits conversational depth. Per-seat pricing is hard to justify for occasional users like founders or marketers.\n\n### 5. Marvin (HeyMarvin) — Best Repository for Existing Interviews\n\n**Best for:** Mature research teams that already conduct moderated interviews and want a single repository to store, code, and search across them.\n\n**Pricing:** Free trial; tiered plans (Free, Team, Enterprise). Specific pricing varies and is generally requested through their site.\n\n**Strengths:**\n- 20+ integrations (Pendo, Qualtrics, Google Forms, Zendesk, Intercom, etc.)\n- Auto-transcription and AI thematic analysis on uploaded recordings\n- \"Deep Research\" prompts for hypothesis testing and report generation across the repository\n\n**Trade-offs vs Koji:** Marvin **does not run the interview**. You still need a moderator (or another tool like Koji) to actually have the conversation. If you want one platform that conducts AND analyzes interviews, you need a true AI moderator.\n\n### 6. Conveo — Best for Quick Conversational Surveys\n\n**Best for:** Marketing teams replacing 5-question feedback surveys with 5-question AI conversations.\n\n**Pricing:** Subscription tiers; entry-level pricing competitive with mid-market survey tools.\n\n**Strengths:**\n- Fast setup, lightweight UI\n- Decent multilingual support\n\n**Trade-offs vs Koji:** Less depth on structured questions and quality gating. Reporting is more basic: fewer dashboards, and no MCP connector to query the data from your own assistant.\n\n### 7. Glaut — Best for European Market Research Agencies\n\n**Best for:** Agencies doing brand or concept testing in European markets.\n\n**Pricing:** Per-study or subscription, sales-led for larger volumes.\n\n**Strengths:**\n- Strong on translation quality across European languages\n- Agency-friendly white-label options\n\n**Trade-offs vs Koji:** Aimed at agencies, less self-serve for in-house teams. Workflow is heavier on study setup time.\n\n### 8. Feedbk — Best for In-Product Micro-Interviews\n\n**Best for:** SaaS product teams embedding 30-second AI conversations inside the app.\n\n**Pricing:** Tiered SaaS pricing.\n\n**Strengths:**\n- Easy embed widget for in-product placement\n- Good for high-volume, low-depth signal\n\n**Trade-offs vs Koji:** Designed for short signals, not deep discovery interviews. Koji also offers an [embed widget](/docs/using-the-embed-widget) plus full long-form interviews.\n\n### 9. User Intuition — Best for Motivational Laddering\n\n**Best for:** Researchers who specifically want laddering interviews with deep \"why\" probing.\n\n**Pricing:** $20 per interview, no monthly fees.\n\n**Strengths:**\n- 30-minute adaptive interviews with 5–7 levels of motivational laddering\n- Pure usage-based pricing\n\n**Trade-offs vs Koji:** Single methodology focus (laddering), no structured questions, no insights repository. Koji supports laddering through its [Mom Test](/docs/mom-test-user-interviews), JTBD, and Customer Discovery methodology frameworks plus other interview formats.\n\n## Side-by-Side: Quick Comparison Matrix\n\n| Platform | Modality | Self-Serve | Free Tier | Starting Price | Best For |\n|---|---|---|---|---|---|\n| **Koji** | Voice + text + structured | Yes | Yes (10 credits) | From €1 per qualified interview | All-around AI research |\n| Listen Labs | Voice + video + text | No (sales-led) | No | ~$3K (entry) | Enterprise knowledge base |\n| Strella | Voice + video | No (sales-led) | No | ~$10K/study | Long-form video |\n| Outset | Video clips + text | Limited | No | ~$20K/seat/yr | Figma prototype tests |\n| Marvin | Repository (no moderator) | Yes | Yes (limited) | Tiered | Existing interview storage |\n| Conveo | Voice + text | Yes | Limited | Mid-market SaaS | Quick conv. surveys |\n| Glaut | Voice + text | Limited | No | Sales-led | EU agencies |\n| Feedbk | Text + voice (short) | Yes | Yes | Tiered SaaS | In-product micro-interviews |\n| User Intuition | Voice (laddering) | Yes | Per-interview | $20/interview | Motivational laddering |\n\n## Honest Recommendation by Use Case\n\n**Customer discovery for an early-stage startup**\nUse Koji. Free tier covers your first interviews, the [Mom Test methodology](/docs/mom-test-user-interviews) framework keeps you honest, and you can ship a study in 15 minutes without a sales call.\n\n**Continuous product discovery for a PM team**\nUse Koji. The [continuous discovery workflow](/docs/continuous-discovery-user-research) plus a public report link you can paste into Slack means insights reach your standup without a hand-off. Listen Labs is overkill unless you have 5+ researchers.\n\n**B2B research with hard-to-reach buyers**\nUse Koji + your CRM. Import contacts, send [personalized links](/docs/personalized-interview-links), and run [async interviews](/docs/async-user-interviews) on the participant's schedule. Listen Labs's panel is mostly B2C.\n\n**Brand tracking across 10 markets**\nListen Labs is the strongest fit if budget exceeds $50K/year. Otherwise Koji handles 30+ languages with [multilingual research](/docs/multilingual-research-guide) at a fraction of the cost, and it recruits in multiple markets for you, quoted per respondent in credits before you commit.\n\n**Long-form, video-required executive interviews**\nStrella for long video sessions. Koji for voice/text at one-tenth the cost.\n\n**You already have 200 interview recordings to analyze**\nMarvin is built for this. But going forward, run new interviews in Koji so analysis happens automatically as conversations finish.\n\n## Why Koji Wins for Most Teams in 2026\n\nThe AI interview category is splitting into two camps: enterprise-only platforms with $20K+ contracts, and self-serve AI-native tools you can actually try. Koji is the most complete self-serve option because it bundles five things most competitors split into separate products:\n\n1. **AI Consultant for brief design** — replaces the research planning step\n2. **AI moderator for voice + text** — replaces the interviewer\n3. **Structured + open-ended questions in one flow** — replaces your survey tool\n4. **Auto-generated reports with every claim linked to its quote**: replaces manual synthesis\n5. **MCP and the REST API**: pipes insights into Claude and your CRM, and the public report link goes straight into Slack\n\nThe outcome: the same job that takes a traditional research team 3 weeks (recruit → schedule → moderate → transcribe → code → report) takes a Koji user about 90 minutes of active work — most of it in the first 15 minutes designing the brief.\n\n## Related Resources\n\n- [Structured Questions in AI Interviews](/docs/structured-questions-guide)\n- [AI Voice Interviews: The Definitive Guide for 2026](/docs/ai-voice-interviews-definitive-guide)\n- [AI Interviews vs. Surveys: Complete Comparison](/docs/ai-interviews-vs-surveys)\n- [Best User Research Tools in 2026](/docs/best-user-research-tools-2026)\n- [Koji vs Listen Labs](/docs/koji-vs-listen-labs)\n- [Koji vs Marvin](/docs/koji-vs-marvin)\n- [How AI Follow-Up Probing Works](/docs/ai-probing-guide)\n\n\n## Further reading on the blog\n\n- [Koji vs Listen Labs: AI Interview Platforms Compared (2026)](/blog/koji-vs-listen-labs-2026) — Listen Labs raised $69M and powers enterprise research at brands across the Fortune 500. Koji is the accessible AI-native alternative starti\n- [Koji vs Marvin: Full-Stack AI Research vs Analysis Repository (2026)](/blog/koji-vs-marvin-2026) — Marvin organizes research you've already collected. Koji runs new AI-moderated interviews from scratch. Here's how to choose — and when you \n- [The 11 Best AI Survey Tools in 2026 (Ranked & Reviewed)](/blog/best-ai-survey-tools-2026) — We tested 11 AI-powered survey platforms head-to-head — from AI question generation to automatic open-ended analysis to fully AI-moderated c\n\n<!-- further-reading:blog -->\n","category":"Comparisons","lastModified":"2026-10-29T13:18:35.581039+00:00","metaTitle":"Best AI Interview Software in 2026: 9 Platforms Compared","metaDescription":"Side-by-side review of Koji, Listen Labs, Strella, Outset, Marvin, Conveo, Glaut, Feedbk, and User Intuition. Pricing, depth, recruitment, and the right pick for each team.","keywords":["ai interview software","best ai interview software","ai interview platform","user research tools","ai moderated interviews","listen labs alternative","strella alternative","outset alternative","marvin alternative","koji"],"aiSummary":"A 2026 buyer-style comparison of nine AI interview platforms. Koji is positioned as the best overall self-serve AI-native option for most teams, with participant recruitment built in and quoted per respondent in credits, and with Listen Labs and Strella recommended for enterprise-only use cases. Each platform is rated on modality, AI probing depth, structured question support, recruitment, analysis, and pricing transparency.","aiPrerequisites":["Basic familiarity with user research workflows"],"aiLearningOutcomes":["Understand the AI interview software landscape in 2026","Compare Koji, Listen Labs, Strella, Outset, Marvin, Conveo, Glaut, Feedbk, and User Intuition","Apply a 6-criterion evaluation framework","Match the right platform to your team size and use case"],"aiDifficulty":"beginner","aiEstimatedTime":"14 minutes"},{"type":"documentation","id":"59f8e38f-5230-4547-b8f0-4c00ef854950","slug":"ai-survey-generator","title":"AI Survey Generator: Build Smart, Adaptive Surveys in Minutes (2026 Guide)","url":"https://www.koji.so/docs/ai-survey-generator","summary":"An AI survey generator translates a research goal into a complete, ready-to-launch questionnaire in seconds — drafting questions, picking response formats from six structured types (open_ended, scale, single_choice, multiple_choice, ranking, yes_no), and ordering them to minimize fatigue. Truly AI-native generators like Koji also probe follow-up questions during the response itself, lifting answer depth 3–5× over static forms while cutting design time 90%. Koji generates a full conversational interview from a two-sentence brief, runs it in voice or text across 30+ languages, and surfaces themes and quotes in real time as responses arrive. Free tier includes 10 credits.","content":"## The Bottom Line\n\nAn **AI survey generator** turns a research goal into a complete, ready-to-launch questionnaire in seconds — drafting the questions, picking response formats, ordering them logically, and (in modern systems) handling adaptive follow-ups during the response itself. Instead of staring at a blank canvas for an hour, you describe what you need to learn; the AI returns a structured guide you can publish immediately.\n\nTraditional survey builders (SurveyMonkey, Google Forms, Typeform) make you write every question yourself and hope respondents have the patience to finish. Koji's AI survey generator drafts the entire questionnaire from a two-sentence brief, mixes six structured question types (open-ended, scale, single-choice, multiple-choice, ranking, yes/no), and converts each \"survey\" into a conversational AI interview that probes intelligently when a participant gives a thin answer. The result: 90% less design time and 3–5× richer responses than a static form.\n\nThis guide walks through what AI survey generators do, how they differ from traditional builders, what to look for in 2026, and how to generate your first adaptive survey with Koji in under five minutes.\n\n## What an AI Survey Generator Actually Does\n\nA real AI survey generator does four things — anything less is a glorified template gallery.\n\n1. **Translates intent into questions.** You write \"I want to understand why beta users uninstall after week one.\" It returns 8–12 questions grouped by theme: onboarding experience, value perception, friction moments, win-back hooks. The questions are phrased neutrally (no leading prompts), avoid double-barrelled traps, and reflect real research methodology — not just keyword matching.\n2. **Picks the right format per question.** Some questions need a 1–10 scale (NPS, satisfaction). Some need single-choice (current tool). Some need open-ended with probing (\"walk me through the moment you decided to uninstall\"). A good generator chooses the format based on what each question is measuring rather than defaulting to multiple choice everywhere.\n3. **Orders questions for low fatigue.** Easy demographics and yes/no warmups go first. Scales and rankings sit in the middle. Open-ended reflection goes at the end when respondents are warmed up. This single ordering decision can lift completion rates by 15–25%, according to widely cited survey methodology research.\n4. **Adapts in real time** — only the best 2026 systems do this. When a respondent gives a one-word answer, the AI probes: \"Can you tell me more about that?\" When they describe a workaround, it asks: \"How often does that happen?\" Static surveys can't. Conversational AI surveys like Koji can.\n\nThe first three are table-stakes. The fourth — adaptive follow-up — is what separates AI-features-bolted-on surveys from a truly AI-native research platform.\n\n## Traditional Surveys vs. AI Survey Generators\n\nThe case for moving on from static survey tools is hard to argue with in 2026.\n\n| Capability | Traditional builder (Typeform, SurveyMonkey, Google Forms) | AI survey generator (Koji) |\n| --- | --- | --- |\n| Time to first draft | 45–90 minutes of manual writing | 30–60 seconds from a research goal |\n| Question quality check | None — you ship what you write | Methodology framework catches leading and double-barrelled questions |\n| Follow-up depth | Static skip logic only | AI probes 1–3 times per open-ended answer |\n| Response format mix | Whatever you remember to add | Auto-balanced across 6 structured types |\n| Multilingual | One survey per language, manually translated | One brief → 30+ languages, voice and text |\n| Analysis | CSV export → spreadsheet → manual coding | Themes, quotes, and quality scores in real time |\n| Time to first insight | Hours to days after collection | Seconds — themes update as each response lands |\n\nSurvey response rates have been declining for a decade — falling from roughly 40% in the mid-2010s to below 10% for many B2B audiences today, according to industry benchmarks tracked across platforms like Pew and Qualtrics. The reason is fatigue: respondents abandon long, impersonal forms. AI-generated conversational surveys reverse this trend because they feel like a chat, not a chore.\n\n## How Koji's AI Survey Generator Works\n\nKoji's generator is built into the platform — there's no separate \"AI mode\" to toggle. Here's the flow.\n\n### 1. Describe what you want to learn\n\nYou write 1–3 sentences in plain English. Examples:\n- \"Why are paying customers churning in month three?\"\n- \"How do RevOps teams evaluate new sales tools?\"\n- \"What features would push free users to upgrade?\"\n\nKoji's AI **Consultant** agent reads the goal and asks 2–4 clarifying questions back: who should participate, what decision the research informs, what methodology fits. This back-and-forth takes 60–90 seconds.\n\n### 2. The brief drafts itself\n\nThe Consultant generates a structured research brief: `problemStatement`, `decisionToInform`, `targetParticipant`, `methodology` (Mom Test, Jobs-to-be-Done, Customer Discovery, or custom), and an ordered list of `StudyQuestion` objects. Each question is tagged with one of six structured types so the system knows how to render it in text mode, ask it conversationally in voice mode, and visualize it in the report.\n\n### 3. You edit, review, publish\n\nThe brief opens in an editable view. You can rewrite any question, change a type from open-ended to scale, drop sections, or reorder. The methodology framework prevents you from accidentally introducing bad-research patterns — for example, the Mom Test framework will flag hypothetical-pricing questions and rewrite them to anchor on past behavior instead.\n\nHit publish and you get a shareable link, an embed snippet, or an API endpoint to inject the interview into your product.\n\n### 4. Responses come in — already coded\n\nEvery response, voice or text, gets analyzed in real time. The Analyst agent extracts themes, pulls verbatim quotes (kept in the participant's original language), assigns a 1–5 quality score per interview, and aggregates structured answers across the cohort. You watch themes form as responses land. No CSV export, no spreadsheet coding, no waiting until 50 respondents are in to spot patterns.\n\n## Six Question Types Every AI Survey Generator Should Support\n\nA serious AI survey generator should mix multiple response formats. Koji's six structured types are:\n\n- **Open-ended** — Free-form qualitative answers, with adaptive AI probing 1–3 times per question. Best for \"walk me through\" or \"tell me about a time\" prompts.\n- **Scale** — Numeric ratings (1–5, 1–10, NPS 0–10) with custom anchor labels. Distribution chart in reports.\n- **Single-choice** — Pick one from a list. Frequency bar chart.\n- **Multiple-choice** — Pick all that apply. Stacked frequency chart.\n- **Ranking** — Order items by preference. Average position chart.\n- **Yes / No** — Binary. Pie or donut chart.\n\nThe right mix typically looks like: 50–60% open-ended (where the qualitative depth lives), 15–25% scales (for benchmarking and trending), and 15–25% choice/ranking/yes-no (for fast segmentation). Most legacy survey tools default to choice questions because they are easiest to chart — but they sacrifice the richest insight in the process.\n\nSee the [structured questions guide](/docs/structured-questions-guide) for a deeper breakdown of when to use each type.\n\n## What to Look For in an AI Survey Generator (2026 Buyer's Checklist)\n\nIf you're evaluating AI survey tools, demand these capabilities. Anything missing is a 2022-era survey tool with a chatbot duct-taped on top.\n\n- **Goal-driven generation** — The generator should accept a research objective, not just keywords. \"Why did churn spike in March\" should yield a different brief than \"evaluate new sales tools.\"\n- **Methodology framework selection** — Mom Test, JTBD, Customer Discovery, and Generative are different research methods with different question conventions. The generator should know the difference.\n- **Adaptive probing** — Not a \"show another question if answer = X\" rule, but an actual LLM follow-up that reads the response and decides what to ask next.\n- **Multiple response formats per study** — Open-ended only is too thin; choice only is too shallow.\n- **Voice + text from the same brief** — One generated guide should run as either a typed conversation or a spoken interview without re-authoring.\n- **Live analysis** — Themes, quotes, and quality scores should appear as responses land, not after the study closes.\n- **Multilingual on day one** — A single brief should run in 28+ languages with native voices, not require manual translation.\n- **API and embed** — The generated survey should be embeddable on a website, runnable inside an app, and trigger-able from CRM events.\n\n## Common Use Cases for AI-Generated Surveys\n\nThe faster you can spin up a survey, the more surveys you can run — which changes what you're willing to ask about. With a 60-second generation cycle, teams using Koji typically run one of these per week instead of one per quarter:\n\n- **Post-onboarding pulse** — Generated from \"what is the new-user experience like?\" Runs as an in-app survey on day 7.\n- **Churn diagnostic** — Generated from \"why did this customer cancel?\" Auto-triggered from the cancel flow.\n- **Feature prioritization** — Generated from \"which of these five features matters most?\" Includes a ranking question and one open-ended \"why.\"\n- **Win/loss interviews** — Generated from \"why did we win or lose this deal?\" Sent to closed-won and closed-lost contacts in CRM.\n- **NPS deep-dive** — Generated from \"what drives our NPS detractors?\" Sent to anyone scoring 0–6 with a 90-second AI follow-up conversation.\n\nFor more on how to scale customer feedback, see [how to automate user research](/docs/how-to-automate-user-research).\n\n## Pricing: What an AI Survey Generator Costs in 2026\n\nMost legacy survey tools price by seat or response volume. Koji uses a credit model that aligns with the actual cost of AI inference:\n\n- **Free tier** — 10 credits on signup (no card required) — enough to run roughly 10 text or 3 voice interviews end-to-end.\n- **Outcome price**: interviews as low as €1 per qualified interview and €3 per qualified voice interview.\n- **Pay as you go**: no subscription, no seats, no minimums. You pay only for the interviews your study actually uses, and volume pricing and plans are there when you want them.\n- **Overage**: there is no overage. You pay as you go, and you only ever pay for the conversations that bring insights. If your balance runs out, your study pauses until you top up. Nothing is ever charged automatically.\n\nCrucially, **low-quality interviews (those scoring 1 or 2 out of 5) don't consume credits** — abandoned or junk sessions are free.\n\nCompare to Typeform Business at $83/month, SurveyMonkey Advantage at $39/month, or Qualtrics Research Core starting at roughly $1,500/year — none of which include AI follow-ups, multilingual voice, or live theme synthesis.\n\n## Generating Your First AI Survey: Step-by-Step\n\n1. **Sign up at koji.so** — Free tier, no credit card.\n2. **Click \"New study.\"**\n3. **Type your goal** — One or two sentences in plain English. \"I want to understand why beta users uninstall after week one\" works.\n4. **Answer the Consultant's 2–4 clarifying questions.** It takes 60–90 seconds.\n5. **Review the auto-generated brief** — Edit questions, change types, reorder sections.\n6. **Hit publish** — Get a shareable link, embed snippet, or API endpoint.\n7. **Share with respondents** — Internal Slack, email blast, in-app trigger, or [Zapier automation](/docs/zapier-research-automation).\n8. **Watch themes form in real time** — No CSV. No waiting. Open the report dashboard and see live aggregations as responses arrive.\n\nThe whole loop — goal to shareable link — takes under five minutes the first time and under 90 seconds once you know the flow.\n\n## When NOT to Use an AI Survey Generator\n\nAI generation is not always the right move:\n\n- **Pre-existing validated instruments.** If you're running a standardized scale (SUS, UEQ, validated psychometric instruments), don't regenerate it — load the canonical items as a template.\n- **Regulatory or legal disclosures.** Generated questions are great drafts but should be reviewed by counsel before going to regulated audiences.\n- **Single-question pulse polls.** A simple \"did this email help?\" thumbs-up/down doesn't need generation — just publish it directly.\n\nFor everything else — discovery, churn analysis, onboarding, NPS follow-ups, feature prioritization, design partner interviews — AI generation is faster, more methodologically sound, and easier to iterate on than manual authoring.\n\n## Related Resources\n\n- [Structured Questions Guide](/docs/structured-questions-guide) — Deep dive on the six question types every AI survey should support\n- [AI Interviews vs. Surveys: Complete Comparison with Data](/docs/ai-interviews-vs-surveys) — Why conversational AI outperforms static forms\n- [Conversational Survey Guide](/docs/conversational-survey-guide) — How chat-based surveys lift response rates and depth\n- [Survey Design Best Practices](/docs/survey-design-best-practices) — Foundational principles for any survey, AI-generated or hand-built\n- [Skip Logic Surveys Guide](/docs/skip-logic-surveys-guide) — How conditional branching works in modern survey tools\n- [Writing Interview Questions](/docs/writing-interview-questions) — Crafting questions that produce actionable answers\n- [Working With the AI Consultant](/docs/working-with-the-ai-consultant) — How Koji's Consultant agent builds your research brief","category":"Research Methods","lastModified":"2026-10-29T13:18:34.722678+00:00","metaTitle":"AI Survey Generator: Build Smart, Adaptive Surveys in Minutes (2026)","metaDescription":"An AI survey generator turns a research goal into a complete adaptive questionnaire in seconds. See how Koji generates 8–12 questions across 6 structured types and probes follow-ups in real time.","keywords":["ai survey generator","ai survey builder","ai questionnaire generator","smart survey tool","adaptive survey software","conversational survey generator","ai survey tool 2026"],"aiSummary":"An AI survey generator translates a research goal into a complete, ready-to-launch questionnaire in seconds — drafting questions, picking response formats from six structured types (open_ended, scale, single_choice, multiple_choice, ranking, yes_no), and ordering them to minimize fatigue. Truly AI-native generators like Koji also probe follow-up questions during the response itself, lifting answer depth 3–5× over static forms while cutting design time 90%. Koji generates a full conversational interview from a two-sentence brief, runs it in voice or text across 30+ languages, and surfaces themes and quotes in real time as responses arrive. Free tier includes 10 credits.","aiPrerequisites":["Basic familiarity with surveys or user research","A Koji account (free tier works)"],"aiLearningOutcomes":["Understand what an AI survey generator actually does (vs. a template gallery)","Compare AI-generated conversational surveys against legacy form builders","Identify the 6 structured question types every modern generator should support","Apply a 2026 buyer's checklist when evaluating AI survey tools","Generate your first adaptive survey in Koji in under 5 minutes"],"aiDifficulty":"beginner","aiEstimatedTime":"11 min read"},{"type":"documentation","id":"76eb2621-af32-4743-8718-34247d0a25c7","slug":"ai-interview-best-practices","title":"AI Interview Best Practices: 14 Rules for Running High-Quality AI-Moderated Customer Research","url":"https://www.koji.so/docs/ai-interview-best-practices","summary":"14 best practices for running high-quality AI-moderated customer interviews on platforms like Koji: write the brief like you're briefing a senior researcher, match each question type to the right job, sequence questions open-to-specific, set probing depth deliberately, choose mode by audience, recruit carefully, trust the quality gate, pilot before launch, avoid pre-coded open-ended questions, use scale + anchor probing, avoid leading questions, run time-boxed waves, chat with your report before presenting, and close the loop with participants and stakeholders.","content":"\n# AI Interview Best Practices: 14 Rules for Running High-Quality AI-Moderated Customer Research\n\nAI-moderated interviews are not the same as surveys, and they are not the same as moderated user interviews. They are a third research method with their own design patterns, failure modes, and best practices. Teams that get this right run 50 high-quality customer interviews in a week with a report ready by the following Monday. Teams that don't get this right ship a \"set and forget\" interview that produces shallow data and stale insights.\n\nThis guide is the practical playbook for running AI interviews that match or exceed the quality of moderated research. Fourteen rules, each with a specific recommendation and the reasoning behind it. Some are about study design, some are about question writing, some are about mode selection and recruiting. All of them are battle-tested by teams running customer research at scale on Koji.\n\nIf you're evaluating AI customer research tools or you've just shipped your first study and want to do the next one better, this is the document for you.\n\n## Rule 1: Write Like You're Briefing a Senior Researcher\n\nThe single highest-leverage thing you can do for AI interview quality is write a clear research brief. The brief is what the AI moderator uses to decide when to probe deeper, when to move on, and how to interpret ambiguous answers.\n\nGood briefs are specific about:\n- **The research goal**: What decision will this research inform?\n- **The hypothesis**: What do you believe, that this research might confirm or refute?\n- **The target audience**: Who should be in this study, and what background do they bring?\n- **What \"good\" looks like**: What kinds of answers would be especially useful to capture verbatim?\n- **What's out of scope**: Topics not to probe on, even if participants mention them.\n\nIf your brief reads like a survey title (\"Customer Satisfaction Q4\"), the AI has no anchor for judgment. If it reads like an internal Slack thread to a senior researcher (\"We're trying to figure out why churn spiked among Pro-tier customers in September — specifically whether the pricing change or the UI redesign drove it\"), the AI moderator has the context to interview like a senior researcher.\n\n## Rule 2: Use the Right Question Type for Each Job\n\nKoji supports six question types — `open_ended`, `scale`, `single_choice`, `multiple_choice`, `ranking`, `yes_no` — and each does one job well. Picking the wrong type at design time means worse data at analysis time.\n\nThe matching rules:\n- **Scale** for numeric ratings you'll track over time (NPS, CSAT, satisfaction)\n- **Single_choice** for mutually exclusive categorical answers (role, primary use case, journey stage)\n- **Multiple_choice** for non-exclusive selections (features used, competitors considered)\n- **Ranking** for explicit preference ordering\n- **Yes/no** for binary checkpoints (have they used X, did they complete Y)\n- **Open-ended** for everything else — discovery, \"why,\" capturing customer language\n\nThe most common mistake is overusing open-ended when a structured type would produce cleaner data, or overusing structured types when an open-ended would surface unexpected insights. A well-designed study uses three to five different types across 8-15 questions.\n\n## Rule 3: Sequence Questions From Open to Specific\n\nThe order of questions in an AI interview matters a lot. The pattern that works:\n\n1. **Warm-up**: An open-ended question that lets the participant orient themselves and the AI (\"Walk me through your role and what you do day to day\").\n2. **Discovery**: Open-ended questions about behavior, workflow, or experience (\"How do you currently handle X?\").\n3. **Structured anchors**: Scale or single_choice questions that capture comparable data.\n4. **Targeted probes**: Specific yes/no, multiple_choice, or scale questions that test hypotheses.\n5. **Catch-all close**: A final open-ended question that lets the participant raise anything important you didn't ask about (\"What's the most important thing about this topic that I didn't ask?\").\n\nThis sequence avoids priming early answers with structured options, and it gives the catch-all close — often where the best insights surface.\n\n## Rule 4: Set Probing Depth Deliberately\n\nEvery Koji question has a `maxFollowUps` setting. Defaults are reasonable but not optimal for every question.\n\nGuidelines:\n- **0 follow-ups**: For demographic questions or fast pulse-checks where the answer is the answer.\n- **1 follow-up (default)**: For most structured questions where you want a qualitative why.\n- **2-3 follow-ups**: For the two or three most important questions in the study — the ones whose answers determine the research outcome.\n\nDon't set everything to 3. Interviews that probe too deeply on every question feel like interrogations and drop completion rates. Reserve depth for the questions where it actually matters.\n\n## Rule 5: Choose Mode by Audience, Not by Researcher Preference\n\nVoice mode and text mode produce different kinds of data. Voice mode tends to produce longer, more reflective open-ended answers and feels more like a moderated interview. Text mode is faster, more comfortable for many B2B professionals, and produces cleaner data when participants want time to compose answers carefully.\n\nThe right mode depends on the audience, not the researcher:\n- **Consumer audiences, narrative topics**: Voice mode usually wins.\n- **B2B professionals, technical topics, work-hours research**: Text mode usually wins.\n- **International audiences with English-as-second-language**: Text mode is gentler.\n- **Sensitive topics**: Anonymous text mode often gets more honest answers.\n\nIf you're not sure, run a small pilot in both modes and compare the depth and completion rates before committing.\n\n## Rule 6: Recruit Like the Quality of the Answers Depends on It\n\nIt does. The single biggest determinant of insight quality in an AI interview is not the AI — it is the participant. Five carefully recruited customers will produce better research than fifty random panel respondents.\n\nFor B2B research:\n- Use targeted recruitment from your CRM, customer list, or LinkedIn outreach.\n- Pay incentives that match the audience (executives need $200+ to participate; power users might accept a $50 gift card).\n- Pre-screen with a one-question screener to confirm relevance before booking the interview.\n\nFor consumer research:\n- Recruit from your existing user base when possible — they have context the AI can work with.\n- Use a participant panel only when targeting an audience you can't access directly.\n- Always screen for relevance before paying the incentive.\n\nGood recruiting is the foundation. With tools like Koji, AI moderation removes the bottleneck on running interviews — but recruitment is still where most of the work happens.\n\n## Rule 7: Use the Quality Gate, Don't Override It\n\nKoji's analysis pipeline scores every interview on quality dimensions: completeness, depth, coherence, and engagement. Low-quality interviews are flagged automatically before they pollute your aggregated report.\n\nResist the temptation to override the quality gate. If 10 of your 50 interviews scored low, those 10 are giving you signal — usually one of three things:\n- The screener missed a participant pool that doesn't match the audience\n- The brief was unclear about what depth of answer counted as complete\n- The interview length was too long for the actual content of the study\n\nUnderstanding why interviews failed quality is often more valuable than rerunning the study. The quality gate is built to protect your final report from noise — let it do its job, and use the failed interviews as a research signal about your study design.\n\n## Rule 8: Pilot Every Study Before Full Launch\n\nThe single best way to catch design mistakes is to run a tiny pilot — three to five interviews — before launching the full study. Pilots catch:\n- Questions that participants consistently misunderstand\n- Options that don't cover the answer space (lots of \"Other\" responses)\n- Probing instructions that produce off-tone follow-ups\n- Study length that feels too long or too rushed\n- Brief language that the AI is interpreting differently than you intended\n\nAfter the pilot, read the transcripts (not just the aggregated themes) and adjust the brief, questions, or probing instructions before launching to the full sample.\n\n## Rule 9: Don't Pre-Code Open-Ended Questions\n\nOne of the most common mistakes in open-ended question design is hiding a structured question inside an open-ended one. \"Tell me about your biggest pain point — is it pricing, ease of use, or integration?\" is not an open-ended question. It is a multiple_choice question with a misleading prompt.\n\nIf you want frequency data on a known set of pain points, use `multiple_choice`. If you want to discover the pain points participants generate spontaneously, use `open_ended` with no leading options. Mixing the two corrupts both.\n\n## Rule 10: Use Scale Questions With Anchor Probing\n\nA scale question without follow-up is just a number. A scale question with `anchor: true` enabled is a research artifact — the AI follows up with \"What would change that?\" or \"Walk me through what would have made that a 9 or 10?\", capturing the qualitative why behind the score.\n\nThis is the single most powerful pattern in Koji: scale + anchor probing. It compresses what would otherwise be two separate questions (\"Rate it\" + \"Why?\") into a single conversational moment, and it produces a structured number plus a synthesized qualitative summary at report time.\n\nDefault to enabling anchor probing on every scale question that matters.\n\n## Rule 11: Avoid Leading and Loaded Questions\n\nThe AI moderator follows the script you write. If your questions lead participants toward a desired answer, the data will reflect that bias. Watch for:\n\n- **Presupposing answers**: \"What did you love about onboarding?\" presupposes love. \"Walk me through your onboarding experience\" is neutral.\n- **Loaded terms**: \"How frustrating was the slow loading?\" is loaded. \"How did the loading experience compare to what you expected?\" is neutral.\n- **Confirmatory probing instructions**: Instructions like \"Probe for the pricing concern\" can bias the AI toward fishing for that specific concern. Instead, write \"Probe for the most important factor in their decision.\"\n\nResearch bias is a deep topic — Koji's separate guide on research bias covers this in more detail. The short version: write questions like you're a journalist, not a marketer.\n\n## Rule 12: Run Studies in Waves, Not Months-Long Continuous Modes\n\nThe biggest mistake teams make with always-on AI interviews is letting them run indefinitely. The data drifts: your product changes, your customer base evolves, the language they use shifts. After three months, the early responses and the late responses are answering subtly different studies.\n\nThe better pattern:\n- Run studies in **time-boxed waves** (1-2 weeks each).\n- Refresh the brief between waves based on what you learned.\n- Tag each wave with a date so you can track how answers change over time.\n- For continuous discovery: think of it as weekly waves rather than always-on, with the brief refreshed weekly.\n\nThis discipline produces longitudinal data you can actually trust.\n\n## Rule 13: Chat With Your Report Before You Present It\n\nKoji generates a structured research report from every study — but the most valuable feature is the ability to chat with your report after it's generated. This is where prep for stakeholder presentations happens.\n\nBefore presenting:\n- Ask the report the questions stakeholders are likely to ask: \"What did Enterprise customers say about pricing?\" \"Did anyone mention competitor X?\"\n- Probe for counter-evidence: \"Are there participants who disagreed with this theme?\"\n- Find your strongest verbatim quotes: \"Give me the most compelling quote about feature Y.\"\n- Stress-test your headline insight: \"How many participants does this finding actually rely on?\"\n\nGoing into a stakeholder meeting having already chatted with the data means you'll handle every \"but what about...\" question with evidence — not vibes.\n\n## Rule 14: Close the Loop With Participants and Stakeholders\n\nThe last 5% of value in any research study is closing the loop. Skip this and the next study will have a harder time recruiting and getting stakeholder buy-in.\n\nWith participants:\n- Send a brief thank-you message summarizing what you learned at a high level (without breaking anonymity).\n- For interview participants who shared specific feature requests, share when those features ship.\n- Build a participant panel of customers who've done research with you — they'll respond faster and give better answers the second time.\n\nWith stakeholders:\n- Tag every research finding to the decision it informed.\n- Track which findings led to shipped changes and which didn't.\n- Send a quarterly \"what research said and what we shipped\" recap to maintain stakeholder buy-in for the research program.\n\nThis is research ops at its best — not a separate role, but a discipline every team can practice.\n\n## What These Rules Add Up To\n\nFollow these 14 rules and your AI interviews will produce research-grade insights at a pace traditional methods can't match: 30 interviews in a week, a report ready the next day, and confidence that the findings will hold up to stakeholder scrutiny.\n\nSkip these rules and your AI interviews will produce noisy, shallow data that confirms whatever bias you walked in with. The technology is the same in both cases — the difference is the rigor you bring to the design.\n\nPlatforms like Koji handle the heavy lifting of moderation, transcription, coding, and synthesis. Your job is to design the studies that make that automation produce trustworthy research. These 14 rules are the foundation.\n\n## Related Resources\n\n- [Structured Questions Guide: All 6 Koji Question Types](/docs/structured-questions-guide)\n- [How to Write a Research Brief](/docs/how-to-write-research-brief)\n- [How Koji's AI Probing Works](/docs/ai-probing-guide)\n- [Understanding Quality Scores](/docs/understanding-quality-scores)\n- [Voice vs Text Interview: When to Use Each Mode](/docs/voice-vs-text-interviews)\n- [Pilot Study in User Research](/docs/pilot-study-user-research-guide)\n- [Research Bias: The Complete Guide](/docs/research-bias-guide)\n","category":"Interview Techniques","lastModified":"2026-10-29T13:18:34.108183+00:00","metaTitle":"AI Interview Best Practices: 14 Rules for Quality Research | Koji","metaDescription":"A practical playbook for running AI-moderated customer interviews that produce research-grade insights. 14 rules covering brief design, question types, probing depth, mode selection, recruiting, quality gates, and stakeholder buy-in.","keywords":["ai interview best practices","ai customer interview tips","how to run ai interviews","ai moderator best practices","ai interview design","quality ai research"],"aiSummary":"14 best practices for running high-quality AI-moderated customer interviews on platforms like Koji: write the brief like you're briefing a senior researcher, match each question type to the right job, sequence questions open-to-specific, set probing depth deliberately, choose mode by audience, recruit carefully, trust the quality gate, pilot before launch, avoid pre-coded open-ended questions, use scale + anchor probing, avoid leading questions, run time-boxed waves, chat with your report before presenting, and close the loop with participants and stakeholders.","aiPrerequisites":["Basic familiarity with AI-moderated interview platforms","Understanding of qualitative vs quantitative research"],"aiLearningOutcomes":["Design AI interview studies that produce research-grade insights","Pick the right Koji question type for each research goal","Configure AI probing depth deliberately rather than using defaults","Choose voice vs text mode based on audience characteristics","Use Koji's quality gate as a research signal rather than overriding it","Run AI interview studies in time-boxed waves for trustworthy longitudinal data"],"aiDifficulty":"intermediate","aiEstimatedTime":"14 min read"},{"type":"blog","id":"01f817b2-bc5f-4114-befe-b89d1987637a","slug":"voxpopme-alternatives-2026","title":"7 Best Voxpopme Alternatives in 2026 (Video and AI Interviews)","url":"https://www.koji.so/blog/voxpopme-alternatives-2026","summary":"The best Voxpopme alternatives in 2026 are Koji, Conveo, Outset, Listen Labs, VideoAsk, Rival Technologies and Qualtrics. Voxpopme offers video surveys, live interviews and an AI Moderator for async video interviews (launched June 2025), with self-serve tiers and quote-based Enterprise. Koji is the best alternative when video is optional: an AI interviewer runs voice or text conversations with live probing, only interviews scoring 3 or more are charged, panel recruitment covers 130+ countries, and pricing starts free. Conveo, Outset and Listen Labs are closest when recorded video is required; VideoAsk is the budget async video option.","content":"The best Voxpopme alternatives in 2026 are **Koji, Conveo, Outset, Listen Labs, VideoAsk, Rival Technologies and Qualtrics**. If you came to Voxpopme for richer open-ended answers than a survey can give, and you do not strictly need video on file, Koji is the strongest alternative: an AI interviewer runs voice or text conversations, probes each answer, and writes the report, with published prices from free. If recorded video is the deliverable, Conveo and Outset are the closest AI-moderated matches, and VideoAsk is the budget option for simple async video.\n\nVoxpopme has been a pioneer in video feedback since 2013 and is used by large consumer brands such as PepsiCo and Mars. In 2025 it added an AI Moderator. But video carries real costs for participants and researchers, and the AI interview category has moved quickly. Here is how the alternatives compare.\n\n## What Voxpopme does today\n\nVoxpopme sells three products: video surveys, live interviews and AI insights ([pricing page](https://www.voxpopme.com/pricing)). In June 2025 it [launched AI Moderator](https://www.voxpopme.com/learn/blog/voxpopme-launches-ai-moderator/), which runs asynchronous AI-moderated video interviews with a built-in screener and demographic targeting. In September 2025 it [added smart probing and stimulus testing](https://www.voxpopme.com/learn/blog/voxpopme-expands-ai-moderator-with-smart-probing-and-stimulus-testing/). It integrates with survey platforms such as Qualtrics, Alchemer and Forsta.\n\nOn pricing, Voxpopme lists self-serve Solo, Team and Pro tiers, with the Solo plan under $200 a month for one user and access to its audience sold as a paid add-on. Enterprise pricing is quote-based. Published figures have varied between snapshots, so check the live page.\n\n## Why teams look for Voxpopme alternatives\n\n**Video asks a lot of participants.** Recording yourself on camera is a bigger ask than typing or talking. Some people decline, others give short, self-conscious answers, and reviewers on [Capterra](https://www.capterra.com/p/189753/Voxpopme/reviews/) mention uneven respondent video quality. Fewer completes means higher cost per usable answer.\n\n**Open-ended answers are hard to get in the first place.** Pew Research Center's analysis of its American Trends Panel found that [closed-ended questions average 1% to 2% nonresponse, while open-ended questions average about 18%](https://www.pewresearch.org/decoded/2021/10/14/why-do-some-open-ended-survey-questions-result-in-higher-item-nonresponse-rates-than-others/), ranging from 3% to over 50%. Asking someone to record a video answer adds friction on top of that. A conversational format that asks one question at a time and follows up tends to be easier to finish.\n\n**Follow-up questions are where the value is.** Qualtrics reported in 2024 that when its AI asked a follow-up question, [respondents added more information about 40% of the time](https://siliconangle.com/2024/05/01/qualtrics-generative-ai-gives-surveys-real-time-twist/). A single recorded answer misses that second layer unless the moderator probes well.\n\n**AI is now the default, not the add-on.** Qualtrics' [2026 Market Research Trends report](https://www.qualtrics.com/articles/strategy-research/market-research-trends) found that **95% of researchers use AI tools regularly or are experimenting with them**. Platforms built around AI from the start often feel simpler than older tools that have added it later.\n\n## Voxpopme alternatives compared\n\n| Tool | Format | AI moderation | Recruiting | Public pricing |\n| --- | --- | --- | --- | --- |\n| **Koji** | Voice and text interviews | Yes, live probing | Built-in panel (130+ countries) or own users | Yes, from free |\n| **Conveo** | Video, audio, text | Yes | Panel incentives on paid plan | Partly |\n| **Outset** | Video, voice, text | Yes | 25+ panel integrations | No |\n| **Listen Labs** | Voice, video, text | Yes | Built-in network | No |\n| **VideoAsk** | Async video | Not its focus | Own audience | Yes |\n| **Rival Technologies** | Video, voice, chat surveys | AI analysis of answers | Own panel (Angus Reid) | No |\n| **Qualtrics** | Surveys with video feedback | AI follow-ups | Panels via partners | No |\n\n## 1. Koji\n\n[Koji](https://www.koji.so) replaces the \"record your answer\" model with a conversation. An AI interviewer talks with each participant by voice or chats by text, asks the questions from your brief, and probes vague answers in real time. People who would rather not appear on camera can still talk it through. Koji then scores each interview and builds an aggregated report with themes, sentiment, segment filters and quotes you can share as a public link.\n\nWhy Koji is a strong Voxpopme alternative:\n\n- **Voice without the camera.** Participants choose to talk or type. Voice interviews capture tone and detail without asking anyone to appear on video. Read about [voice vs text interviews](/docs/voice-vs-text-interviews).\n- **Only pay for good answers.** Every interview gets a quality score, and only interviews scoring 3 or more use credits or enter the report. The score is capped by how much the participant actually said, so one-word answers and filler do not count. See [understanding quality scores](/docs/understanding-quality-scores).\n- **Panel built in.** Recruit from panel partners in 130+ countries from the study's Panel recruit tab, see the price before you launch, and soft-launch part of the goal first to check quality. Credits are held at launch and unused credits come back. See [panel recruitment](/docs/panel-recruitment).\n- **Screeners on every route.** [Screening questions](/docs/adding-screening-questions-to-your-study) run before every interview, whether the person came from the panel, an email or your website.\n- **Clear pricing.** Free (10 one-time credits), Insights €29/month, Interviews €79/month, Enterprise from €500/month. A text interview costs 1 credit and a voice interview 3. See the [plan comparison guide](/docs/plan-comparison-guide).\n- **For agencies and brands.** Enterprise plans add [brand themes](/docs/brand-themes) so each client's interviews carry their logo and colours. Interviews run in 30+ languages.\n\nThe honest trade-off: Koji does not record participant video, so it will not produce a video highlight reel of faces for a boardroom. If that reel is the deliverable, choose Conveo or Outset. If the deliverable is understanding why customers behave the way they do, Koji is built for exactly that, with a lower entry price and no sales call. See our full [Koji vs Voxpopme](https://www.koji.so/blog/koji-vs-voxpopme-2026) comparison.\n\n## 2. Conveo\n\nConveo is the closest modern match for Voxpopme's consumer brand customers. Its AI interviewer asks dynamic follow-up questions, and participants answer by video, audio or text. It raised a [$50M Series A led by DST Global](https://conveo.ai/mediakit/series-a) in September 2026 and says more than 400 enterprises use it, including Google, Unilever and AB InBev. Conveo publishes a Free plan and a Starter plan at $1,250/month billed annually, with Enterprise on request.\n\n## 3. Outset\n\nOutset runs AI-moderated interviews by video, voice and text in more than 40 languages and connects to more than 25 recruiting panels. It raised a [$30M Series B](https://outset.ai/series-b) in December 2025 and serves enterprise insights teams at companies such as Microsoft and Nestlé. Pricing is custom.\n\n## 4. Listen Labs\n\nListen Labs runs AI interviews at large scale and is being acquired by Salesforce, which [signed a definitive agreement](https://www.salesforce.com/news/stories/salesforce-signs-definitive-agreement-to-acquire-listen-labs/) on September 29, 2026. Salesforce cites a network of more than 50 million potential participants and support for more than 120 languages. Pricing is not public.\n\n## 5. VideoAsk\n\nVideoAsk, from Typeform, is a simple async video tool: you record a question, people reply by video, audio or text. It has a free tier and low-cost paid plans priced by processing minutes. It is built for quick async replies rather than AI-moderated probing interviews, so it suits testimonials and lightweight feedback more than research.\n\n## 6. Rival Technologies\n\nRival runs mobile-first chat surveys with video and voice answers. It [merged with Angus Reid Group](https://www.mrweb.com/drno/news38643.htm) in 2025 to form Rival Group, which gives it its own community panels. Pricing is not public. A good fit for market research teams that want community-style engagement.\n\n## 7. Qualtrics\n\nIf you already run Qualtrics, its video feedback and AI follow-up questions may cover part of what you used Voxpopme for, without another vendor. Qualtrics made AI video summaries [generally available in 2023](https://www.businesswire.com/news/home/20230815261930/en/Qualtrics-Announces-General-Availability-of-new-AI-Powered-Qualitative-Research-Tools-That-Provide-Smarter-Faster-Insights). Pricing is quote-based and usually part of a wider contract.\n\n## How to choose\n\n- **You need video clips of real customers:** Conveo, Outset or Listen Labs, or stay with Voxpopme.\n- **You need depth and volume, and video is optional:** Koji. Voice and text interviews with AI probing, a quality gate and a built-in panel.\n- **You need quick async video replies on a budget:** VideoAsk.\n- **You are standardised on Qualtrics:** try its built-in video and AI follow-ups first.\n\nDemand for this kind of work keeps rising. Maze's [Future of User Research 2026](https://maze.co/blog/future-user-research-2026/) report found that **research demand grew 20% year over year**. The winning setup is the one that turns more conversations into decisions with the same team.\n\n## Try Koji\n\nKoji gives you the depth Voxpopme promised, without asking every participant to go on camera. Write your brief, [preview the interview](/docs/previewing-your-interview) for free, recruit from the panel or share your link, and pay only for interviews that pass the quality gate.\n\n[Start free on Koji](https://www.koji.so/signup) and hear from your customers this week.","category":"Comparisons","lastModified":"2026-10-29T13:18:33.839847+00:00","metaTitle":"7 Best Voxpopme Alternatives in 2026 | Video and AI Interviews","metaDescription":"Compare Voxpopme alternatives for 2026: Koji, Conveo, Outset, Listen Labs, VideoAsk, Rival Technologies and Qualtrics. Formats, AI moderation, panels, pricing.","keywords":["voxpopme alternatives","voxpopme competitors","voxpopme pricing","video survey alternatives","video feedback platforms","ai video interviews","voxpopme vs koji"],"aiSummary":"The best Voxpopme alternatives in 2026 are Koji, Conveo, Outset, Listen Labs, VideoAsk, Rival Technologies and Qualtrics. Voxpopme offers video surveys, live interviews and an AI Moderator for async video interviews (launched June 2025), with self-serve tiers and quote-based Enterprise. Koji is the best alternative when video is optional: an AI interviewer runs voice or text conversations with live probing, only interviews scoring 3 or more are charged, panel recruitment covers 130+ countries, and pricing starts free. Conveo, Outset and Listen Labs are closest when recorded video is required; VideoAsk is the budget async video option.","aiKeywords":["voxpopme alternatives","video feedback","ai moderated interviews","voice interviews","conveo","videoask","open-ended responses","koji"],"aiContentType":"comparison","faqItems":[{"answer":"The top Voxpopme alternatives in 2026 are Koji, Conveo, Outset, Listen Labs, VideoAsk, Rival Technologies and Qualtrics. Koji is best when you want depth without requiring video; Conveo and Outset are best when recorded video is essential.","question":"What are the best Voxpopme alternatives?"},{"answer":"Voxpopme lists self-serve Solo, Team and Pro plans on its pricing page, with the Solo plan under $200 a month for one user and audience access sold as an add-on. Enterprise pricing is quote-based. Published figures have changed over time, so check the live page.","question":"How much does Voxpopme cost?"},{"answer":"Yes. Voxpopme launched AI Moderator in June 2025 for asynchronous AI-moderated video interviews and added smart probing and stimulus testing in September 2025.","question":"Does Voxpopme have AI-moderated interviews?"},{"answer":"It depends on the deliverable. Video is useful when you need clips of real customers for stakeholders. Voice and text interviews ask less of participants, which can make them easier to complete, while still capturing detailed answers that an AI interviewer can probe.","question":"Is voice or video better for customer research?"},{"answer":"No. Koji runs AI-moderated voice and text interviews and does not record participant video. It offers a built-in panel in 130+ countries, a quality gate so you only pay for interviews scoring 3 or more, and plans from free.","question":"Does Koji support video interviews?"}],"relatedTopics":["video feedback","ai moderated interviews","voice interviews","open-ended survey responses","voxpopme pricing"]},{"type":"blog","id":"e7a9b6c7-7657-4324-83ed-3d438512d791","slug":"usertesting-vs-user-interviews-2026","title":"UserTesting vs User Interviews (2026): Same Owner, Different Jobs","url":"https://www.koji.so/blog/usertesting-vs-user-interviews-2026","summary":"UserTesting and User Interviews share an owner since UserTesting acquired User Interviews on January 7, 2026; the company rebranded as Auros in September 2026, keeping both as product brands. UserTesting is a research platform for recorded unmoderated tests and live moderated sessions, sold on annual contracts with a Vendr-reported median of $41,200 a year. User Interviews is a recruiting marketplace that finds, screens, schedules and pays participants, from $49 per completed consumer session and $98 for B2B, with incentives extra. Neither runs and analyses interviews for you. Koji is an AI interview alternative with published per-interview pricing, panel recruitment on paid plans and automatic reports, but no video recording.","content":"# UserTesting vs User Interviews (2026): Same Owner, Different Jobs\n\n**Short answer:** UserTesting and User Interviews now belong to the same company, but they still do different jobs. **UserTesting** is a research platform: you build tests, its contributor network takes them, and you watch videos of real people using your product, unmoderated or live. **User Interviews** is a recruiting marketplace: it finds, screens, schedules and pays participants, and you run the sessions in your own tools. UserTesting is sold on annual contracts with no public prices; Vendr puts the median at **$41,200 a year**. User Interviews publishes a rate card starting at **$49 per completed session**, with incentives on top. Since January 2026 both sit under one owner, which in September 2026 renamed itself **Auros**. If what you really need is to talk to customers and get the analysis done for you, [Koji](https://www.koji.so) runs AI-moderated interviews at a published price per interview.\n\n---\n\n## What changed in 2026\n\nThis comparison used to be about two separate vendors. Not any more.\n\n- **January 7, 2026:** UserTesting announced it had acquired User Interviews ([Business Wire](https://www.businesswire.com/news/home/20260107716032/en/UserTesting-Acquires-User-Interviews-to-Strengthen-the-Industrys-Most-Comprehensive-Customer-Insights-Solution-for-the-AI-Era)). At the time, UserTesting said it served **more than 3,000 customers, including 75 of the Fortune 100**, and User Interviews said its network held **more than 6 million participants**. Both said there were no immediate changes to products, pricing or workflows.\n- **August 12, 2026:** UserTesting launched Advanced Targeting, which brings the User Interviews participant network directly into UserTesting, giving customers access to **more than 7 million participants across 34 countries** ([UserTesting](https://www.usertesting.com/company/newsroom/press-releases/usertesting-introduces-advanced-targeting-bringing-user-interviews)).\n- **Late September 2026:** the company rebranded as **Auros**, \"the human network behind AI\", built on what it calls **7.6 million verified people** ([Business Wire](https://www.businesswire.com/news/home/20260929631519/en/UserTesting-Announces-Rebranding-as-Auros-Providing-the-Human-Network-Behind-AI)). Auros is now the corporate brand and the name for its AI data, evaluation and research services. UserTesting and User Interviews continue as product brands, and the company says contracts and day-to-day operations are unchanged.\n\nSo the real question for a buyer is no longer \"which company?\" It is \"which of these two products does the job I have?\", and whether either is the right shape for how your team researches.\n\n---\n\n## UserTesting vs User Interviews at a glance\n\n![Table comparing UserTesting and User Interviews on what each is, participants, who runs sessions, analysis, pricing model and ownership.](/visuals/usertesting-vs-user-interviews-comparison-table/usertesting-vs-user-interviews-comparison-table.png \"One owner, two products: one to run and watch tests, one to find the people.\")\n\n| | UserTesting | User Interviews |\n|---|---|---|\n| What it is | A research platform for watching people use your product | A recruiting marketplace and participant CRM |\n| Core job | Unmoderated video tests and live moderated sessions | Find, screen, schedule and pay participants |\n| Participants | Its contributor network, your own users, or Advanced Targeting | Its own panel, or your customers through Research Hub |\n| Who runs the session | The platform records it; a person moderates live sessions | You do, in your own video or testing tool |\n| Analysis | AI summaries and highlight reels on recorded tests | Not included |\n| Pricing model | Annual contract with seats and credits, no public prices | Published fee per completed session, incentives extra |\n| Ownership | Auros, since the 2026 rebrand | Auros, after UserTesting bought it in January 2026 |\n\n---\n\n## UserTesting: watch people use your product\n\nUserTesting is built around video. You write a test with tasks and questions, send it to participants, and get back recordings of them thinking aloud as they use your website, app or prototype. For live work, its Live Conversation feature lets you schedule a moderated interview with someone from the contributor network or one of your own customers.\n\n**Where it is strong:**\n\n- Fast unmoderated usability tests on consumer products, with results often back the same day.\n- Video clips that show a stakeholder exactly where people get stuck.\n- Enterprise features: single sign-on, role controls, a customer success manager, and now the User Interviews network through Advanced Targeting.\n\n**What it costs:** UserTesting does not publish prices. You buy an annual contract made of platform seats and credits that pay for participant sessions. Vendr's data across 359 purchases puts the **median at $41,200 a year**, with a range from about $12,000 to $113,642 ([Vendr](https://www.vendr.com/marketplace/usertesting)). Our [UserTesting pricing guide](/blog/usertesting-pricing-2026) breaks down seats, credits and renewal terms.\n\n**Where it is weaker:** the panel leans toward consumers, so senior B2B roles are slower and dearer to fill. And live interviews still need a person to moderate every one.\n\n---\n\n## User Interviews: find the right people\n\nUser Interviews does not run your research. It gets the right people into your calendar. You describe who you need, write a screener, and it recruits from its panel, schedules sessions, sends reminders and pays incentives. Research Hub adds a participant CRM so you can manage your own customers as a panel.\n\n**Where it is strong:**\n\n- A large panel and detailed targeting, including B2B roles through an Advanced B2B targeting add-on.\n- A published rate card, which is rare in this market.\n- Research Hub, for managing your own customers as a reusable panel.\n\n**What it costs:** User Interviews' pricing page lists **$49 per completed session** on Pay As You Go for consumer participants and **$98 for B2B**, with Advanced B2B targeting as an add-on ([User Interviews](https://www.userinterviews.com/pricing)). Incentives are paid on top. The same page now warns that **Pay As You Go usage limits are coming soon** and steers buyers toward subscriptions, so check the terms before you plan a year around it. We worked through real study budgets in [User Interviews pricing](/blog/user-interviews-pricing-2026).\n\n**Where it is weaker:** it stops at the calendar invite. You still moderate every session, take notes, and do the analysis.\n\n---\n\n## Running the sessions is the hidden cost\n\nNeither product removes the most expensive part of qualitative research: the hours a person spends running and making sense of conversations. With User Interviews, a researcher moderates every call. With UserTesting's live sessions, the same is true; its unmoderated tests skip moderation but cannot ask a follow-up when an answer gets interesting.\n\n![A researcher on a video call at a laptop taking notes, the hours of moderation a recruited session still needs](https://images.unsplash.com/photo-1616587226960-4a03badbe8bf?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHwxfHxyZXNlYXJjaGVyJTIwdmlkZW8lMjBjYWxsJTIwbGFwdG9wJTIwbm90ZXN8ZW58MXwwfHx8MTc5MjM4OTAxNHww&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [LinkedIn Sales Solutions](https://unsplash.com/@linkedinsalesnavigator?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\nThat time adds up. [Lyssna's 2025 synthesis survey](https://www.lyssna.com/blog/research-synthesis-challenges/) of 300 practitioners found that **60.3% say time-consuming manual work is their biggest frustration**, and reading through responses was the most time-consuming step for **59%**. Meanwhile demand keeps climbing: Maze's 2026 Future of User Research report found that **69% of researchers now use AI in at least some projects, up 19 points in a year** ([Maze](https://maze.co/blog/future-user-research-2026/)), largely to get through that backlog.\n\nThis is why many teams now pair a recruiting source with an AI interviewer instead of a human moderator:\n\n![Process showing four steps to pair a recruiting panel with AI interviews: recruit your sample, send an interview link, let the AI moderate, read the live report.](/visuals/recruiting-panel-plus-ai-interview-workflow/recruiting-panel-plus-ai-interview-workflow.png \"Keep your recruiting source and hand the moderation and analysis to an AI interviewer.\")\n\n1. **Recruit your sample.** Use Prolific, Respondent, User Interviews or another panel.\n2. **Send an interview link.** Share a Koji link, or import participants by CSV.\n3. **Let the AI moderate.** Adaptive follow-ups, plus a quality gate on each conversation.\n4. **Read the live report.** Themes, quotes and charts, ready to share the same day.\n\n---\n\n## Which one should you choose?\n\n**Choose UserTesting if:**\n\n- Your main question is about usability: can people find, complete or understand something?\n- You need video of real behaviour for stakeholders.\n- You run tests every week and have a budget for an annual contract.\n\n**Choose User Interviews if:**\n\n- You already have a moderation and analysis process and just need the right people.\n- You recruit B2B roles that consumer panels struggle to reach.\n- You want a published price per session rather than a negotiated contract.\n\n**Choose both** (now easier under one owner) if you run usability tests and recruit hard-to-find interviewees for moderated work.\n\n**Choose something else** if the bottleneck is not finding people but having the conversations and analysing them.\n\n---\n\n## The AI interview alternative: Koji\n\n[Koji](https://www.koji.so) is an AI-native research platform that runs the interviews and the analysis for you. You type what you want to learn, an assistant drafts the brief, and an AI interviewer holds each conversation by voice or text in 30+ languages, asking follow-up questions as it goes.\n\n- **Published, per-interview prices.** A qualified text interview uses 1 credit and a voice interview uses 3. Credits cost €1. Start with 10 free credits and no card, buy credit packs as you go, or choose Insights at €29 a month (29 credits) or Interviews at €79 a month (79 credits).\n- **You only pay for usable interviews.** Every interview gets a [quality score from 1 to 5](/docs/understanding-quality-scores), and only those scoring 3 or more use credits or enter the report.\n- **Recruit from Koji's panel.** On a paid plan, [describe your audience](/docs/panel-recruitment), add screening criteria and see the cost per interview before you launch. Credits are held at launch, unused credits come back at the end, and fraudulent completes you report are refunded once the report is accepted.\n\n![Koji panel recruitment wizard with audience targeting by age and employment and a supply feasibility estimate](/email/shots/shot-panel-recruit.jpg \"Describe who you want to hear from and Koji shows whether the panel can supply them.\")\n\n- **Or use your own people.** Share one link, or [import a CSV](/docs/importing-participants-csv) so each customer gets a personal invite link (Interviews and Enterprise plans). [Screening questions](/docs/adding-screening-questions-to-your-study) work the same whichever way people arrive.\n- **Analysis included.** Reports build as interviews arrive, with themes, findings, recommended actions and personas, and you can [trace every finding to the quotes behind it](/docs/tracing-report-findings-to-evidence).\n\nWhere UserTesting and User Interviews are ahead: Koji does not record video of people using your product, so for watching someone struggle with a prototype, UserTesting is the better tool. And if you need a human researcher to run a sensitive live conversation, User Interviews will fill that calendar.\n\n---\n\n## Try it on your next study\n\nBefore you sign an annual contract or prepay a block of sessions, run one study the new way. Write your research question in Koji, [preview the interview](/docs/previewing-your-interview) for free, and send it to ten customers or a small panel sample. You will have transcripts and a report the same day, and a clear sense of which parts of your research still need a human in the room.\n\n[Start free with Koji](https://www.koji.so). 10 credits, no card.\n\nMore reading: [Koji vs UserTesting](/blog/koji-vs-usertesting-2026), [Koji vs User Interviews](/blog/koji-vs-userinterviews-2026), [UserTesting alternatives](/blog/usertesting-alternatives-2026) and [User Interviews alternatives](/blog/userinterviews-alternatives-2026).\n\n---\n\n*Sources: UserTesting and User Interviews acquisition announcement (Business Wire, January 2026); UserTesting Advanced Targeting announcement (August 2026); Auros rebrand announcement (Business Wire, September 2026); Vendr UserTesting pricing data; User Interviews pricing page; Lyssna, Research Synthesis Report 2025; Maze, Future of User Research Report 2026. Prices change often; check each vendor's current terms.*\n","category":"Comparisons","lastModified":"2026-10-29T13:18:33.553759+00:00","metaTitle":"UserTesting vs User Interviews 2026: Now Both Owned by Auros","metaDescription":"UserTesting runs and records tests on annual contracts; User Interviews recruits participants from $49 a session. Both now belong to Auros. See which job each does, and an AI option.","keywords":["usertesting vs user interviews","user interviews vs usertesting","usertesting user interviews acquisition","auros usertesting","participant recruitment platform","usability testing platform"],"aiSummary":"UserTesting and User Interviews share an owner since UserTesting acquired User Interviews on January 7, 2026; the company rebranded as Auros in September 2026, keeping both as product brands. UserTesting is a research platform for recorded unmoderated tests and live moderated sessions, sold on annual contracts with a Vendr-reported median of $41,200 a year. User Interviews is a recruiting marketplace that finds, screens, schedules and pays participants, from $49 per completed consumer session and $98 for B2B, with incentives extra. Neither runs and analyses interviews for you. Koji is an AI interview alternative with published per-interview pricing, panel recruitment on paid plans and automatic reports, but no video recording.","aiKeywords":["UserTesting","User Interviews","Auros","UserTesting vs User Interviews","participant recruitment","usability testing","Koji"],"aiContentType":"comparison","faqItems":[{"answer":"Yes. UserTesting announced it had acquired User Interviews on January 7, 2026. In September 2026 the company rebranded as Auros, and UserTesting and User Interviews continue as product brands under it.","question":"Is User Interviews owned by UserTesting?"},{"answer":"UserTesting is a research platform where participants take your tests and you watch video of them using your product, unmoderated or live. User Interviews is a recruiting marketplace: it finds, screens, schedules and pays participants, and you run the sessions in your own tools.","question":"What is the difference between UserTesting and User Interviews?"},{"answer":"UserTesting has no public prices; Vendr reports a median annual contract of $41,200 across 359 purchases. User Interviews lists $49 per completed consumer session and $98 for B2B on Pay As You Go, plus participant incentives.","question":"How much do UserTesting and User Interviews cost?"},{"answer":"Yes. In August 2026 UserTesting launched Advanced Targeting, which brings the User Interviews participant network into UserTesting, covering more than 7 million participants across 34 countries.","question":"Can I use User Interviews participants inside UserTesting?"},{"answer":"Koji runs AI-moderated voice and text interviews and writes the report for you. A qualified text interview costs 1 credit and a voice interview 3, at €1 per credit, with 10 free credits to start. Koji does not record video, so UserTesting remains the better choice for watching usability sessions.","question":"Is there an alternative that runs the interviews for me?"}],"relatedTopics":["participant recruitment","usability testing","AI-moderated interviews","research platform pricing","research tool consolidation"]},{"type":"blog","id":"fae8485a-4652-43b2-8c3c-294d3d65df7a","slug":"trymata-alternatives-2026","title":"8 Best Trymata Alternatives in 2026 (After the QuestionPro Merger)","url":"https://www.koji.so/blog/trymata-alternatives-2026","summary":"Trymata (formerly TryMyUI) announced a merger with sister company QuestionPro on February 3, 2026, moving its usability testing and 25,000-tester panel into QuestionPro UX, with post-merger pricing quoted by sales. The best Trymata alternatives in 2026 are Koji (AI-moderated voice and text interviews in 30+ languages with built-in panel recruitment, from free with 10 credits, Insights €29/mo, Interviews €79/mo), Maze (prototype testing, Starter about $99/mo), Lyssna (design tests, Growth $165/mo annual), Userbrain (recorded think-aloud tests), Loop11 (published pricing, $199 and $399/mo), UXtweak (free tier, Business about €92/mo), UserTesting (enterprise video, median about $40,000/yr per Vendr) and QuestionPro UX itself. Pick a task-based tool for screen recordings and Koji for the reasons behind user behaviour.","content":"Trymata, the usability testing platform that started life as TryMyUI, is being folded into QuestionPro. The two sister companies announced their merger on February 3, 2026, and new usability work is moving under the QuestionPro UX brand. If you are a Trymata customer deciding whether to follow it there, or a team that was about to buy Trymata, the realistic alternatives in 2026 are **Koji, Maze, Lyssna, Userbrain, Loop11, UXtweak, UserTesting and QuestionPro UX itself**.\n\nThe short version: if you used Trymata to watch people complete tasks on your site, Loop11, Userbrain, Maze and Lyssna are the closest replacements. If you used it mostly to hear _why_ people struggle, an AI-moderated interview platform like Koji gives you more of that for less money, in voice or text and in 30+ languages.\n\n## What changed at Trymata\n\nAccording to the announcement covered by [Daily Research News Online](https://www.mrweb.com/drno/news39372.htm), QuestionPro reached an agreement to merge with Trymata, its long-time sister company. Trymata had been owned by Survey Analytics, a QuestionPro division, for about a decade. Three details matter to buyers:\n\n- **The panel moves.** Trymata's panel of roughly **25,000 usability testers** is being integrated into QuestionPro's audience offering.\n- **Contracts are honoured.** QuestionPro said it would honour existing Trymata contracts and that tests would work the same way in the near term.\n- **Pricing was reset.** At announcement, QuestionPro said plans and prices for the combined product were \"not yet\" decided and would arrive with a beta in March 2026. We could not find a public QuestionPro UX price list as of October 2026, so expect a sales conversation.\n\nFor reference, Trymata's last published pricing was annual only. Its own FAQ described a Team plan at about $4,000 a year with 10 tester credits a month, and third-party listings put Enterprise between roughly $1,667 and $2,000 a month. None of that is guaranteed to survive the merger.\n\n## Why teams look for Trymata alternatives now\n\nA merger is a natural moment to re-check what you actually use. Most teams that bought Trymata did so for one of three jobs:\n\n1. **Task-based usability tests**: recorded sessions of people trying to complete a flow, with think-aloud audio.\n2. **Quick qualitative feedback**: first impressions, confusion points and the reasons behind them.\n3. **Access to a tester panel**: so you did not have to recruit anyone yourself.\n\nThese jobs are now sold separately by different kinds of tools. The question to answer before you migrate is which of the three you need most, because no single tool is the cheapest at all of them.\n\n## The 8 best Trymata alternatives in 2026\n\n### 1. Koji: best for the \"why\" behind usability problems\n\nKoji is an AI-native research platform. Instead of recording a screen, an AI interviewer holds a real conversation with each participant by voice or text, asks unscripted follow-up questions when an answer is vague, and themes every conversation into a single report you can share with a public link. You can mix open questions with six structured question types (open-ended, scale, single choice, multiple choice, ranking and yes/no), so one study gives you both quotes and numbers.\n\nWhat makes it a strong Trymata replacement:\n\n- **Recruitment is built in.** Describe the audience on the Panel recruit tab and Koji finds participants across 130+ countries, showing the cost before you launch. Credits are set aside at launch and anything unused goes back to your balance. Panel recruitment needs a paid plan. See [panel recruitment](/docs/panel-recruitment).\n- **Screening before the interview.** Add [screening questions](/docs/research-screener-questions), mark which answers qualify or rule someone out, and anyone who does not fit is turned away whatever link they came in by.\n- **30+ languages.** Each participant picks their language on the welcome screen, and every answer lands in one report.\n- **Free previews.** Take your own interview by text or voice before launch. Previews never use credits.\n\n**Pricing:** Start free with 10 credits and no card. Insights is €29 a month with 29 credits; Interviews is €79 a month with 79 credits and 3 seats. A text interview costs 1 credit and a voice interview 3 credits, extra credits are €1, and low-quality conversations are never charged.\n\n**Where it falls short:** Koji does not record screens or measure task success. If your stakeholders need a video of someone failing a checkout, pair Koji with one of the task-based tools below, or read our guide to [AI usability testing](/docs/ai-usability-testing-guide) for how teams combine the two.\n\n### 2. Maze: best for prototype testing at scale\n\nMaze is built around unmoderated tests of Figma prototypes and live sites, with click paths, misclick rates and success metrics. Third-party reviews put its Starter plan at about **$99 a month**, while Team and Organization pricing is quoted by sales; one purchase dataset cited by Usercall shows contracts averaging around $29,000 a year. Maze is quantitative first. You learn where people click, less about why.\n\n### 3. Lyssna: best for fast design tests\n\nLyssna (formerly UsabilityHub) handles five-second tests, first-click tests, preference tests and prototype tests. Its Growth plan is listed at **$165 a month billed annually or $199 monthly**, and panel responses are billed separately by test length, roughly a credit per participant minute. It is quick and well designed, but most studies are short and structured.\n\n### 4. Userbrain: closest to the Trymata workflow\n\nUserbrain sells exactly what Trymata was known for: recorded remote sessions of real testers thinking aloud. Public plan names and prices have changed recently and sources disagree, so check its pricing page; third-party listings range from about $49 to €299 a month depending on plan, with panel testers either included per month or bought per session.\n\n### 5. Loop11: best for published, all-in-one pricing\n\nLoop11 publishes its prices, which is rare in this category. Its Rapid Insights plan is **$199 a month** (or $179 a month billed annually) with 5 seats, and Pro is **$399 a month** with heatmaps, clickstreams and path analysis. It supports moderated and unmoderated testing. Recruitment and moderation services sit outside the subscription.\n\n### 6. UXtweak: best free tier\n\nUXtweak bundles session recording, card sorting, tree testing and prototype testing. Its free plan includes 15 responses a month and one active study, and its Business plan is listed at about **€92 a month billed annually**. A good fit for teams that want many methods on a small budget.\n\n### 7. UserTesting: best for enterprise video evidence\n\nUserTesting remains the largest name in recorded usability research, with a large vetted panel and strong highlight-reel tooling. Prices are quote-only. Vendr's buyer data puts the [median annual contract at about $40,000](https://www.vendr.com/marketplace/usertesting), with deals from roughly $12,000 to over $110,000. Worth it if you need enterprise governance and a lot of video; expensive if you do not.\n\n### 8. QuestionPro UX: best if you want to stay\n\nIf your Trymata contract is working, the path of least effort is to follow it into QuestionPro UX, especially if you already use QuestionPro surveys. The combined product pairs Trymata's usability tools with QuestionPro's survey and audience features. The trade-off is uncertainty: pricing and packaging are still being set, and migrations mid-merger can be bumpy.\n\n## How to choose\n\n| If you mostly need...                                     | Pick                             |\n| --------------------------------------------------------- | -------------------------------- |\n| Recorded task sessions with think-aloud                   | Userbrain, Loop11 or UserTesting |\n| Prototype click metrics                                   | Maze or Lyssna                   |\n| Many methods on a small budget                            | UXtweak                          |\n| The reasons behind behaviour, at scale, in many languages | Koji                             |\n| Minimal change from today                                 | QuestionPro UX                   |\n\nTwo pieces of evidence help frame the decision. First, Nielsen and Landauer's classic model suggests that **five users uncover about 85% of the usability problems that affect at least 31% of users** on a given task, as [MeasuringU explains](https://measuringu.com/five-user-85/). Task testing hits diminishing returns quickly, so you rarely need a big recorded-session budget. Second, the bottleneck has moved to understanding. In Qualtrics' 2025 market research trends study, **89% of researchers** said they were already using AI tools regularly or experimentally ([Research Live](https://www.research-live.com/article/news/researchers-predict-synthetic-responses-will-dominate-in-coming-years/id/5131923)), and ESOMAR's latest industry figures show research software growing **11.5%**, faster than any other part of the insights industry ([ResearchWorld](https://researchworld.com/articles/inside-the-153bn-insights-industry)).\n\nThe practical setup many teams land on: five recorded task sessions in a cheap usability tool to find _where_ the flow breaks, then thirty or more AI-moderated interviews in Koji to understand _why_, across segments and markets, without scheduling a single call. Our guide to [unmoderated usability testing](/docs/unmoderated-usability-testing-guide) covers the first half; [voice vs text interviews](/docs/voice-vs-text-interviews) helps with the second.\n\n## Migrating off Trymata\n\n- **Export your recordings and notes first.** Do it before your contract renews or the platform changes.\n- **Rebuild your screener.** Most tools let you qualify or disqualify on answers. In Koji, screening questions run before every interview.\n- **Re-run one benchmark study** in the new tool so you have a baseline that does not depend on the old panel.\n- **Decide who owns recruitment.** If you relied on Trymata's testers, check panel reach and price per participant in the new tool before committing.\n\n## Try Koji as your Trymata replacement\n\nIf what you valued in Trymata was hearing real people explain what confused them, Koji does that conversation for you, by voice or text, with follow-up questions, and turns it into a shareable report the same day. Start free with 10 credits and no card at [koji.so](https://www.koji.so), and recruit from the panel when you are ready to scale.\n","category":"Comparisons","lastModified":"2026-10-29T13:18:33.282228+00:00","metaTitle":"8 Best Trymata Alternatives (2026) After the QuestionPro Merger","metaDescription":"Trymata is merging into QuestionPro UX. Compare Koji, Maze, Lyssna, Userbrain, Loop11, UXtweak and UserTesting on pricing, panels and methods.","keywords":["trymata alternatives","trymata competitors","trymata pricing","trymata review 2026","questionpro ux","usability testing tools 2026","trymata vs koji"],"aiSummary":"Trymata (formerly TryMyUI) announced a merger with sister company QuestionPro on February 3, 2026, moving its usability testing and 25,000-tester panel into QuestionPro UX, with post-merger pricing quoted by sales. The best Trymata alternatives in 2026 are Koji (AI-moderated voice and text interviews in 30+ languages with built-in panel recruitment, from free with 10 credits, Insights €29/mo, Interviews €79/mo), Maze (prototype testing, Starter about $99/mo), Lyssna (design tests, Growth $165/mo annual), Userbrain (recorded think-aloud tests), Loop11 (published pricing, $199 and $399/mo), UXtweak (free tier, Business about €92/mo), UserTesting (enterprise video, median about $40,000/yr per Vendr) and QuestionPro UX itself. Pick a task-based tool for screen recordings and Koji for the reasons behind user behaviour.","aiKeywords":["trymata alternatives","questionpro ux","usability testing tools","ai moderated interviews","trymata pricing"],"aiContentType":"comparison","faqItems":[{"answer":"Trymata, formerly TryMyUI, announced a merger with its long-time sister company QuestionPro on February 3, 2026. Its usability testing tools and roughly 25,000-tester panel are moving into QuestionPro UX, and QuestionPro said it would honour existing Trymata contracts.","question":"What happened to Trymata?"},{"answer":"There is no public post-merger price list. Trymata's last published pricing was annual, with a Team plan at about $4,000 a year including 10 tester credits a month, and Enterprise listed by third parties at roughly $1,667 to $2,000 a month. Expect to request a QuestionPro UX quote.","question":"How much does Trymata cost in 2026?"},{"answer":"For recorded think-aloud usability sessions, Userbrain and Loop11 are the closest matches, with UserTesting at the enterprise end. For prototype click metrics, Maze and Lyssna. For understanding why users struggle, Koji runs AI-moderated voice and text interviews with follow-up questions and a themed report.","question":"What is the closest alternative to Trymata?"},{"answer":"Koji replaces the qualitative half of Trymata: hearing people explain what confused them, at scale and in 30+ languages, with built-in panel recruitment on paid plans. It does not record screens or measure task success, so teams that need video often pair Koji with a lighter task-based tool.","question":"Can Koji replace Trymata?"},{"answer":"Nielsen and Landauer's model suggests about five users uncover roughly 85% of the problems that affect at least 31% of users on a given task. Rare problems, complex flows or several user groups need more participants.","question":"How many users do I need for a usability test?"}],"relatedTopics":["usability testing tools","unmoderated usability testing","ai moderated interviews","trymata pricing"]},{"type":"blog","id":"4c82377e-f488-433a-92f6-b18ef554c179","slug":"toluna-alternatives-2026","title":"Toluna Alternatives (2026): 7 Options for Panels, Surveys and Interviews","url":"https://www.koji.so/blog/toluna-alternatives-2026","summary":"Toluna Start combines a consumer panel of more than 81 million members in 70+ countries with a survey builder, QProbe AI follow-ups and synthetic personas, with pricing by quote only. Alternatives by need: Cint Exchange for programmatic sample from 800+ suppliers in 130+ countries, Dynata for a 70 million+ first-party panel with deep profiling, Pollfish for DIY surveys from $0.95 per complete, Attest for an annual credit subscription with a flat audience fee across 59 countries, Suzy for enterprise brand testing, and Prolific for fast samples with a 42.8% corporate fee. Koji recruits panel participants inside a study, with the cost per interview shown before launch, and runs AI-moderated text and voice interviews in 30+ languages with a quality gate and an automatic report; plans start free with 10 credits.","content":"**Short answer:** Toluna Start pairs a large consumer panel with a do-it-yourself survey suite, sold through sales on quoted terms. The best alternative depends on which half you want to replace. For raw sample at the lowest unit cost, buy through the Cint Exchange or Pollfish. For a first-party panel with deep profiling, Dynata. For an annual research subscription with a flat audience fee, Attest or Suzy. For fast academic-style samples, Prolific. And if a panel survey keeps giving you the _what_ without the _why_, [Koji](https://www.koji.so) recruits from a panel inside your study and runs AI-moderated interviews with each person, so the report explains the numbers.\n\n## Why teams are comparing Toluna right now\n\nToluna has been around for 25 years, and its founder has just handed over the CEO role. The company [appointed Amy Fenton as chief executive, effective October 15, 2026](https://www.itweb.co.za/article/toluna-appoints-amy-fenton-as-chief-executive-officer/Pero37Z3V4GMQb6m), succeeding founder Frédéric Charles Petit, who stays on the board. A leadership change does not change your contract. But it is a natural moment for an insights team to check whether the platform still fits how they work, especially with renewals coming up.\n\n![An insights team talking around a table with laptops, the kind of group that checks whether a research platform still fits before renewal](https://images.unsplash.com/photo-1563461660947-507ef49e9c47?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHwxMHx8aW5zaWdodHMlMjB0ZWFtJTIwbWVldGluZyUyMGxhcHRvcHN8ZW58MXwwfHx8MTc5MzA4MzgxMnww&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Product School](https://unsplash.com/@productschool?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\nThe bigger reason is that the job itself is shifting. In Qualtrics' [2026 Market Research Trends report](https://www.qualtrics.com/articles/strategy-research/market-research-trends/), based on more than 3,000 researchers, 95% said they now use AI tools regularly or are experimenting with them. Toluna has answered with synthetic respondents. Other teams would rather use AI to talk to more real people, faster. Which camp you are in decides which alternative makes sense.\n\nThere is also the data quality question that every panel buyer now faces. In 2025, Dartmouth's Sean Westwood showed in PNAS that an AI agent [passed 99.8% of standard survey attention checks](https://www.404media.co/a-researcher-made-an-ai-that-completely-breaks-the-online-surveys-scientists-rely-on/), a study that later [won a PNAS prize](https://fas.dartmouth.edu/news/2026/04/sean-westwood-wins-pnas-prize-ai-polling-study). Tick-box screening no longer proves a human is answering. That puts pressure on every platform below to show how it keeps bots out.\n\n## What Toluna offers\n\nToluna Start is a self-service and managed platform built around Toluna's own consumer community. Its [August 2026 panel book](https://tolunacorporate.com/wp-content/uploads/2026/08/Global-Community-Panel-Book-2026_v2_20260825.pdf) lists more than 81 million members across more than 70 countries, with 350+ profiling points for targeting. On top of the panel sit a drag-and-drop survey builder, automated targeting, dashboards and [QProbe](https://tolunacorporate.com/ai-and-innovation/qprobe/), an AI feature that spots thin open-text answers and asks a follow-up. Its HarmonAIze suite adds synthetic personas, which Toluna says are built from anonymised data from its US panel.\n\nToluna does not publish prices. Both [G2](https://g2.com/products/toluna-start/pricing) and Capterra say you need to contact Toluna for a quote if you need respondents.\n\nTeams usually look elsewhere for one of four reasons: they want published, per-study pricing; they want a different or broader panel; they want qualitative depth rather than survey grids; or they are not comfortable replacing real respondents with synthetic ones.\n\n## Toluna alternatives at a glance\n\nHere is how the seven alternatives line up, by what each is best at.\n\n![Table comparing seven Toluna alternatives, Koji, Cint, Dynata, Pollfish, Attest, Suzy and Prolific, by what each is best for and what you get](/visuals/toluna-alternatives-comparison-table/toluna-alternatives-comparison-table.png \"Pick by the job you need done, not by panel size alone\")\n\n## 1. Koji: panel recruits plus AI-moderated interviews\n\n[Koji](https://www.koji.so) is an AI-native research platform. Instead of a survey, each participant has a conversation with an AI interviewer, by text or voice, that asks your questions and follows up on what they say. Structured questions (scales, single and multiple choice, ranking) sit inside the same conversation, so you get charts and the reasons behind them in one pass.\n\nSince September 2026, Koji can supply the people too. On a study's Panel recruit tab you describe the audience in your own words, refine it with targeting criteria, and see the cost per interview in credits before you launch. Koji recruits through the Cint Exchange. A few details matter if you are coming from Toluna:\n\n- **Soft launch.** Field the first batch, around 10% of your goal, then pause and check the answers before releasing the rest.\n- **Credits held, not spent.** Credits for the whole goal are set aside at launch and anything unused comes back when the recruitment ends.\n- **Quality gate.** Every interview gets a quality score from 1 to 5, and only interviews scoring 3 or more use credits or enter the report. Screened-out and quality-rejected panel interviews don't count toward your participant limit. See [how quality scores work](/docs/understanding-quality-scores).\n- **30+ languages.** When a study runs in several languages, every question, answer option and scale label is translated at publish, and you can [choose the language the report is written in](/docs/multilingual-research-guide).\n\n![Koji panel recruitment wizard with audience targeting by age and employment and a supply feasibility estimate](/email/shots/shot-panel-recruit.jpg \"Describe the audience in plain words, then check feasibility before you launch\")\n\nYou can also skip the panel and interview your own customers by link, embed, CSV import or phone, and pay them through Koji's [built-in incentives](/docs/research-participant-incentives), from €10 to €500 per person on paid plans.\n\n**Pricing:** published. Free starts with 10 credits. Insights is €29 a month with 29 credits, and Interviews is €79 a month with 79 credits. A text interview costs 1 credit and a voice interview 3; panel recruits add the panel price per interview, shown before launch. Launching a panel recruitment needs a paid plan, and voice interviews are on Interviews and Enterprise.\n\n**Best for:** teams who need to know _why_ consumers choose, switch or stay, without a separate qual project or an annual panel contract.\n\n**Watch out for:** Koji is built for conversations, not for 40-question tracking grids sent to thousands of people. If you need a national brand tracker, keep a quant tool for that.\n\n## 2. Cint: the sample exchange\n\n[Cint](https://www.cint.com/how-it-works/) runs the Cint Exchange, a programmatic marketplace that connects buyers to more than 800 sample suppliers, with access to respondents in over 130 countries. It is the supply layer much of the industry buys from. You bring the survey (from your own tool or a platform plugged into the exchange) and pay per completed interview.\n\n**Pricing:** per complete, quoted by audience, length and incidence. Not published as a rate card.\n\n**Best for:** teams with their own survey tool who want reach across many countries and suppliers.\n\n**Watch out for:** Cint gives you respondents, not a study. Questionnaire design, analysis and reporting stay with you.\n\n## 3. Dynata: first-party panel at scale\n\n[Dynata](https://www.dynata.com/data-surveys/global-panel/) runs one of the largest first-party panels, citing 70 million+ panelists and 125 million+ completed surveys a year. Its [panel data page](https://www.dynata.com/dynata-data) claims an average of 647 data points per respondent across 82 countries, which helps with hard-to-reach targeting. Managed services are available if you want Dynata to run the project.\n\n**Pricing:** quote-based.\n\n**Best for:** large quantitative studies and B2B or niche consumer targets where profiling depth matters.\n\n**Watch out for:** like Toluna, it is sales-led, and the analysis layer is light unless you buy services.\n\n## 4. Pollfish: cheapest DIY surveys\n\n[Pollfish](https://ws.pollfish.com/resources/blog/uncategorised/evolving-pollfish-pricing-to-better-support-your-research-needs/) is a self-serve survey platform owned by Prodege. Its DIY tier starts from $0.95 per complete, with full-service projects from $5,000 and enterprise from $10,000 a year. Pricing changed on July 15, 2026: short surveys with broad targeting got cheaper, long surveys with narrow audiences may cost more.\n\n**Best for:** quick, short consumer polls on a small budget.\n\n**Watch out for:** short and broad is where it shines. Detailed screening and open-ended depth are not its strengths.\n\n## 5. Attest: annual subscription, flat audience fee\n\n[Attest](https://www.askattest.com/pricing) sells consumer research as an annual subscription measured in credits, where one credit is one answer from one person. It says its multi-panel audience covers 150 million+ people in 59 countries, at one flat audience cost wherever they are. Seats are unlimited and a research expert is included. Attest does not offer pay-as-you-go.\n\n**Best for:** brand and marketing teams who run steady survey volume all year and want predictable cost.\n\n**Watch out for:** you commit for a year up front, and the output is survey data.\n\n## 6. Suzy: agile insights for enterprise brands\n\n[Suzy](https://www.suzy.com) is an enterprise consumer insights platform with access to 70+ panels in 130+ markets, bot detection it calls Biotic, and Suzy Speaks, its AI-moderated conversational product. It does not publish pricing. Third-party estimates based on Vendr data put typical contracts around $88,000 a year, so treat that as a ballpark, not a quote.\n\n**Best for:** large CPG and consumer brands that test concepts and claims every week.\n\n**Watch out for:** sales-led and priced for enterprise budgets.\n\n## 7. Prolific: fast samples, transparent fees\n\n[Prolific](https://researcher-help.prolific.com/en/articles/445239-what-is-your-pricing) is popular with academics and product teams for quick, well-vetted samples. You set the participant reward and Prolific adds a platform fee on top: 42.8% for companies, 33.3% for academic and non-profit customers. It says it has a pool of more than a million people, with 200,000+ active.\n\n**Best for:** behavioural studies, quick validation and unmoderated tasks with general audiences.\n\n**Watch out for:** it is recruitment only. You need your own survey, test or interview tool.\n\n## Why a panel survey alone often misses the point\n\nMost panel platforms, Toluna included, are built to field closed questions fast. That is useful, but the reasons behind an answer usually live in open text, and open text is where surveys fall down. Pew Research Center found that [open-ended questions on its online panel average about 18% item nonresponse, against 1% to 2% for closed questions](https://www.pewresearch.org/decoded/2021/10/14/why-do-some-open-ended-survey-questions-result-in-higher-item-nonresponse-rates-than-others/). And the people who do answer often write a few words.\n\n![Comparison of a panel survey and an AI interview with panel recruits, showing how each handles open answers, follow-ups and analysis](/visuals/panel-survey-vs-ai-interview-comparison/panel-survey-vs-ai-interview-comparison.png \"Same panel, different depth: one asks, the other listens and follows up\")\n\nAn AI interviewer changes that. It asks the open question, hears a short answer, and asks what happened next or why that mattered, the way a moderator would. A conversation is also harder for a bot to fake than a grid of radio buttons, and Koji's quality gate keeps low-effort interviews out of the report. You can read more on [when to use an AI interview instead of a survey](/docs/ai-interviews-vs-surveys) and on [spotting fake and low-effort respondents](/docs/survey-fraud-respondent-quality).\n\n## How to choose\n\nAsk four questions before you sign anything:\n\n1. **Do you need sample, or a study?** Cint, Dynata and Prolific supply people. Attest, Suzy, Pollfish and Toluna add a survey tool. Koji adds the interview and the analysis.\n2. **How will you prove respondents are real?** Ask every vendor how they catch bots now that attention checks can be passed by AI.\n3. **Can you see the price before you commit?** Pollfish, Prolific and Koji publish theirs. Toluna, Dynata and Suzy quote.\n4. **Do you need the why?** If your team keeps running a survey and then a separate qual round to explain it, combine them.\n\nIf you are still building your recruiting approach, our guide to [finding research participants](/docs/finding-research-participants) covers panels, your own users and communities.\n\n## Try Koji\n\nKeep a panel for your big quant waves if you need one. For the questions where the reasons matter more than the percentages, open a [free Koji account](https://www.koji.so), describe who you want to hear from, and check the cost per interview before you launch. Run a soft launch, read the first interviews, then release the rest. You'll have a report built from real conversations, with the quotes behind every theme, without an annual contract. The [panel recruitment guide](/docs/panel-recruitment) walks through every step.\n","category":"Comparisons","lastModified":"2026-10-29T13:18:33.015512+00:00","metaTitle":"Toluna Alternatives 2026: 7 Panel and Survey Tools Compared","metaDescription":"Compare 7 Toluna alternatives for consumer research: Koji, Cint, Dynata, Pollfish, Attest, Suzy and Prolific, by panel, pricing model and depth.","keywords":["toluna alternatives","toluna competitors","toluna start alternatives","consumer research panels","survey panel providers","cint vs toluna","dynata vs toluna"],"aiSummary":"Toluna Start combines a consumer panel of more than 81 million members in 70+ countries with a survey builder, QProbe AI follow-ups and synthetic personas, with pricing by quote only. Alternatives by need: Cint Exchange for programmatic sample from 800+ suppliers in 130+ countries, Dynata for a 70 million+ first-party panel with deep profiling, Pollfish for DIY surveys from $0.95 per complete, Attest for an annual credit subscription with a flat audience fee across 59 countries, Suzy for enterprise brand testing, and Prolific for fast samples with a 42.8% corporate fee. Koji recruits panel participants inside a study, with the cost per interview shown before launch, and runs AI-moderated text and voice interviews in 30+ languages with a quality gate and an automatic report; plans start free with 10 credits.","aiKeywords":["Toluna","Koji","Cint","Dynata","Pollfish","Attest","Suzy","Prolific","consumer panel","panel recruitment","AI-moderated interviews"],"aiContentType":"comparison","faqItems":[{"answer":"It depends on what you want to replace. Cint and Dynata supply sample at scale, Pollfish is the cheapest for short DIY surveys, Attest and Suzy sell research subscriptions, and Prolific is quick for general samples. If you need to know why people answer the way they do, Koji recruits from a panel and runs AI-moderated interviews in the same study.","question":"What is the best alternative to Toluna?"},{"answer":"Toluna does not publish prices. G2 and Capterra both say you need to contact Toluna for a quote if you need respondents, so costs depend on audience, country and survey length.","question":"How much does Toluna cost?"},{"answer":"Toluna's August 2026 panel book lists more than 81 million members across more than 70 countries, with 350+ profiling points for targeting. Older Toluna materials quote 79 million.","question":"How big is Toluna's panel?"},{"answer":"Yes. On a study's Panel recruit tab you describe the audience in your own words and see the cost per interview in credits before you launch. Launching needs a paid plan. You can run a soft launch of about 10% of your goal first, and unused credits come back when the recruitment ends.","question":"Can I recruit panel participants in Koji?"},{"answer":"Yes. In a 2025 PNAS study, Dartmouth's Sean Westwood showed an AI agent passing 99.8% of standard attention checks. Ask every panel vendor how they catch bots. A conversation with follow-up questions, plus a quality score on each interview, is harder to fake than a grid of radio buttons.","question":"Are AI survey responses a problem for panel research?"}],"relatedTopics":["Toluna alternatives","Consumer research panels","Panel recruitment","Survey data quality","AI-moderated interviews","Market research platforms"]},{"type":"blog","id":"21c16da9-af5b-4418-867c-2099f49490d5","slug":"strella-pricing-2026","title":"Strella Pricing in 2026: What It Costs and How to Budget","url":"https://www.koji.so/blog/strella-pricing-2026","summary":"Strella does not publish prices; the path to a number is a demo and a custom quote. The most consistent figure in 2026 buyer guides is about $10,000 to $25,000 per study, which works out to roughly $500 to $1,250 per interview for a 20-interview study. An older directory listing of $150 and $400 monthly plans dates from March 2025 and looks out of date. Strella raised a $14 million Series A led by Bessemer in October 2025 and offers AI-moderated voice interviews with webcam and screen recording, human-led sessions and highlight reels. Koji is the self-serve alternative with published prices: 1 credit per qualified text interview and 3 per voice interview at €1 per credit, billing only interviews scoring 3 or more, with 10 free credits at signup. Koji does not record video.","content":"# Strella Pricing in 2026: What It Costs and How to Budget\n\n**Short answer:** Strella does not publish prices. There is no price list on its site, and the way to a number is a demo and a custom quote. The most consistent figure in 2026 buyer guides is **roughly $10,000 to $25,000 per study**. An older listing shows monthly plans at $150 and $400, but it was last updated in March 2025 and does not match the quote-based model buyers describe today. If you want to run AI-moderated interviews this week without a quote, [Koji](https://www.koji.so) publishes its prices: 1 credit per qualified text interview, 3 per voice interview, €1 per credit, and 10 free credits to start.\n\nThis guide covers what is known about Strella's pricing, why the public numbers disagree, what a study is likely to cost per interview, and when a pay per interview tool is the better fit.\n\n---\n\n## Why teams are pricing Strella now\n\nAI-moderated interviews moved from experiment to habit in about two years. In the 2026 State of User Research report from User Interviews, **80% of researchers said they use AI, up 24 points in a year**. Maze's 2026 Future of User Research report puts the figure lower but still rising fast: **69% of researchers use AI in at least some projects, up 19 points**. The worry has grown alongside the use: the same User Interviews report found that **91% of researchers are concerned about hallucinations**, which is why buyers now look closely at how a platform moderates and analyses before they sign.\n\n![A small research team around a table with laptops, weighing which AI interview platform to budget for](https://images.unsplash.com/photo-1760346546771-a81d986459ff?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHw1fHxyZXNlYXJjaCUyMHRlYW0lMjBsYXB0b3BzJTIwbWVldGluZyUyMHRhYmxlfGVufDF8MHx8fDE3OTIxMjk4MTV8MA&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Bluestonex](https://unsplash.com/@bluestonex_apphaus?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\nStrella is one of the platforms riding that wave. It was founded by Lydia Hylton (CEO) and Priya Krishnan (COO), and in October 2025 it raised a **$14 million Series A led by Bessemer Venture Partners**, bringing its total funding to about $18 million. The company said the money would go toward mobile research and a larger engineering team. Reported customers include Amazon and Duolingo.\n\n---\n\n## What Strella sells\n\nStrella is an AI moderator with a research workflow around it:\n\n- **Real-time voice interviews.** The AI moderator holds voice-to-voice conversations and asks follow-up questions, with sessions that can run up to 90 minutes.\n- **Webcam and screen recording**, so participants can react to a prototype or a website while they talk.\n- **Human and AI moderation in one project.** Teams can mix AI-moderated sessions with human-led ones.\n- **Synthesis and highlight reels.** Themes and video clips for stakeholders.\n- **Recruitment.** Comparisons describe recruitment through a User Interviews integration; one directory lists a much larger global panel. We could not confirm which is current, so ask on the demo.\n- **Languages and connectors.** Directories list support for more than 40 languages, and in March 2026 Strella announced an AI connector that brings its research into the tools teams already use.\n\nThat is a strong product for a team that wants video, long sessions and a mix of human and AI moderation.\n\n---\n\n## What Strella actually costs\n\nBecause Strella's site has no price list, every public number comes from a third party. Here is what each type of source says:\n\n| Source type | What it reports | How much to trust it |\n|---|---|---|\n| 2026 buyer guides and comparisons | About $10,000 to $25,000 per study, sold as quote-based plans | The most consistent figure, still an estimate |\n| Software directory (last updated March 2025) | Starter at $150 a month with 16 completes, Team at $400 a month | Likely out of date |\n| Competitor comparisons | Usage-based, no per-seat charge, no price list | Comes from a competitor, so check it |\n| Free trial | One guide mentions a 14-day trial, others find none | Unconfirmed |\n\nRead together, the picture is a quote-based, per-project purchase in the five-figure range. The low monthly plans look like an earlier self-serve offer that Strella no longer promotes.\n\n### What that means per interview\n\nA per-study price only makes sense once you divide it by the interviews you get. Using the reported range, and a typical study of 20 interviews:\n\n- At **$10,000 per study**, each interview costs about **$500**.\n- At **$25,000 per study**, each interview costs about **$1,250**.\n- At **40 interviews**, the same range works out to **$250 to $625** per interview.\n\nThese are illustrations, not quotes. A Strella study may include recruitment, analysis and support that you would otherwise pay for separately, and that is worth a lot when you need panel participants and video.\n\n---\n\n## What sets your Strella quote\n\nStrella has not published its pricing model, so take these as the questions that usually decide a quote for an enterprise AI interview platform:\n\n1. **Number of studies a year.** A per-project price rewards teams that run a few large studies, not many small ones.\n2. **Interviews per study.** More completes means more cost, especially with panel recruitment.\n3. **Audience.** Hard-to-reach B2B roles cost far more to recruit than general consumers.\n4. **Format.** Video with screen sharing and long sessions cost more to field than short voice or text interviews.\n5. **Support.** Whether Strella's team helps with the guide, recruitment and synthesis.\n\nBefore the demo, write down your answer to each. Then ask for the following in writing:\n\n![Checklist of questions to ask before signing an AI interview platform contract, covering billing, quality, recruitment, data and exit terms.](/visuals/ai-interview-platform-contract-questions-checklist/ai-interview-platform-contract-questions-checklist.png \"Ask Strella each of these, and keep the answers in writing.\")\n\n1. What the platform fee includes: seats, studies, languages, support.\n2. The credit or session price for each audience you plan to recruit.\n3. What counts as a billable complete.\n4. The rate for interviewing your own customers.\n5. Whether unused credits expire, and when.\n6. The renewal increase, in writing.\n7. Export of raw transcripts and recordings if you leave.\n\n---\n\n## Who Strella is built for\n\nStrella fits a research team that runs a handful of large, planned studies a year and wants them to look and feel like moderated sessions: webcam, screen sharing, long conversations and video clips for leadership. If your budget is already set at the project level, a per-study quote is easy to approve.\n\nIt fits less well when research is a weekly habit. A product manager who needs 15 conversations before a planning meeting, a founder testing a pricing idea, or a customer success lead who wants to hear from every churned account each month will struggle to justify a five-figure study each time. Lyssna's 2025 synthesis survey found that **60.3% of practitioners say time-consuming manual work is their biggest frustration**, and a slow buying process adds to that before a single interview happens.\n\nThe choice comes down to two buying models:\n\n![Comparison of an enterprise AI interview contract and pay per interview pricing, showing differences in commitment, trial access and who recruits.](/visuals/enterprise-contract-vs-pay-per-interview-comparison/enterprise-contract-vs-pay-per-interview-comparison.png \"Which model fits depends on how often you research and who you talk to.\")\n\n- **Enterprise contract:** an annual or project fee before any interviews, a price quoted after a demo or pilot, recruitment and moderation bundled, best for high and steady volume.\n- **Pay per interview:** no fee until you run interviews, a published price per interview, you bring participants or buy a panel quote, best for small or irregular studies.\n\n---\n\n## The pay per interview alternative\n\nIf a per-study quote is more than your research needs, [Koji](https://www.koji.so) runs AI-moderated interviews with its prices on the website:\n\n- **Published, per-interview pricing.** A qualified text interview uses 1 credit and a voice interview uses 3. Credits cost €1. A 20-interview voice study with your own customers uses about 60 credits, or about €60.\n- **Billing tied to quality.** Each interview gets a [quality score from 1 to 5](/docs/understanding-quality-scores). Only interviews scoring 3 or more use credits or appear in the report.\n- **No sales call.** Sign up, get 10 free credits with no card, and run a real study. Buy credit packs as you go, or choose a monthly plan: Insights at €29 (29 credits) or Interviews at €79 (79 credits). Credits stay valid for 12 months.\n- **Voice, text and phone.** [AI voice interviews](/docs/ai-voice-interviews) and text chat in 30+ languages, plus phone interviews billed by the minute.\n- **Numbers and reasons in one study.** Six question types (open ended, scale, single choice, multiple choice, ranking and yes/no), with AI follow-up probing on the answers.\n- **Recruit your way.** Share a link, send [personal invite links from a CSV](/docs/importing-participants-csv), or on a paid plan use [panel recruitment](/docs/panel-recruitment) and see the cost in credits before you launch.\n- **Try it first, free.** [Preview your interview](/docs/previewing-your-interview) by voice or text as a participant would. Previews never use credits.\n\n![Koji voice interview in progress with an AI orb asking a follow-up question and a microphone button](/email/shots/shot-voice-interview.jpg \"A Koji voice interview: the AI asks a follow-up based on what the participant just said.\")\n\nReports build as interviews arrive, with [thematic analysis](/docs/thematic-analysis-guide), findings backed by quotes, and personas.\n\nWhere Strella is ahead: Koji does not record webcam video or screen shares, and it does not mix human-moderated sessions into a study. If video and long moderated sessions are what your stakeholders need, Strella is the better choice.\n\n---\n\n## Strella pricing vs Koji at a glance\n\n| | Strella | Koji |\n|---|---|---|\n| Published pricing | No, custom quote after a demo | Yes |\n| Reported cost | About $10,000 to $25,000 per study | Free to start with 10 credits |\n| Billing unit | Per study or usage, quote-based | Credits per qualified interview |\n| Interview formats | Voice with webcam and screen recording, plus human-led sessions | Voice, text and phone |\n| Video highlight reels | Yes | No |\n| Recruitment | Partner panel | Your own users, CSV invites, or panel priced per respondent |\n| Try before you buy | Demo | Self-serve, no card |\n\nMore reading: [Koji vs Strella](/blog/koji-vs-strella-2026), [Strella alternatives](/blog/strella-alternatives-2026), [Listen Labs vs Strella](/blog/listen-labs-vs-strella-2026) and [how AI-moderated interviews work](/docs/ai-moderated-interviews).\n\n---\n\n## Start with ten interviews, not a quote\n\nThe fastest way to judge AI-moderated interviews is to run some. Type your research question into Koji, let it draft the brief, preview the interview for free, then send it to ten customers. You will have transcripts, themes and a report before most enterprise demos are booked.\n\n[Start free with Koji](https://www.koji.so). 10 credits, no card, and you only pay for interviews that pass the quality bar.\n\n---\n\n*Sources: Strella Series A announcement and Bessemer Venture Partners (October 2025); third-party buyer guides and software directories for reported pricing; User Interviews, State of User Research 2026; Maze, Future of User Research Report 2026; Lyssna, Research Synthesis Report 2025. Strella figures are estimates. Ask Strella for a current quote.*\n","category":"Comparisons","lastModified":"2026-10-29T13:18:32.723696+00:00","metaTitle":"Strella Pricing 2026: Reported Costs, Quotes and What to Ask","metaDescription":"Strella has no public prices. Buyer guides report about $10K to $25K per study. See the cost per interview, what sets the quote, and a pay per interview option.","keywords":["strella pricing","strella cost","strella ai pricing","strella free trial","strella alternative","ai interview platform pricing"],"aiSummary":"Strella does not publish prices; the path to a number is a demo and a custom quote. The most consistent figure in 2026 buyer guides is about $10,000 to $25,000 per study, which works out to roughly $500 to $1,250 per interview for a 20-interview study. An older directory listing of $150 and $400 monthly plans dates from March 2025 and looks out of date. Strella raised a $14 million Series A led by Bessemer in October 2025 and offers AI-moderated voice interviews with webcam and screen recording, human-led sessions and highlight reels. Koji is the self-serve alternative with published prices: 1 credit per qualified text interview and 3 per voice interview at €1 per credit, billing only interviews scoring 3 or more, with 10 free credits at signup. Koji does not record video.","aiKeywords":["strella","strella pricing","ai moderated interviews","ai interview platform pricing","custom quote","koji"],"aiContentType":"comparison","faqItems":[{"answer":"No. Strella's site has no price list. You get a custom quote after a demo.","question":"Does Strella publish its pricing?"},{"answer":"2026 buyer guides most often put Strella at about $10,000 to $25,000 per study. Strella has not confirmed this publicly, so treat it as an estimate and ask for a written quote.","question":"How much does Strella cost?"},{"answer":"A software directory lists Starter at $150 a month and Team at $400 a month, but that page was last updated in March 2025. Strella does not promote these plans today, so assume they are out of date.","question":"Does Strella have a $150 a month plan?"},{"answer":"Sources disagree. One guide mentions a 14-day trial and others find none. There is no self-serve signup on the public site. Koji gives 10 free credits at signup with no card.","question":"Does Strella offer a free trial?"},{"answer":"At the reported $10,000 to $25,000 per study, a 20-interview study works out to about $500 to $1,250 per interview, and a 40-interview study to about $250 to $625. These are illustrations, not quotes.","question":"What does a Strella interview cost per interview?"},{"answer":"Koji publishes its prices: 1 credit per qualified text interview and 3 per voice interview, at €1 per credit, and only interviews scoring 3 or more on quality are billed. Koji does not record webcam video, so Strella remains the better fit when video matters.","question":"What is a cheaper alternative to Strella?"}],"relatedTopics":["Strella","AI interview platform pricing","AI moderated interviews","research procurement","pay per interview"]},{"type":"blog","id":"698b518a-40f4-45c9-ad2a-61e0b16f4b46","slug":"respondent-pricing-2026","title":"Respondent Pricing in 2026: Session Fees, Incentives and Annual Plans","url":"https://www.koji.so/blog/respondent-pricing-2026","summary":"Respondent (respondent.io) charges a recruiting fee per completed session: $40 for consumer and $80 for B2B participants on pay as you go. Annual plans require $7,500, $20,000, $50,000 or $500,000+ commitments, with consumer rates of about $36, $34, $32 and $28 per session. Incentives are always extra, set by the researcher from $5 to $1,000; Respondent suggests $100 to $175 for remote US professionals. No fee is charged for no-shows, screen-outs or drop-outs. Worked examples: 12 B2B interviews cost about $2,160 ($180 each); 20 consumer interviews about $2,300 ($115 each), before moderation time. Koji is an AI interviewer alternative: no recruiting fee for your own participants, 1 credit per text and 3 per voice interview at EUR 1 per credit, only interviews scoring 3+ of 5 use credits, and panel recruitment on paid plans with the price per completed respondent shown before launch.","content":"Respondent charges a **recruiting fee per completed session: $40 for consumer (B2C) participants and $80 for business (B2B) professionals** on pay-as-you-go, with no commitment. Annual plans start at a $7,500 commitment and bring the per-session rate down, to about $28 at the $500,000+ tier. **Participant incentives are always extra**, and you set them yourself, from $5 to $1,000 per person. You are not charged the fee for no-shows, screen-outs or people who drop out mid-interview.\n\nSo a Respondent session costs the fee plus the incentive. For a B2B interview that is often $80 plus $100 to $175, before anyone on your team spends an hour moderating. This guide breaks down each part of Respondent pricing, works through two real budgets, and shows when a different setup costs less.\n\n## Respondent pricing at a glance\n\n| | Pay as you go | Annual plans |\n|---|---|---|\n| Commitment | None | $7,500, $20,000, $50,000 or $500,000+ a year |\n| Recruiting fee, consumer | $40 per completed session | About $36, $34, $32 and $28 per session by tier |\n| Recruiting fee, B2B | $80 per completed session | Lower per-session rates, confirmed at commitment |\n| Incentives | Extra, set by you ($5 to $1,000) | Extra, set by you |\n| No-shows and screen-outs | Not charged | Not charged |\n| Included | Screeners, scheduling, AI moderator, recordings, synthesis, highlight reels | Same |\n\nSource: [Respondent's pricing page](https://www.respondent.io/pricing) and [help center](https://help.respondent.io/en/articles/7277356-pricing-details-for-researchers), as indexed in October 2026. Respondent notes that \"rates and session counts are estimates; final pricing is confirmed at commitment\", and it does not publish per-tier B2B rates, so get a written quote before you sign an annual plan.\n\n## How each part of the price works\n\n### The recruiting fee\n\nThe fee is what you pay Respondent for finding, screening and scheduling a participant. You pay it only when a session is completed. If someone fails your screener, does not show up or leaves partway through, there is no fee. That is a real advantage over panels that charge per invite or per qualified lead.\n\nThe B2B fee is double the consumer fee because professionals are harder to find. Respondent says its panel holds 4.3 million verified participants, including 3.1 million professionals and 450,000 subject-matter experts, verified by phone, email and LinkedIn.\n\n### Incentives\n\nIncentives are paid on top, and Respondent is clear that they are \"never included in pricing plans or in PAYG pricing\". You set the amount in $5 steps between $5 and $1,000. Respondent's own guidance is to base professional incentives on the person's hourly rate and consider at least double it. Its [example rates](https://help.respondent.io/en/articles/5471087-what-is-the-correct-incentive-amount-for-my-research-project) for a remote US session:\n\n- Security admin or software engineer: $100\n- UX designer or web developer: $125\n- Data analyst, IT consultant or software developer: $150\n- Cybersecurity analyst: $175\n\nIn-person sessions run about double. Respondent's help pages also describe a 5% processing fee ($1 minimum) taken from the participant's payout, though some marketing pages say \"no fees\", so check the current terms.\n\n### Credits and annual plans\n\nOn an annual plan you commit up front and draw sessions down at a lower rate. An older help article says pay-as-you-go credits expire 60 days after purchase, plan credits roll over at renewal, unused prepayments are not refunded and invoicing needs a $5,000 minimum. These rules may have changed with the newer per-session pricing, so ask.\n\n### AI interviews\n\nRespondent now offers an [AI Interviewer](https://www.respondent.io/platform/ai-moderator) that runs AI-moderated interviews with its participants. According to its help center, AI interviews are billed like moderated sessions, only when completed, with the estimated cost held on the project at launch. Incentives still apply.\n\n## Two Respondent budgets, worked out\n\n**12 B2B interviews with software engineers, pay as you go:**\n\n- Recruiting fee: 12 × $80 = $960\n- Incentives: 12 × $100 = $1,200 (Respondent's example rate)\n- Total: **$2,160, or $180 per completed interview**\n\n**20 consumer interviews, pay as you go:**\n\n- Recruiting fee: 20 × $40 = $800\n- Incentives: 20 × $75 = $1,500 (the example in Respondent's [incentive calculator](https://www.respondent.io/resources/incentive-calculator))\n- Total: **$2,300, or $115 per completed interview**\n\nNeither figure includes the time to write the screener, schedule, moderate and analyze, and Respondent's calculator itself advises over-recruiting by 20 to 30%.\n\n![Process for estimating a participant recruiting budget, from counting sessions to adding moderator time](/visuals/participant-recruiting-budget-estimate-process/participant-recruiting-budget-estimate-process.png \"Most budgets miss the last two steps\")\n\n## The cost no one budgets: no-shows and moderation\n\nRespondent does not charge for no-shows, but they still cost you the time slot. MeasuringU's analysis of about 14,000 moderated studies on User Interviews found an [average no-show rate of about 8%](https://measuringu.com/typical-no-show-rate-for-moderated-studies/), with some studies as high as 34%. Incentives help: User Interviews' [incentives report](https://www.userinterviews.com/blog/research-incentives-report) saw no-shows fall from 10% at $60 an hour to 6% at $100 an hour.\n\n![A person waiting alone on a video call at a laptop, the empty time slot a research no-show leaves behind](https://images.unsplash.com/photo-1698891667813-b445f80aace7?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHw5fHxwZXJzb24lMjB3YWl0aW5nJTIwdmlkZW8lMjBjYWxsJTIwbGFwdG9wfGVufDF8MHx8fDE3OTIyMTYyMTN8MA&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Resume Genius](https://unsplash.com/@resumegenius?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\nThen there is the hour of moderation, the prep and the synthesis for each session. That is where research time goes. In User Interviews' [State of User Research 2025](https://www.userinterviews.com/state-of-user-research-report), 54% of researchers said recruiting on their own takes too long, and in [Hubble's 2026 survey](https://www.usehubble.io/blog/state-of-user-research), 36.3% named recruitment as one of the biggest causes of project delays.\n\n## What reviewers say about Respondent\n\nResearchers rate Respondent well: 4.6 out of 5 from more than 500 reviews on [G2](https://www.g2.com/products/respondent/reviews), with praise for ease of use and the quality of B2B participants. The common complaints are that it is costly for small studies and that support can be slow at busy times. Ratings on Trustpilot are lower (about 3.9), but most of those reviews come from participants, about cancelled studies and payment delays, not from researchers.\n\n## When Respondent is the right choice\n\nRespondent is a strong pick when:\n\n- You need specific professionals (engineers, IT buyers, security leads) and cannot reach them through your own customer base.\n- You run moderated sessions and want screening, scheduling and Zoom links handled for you.\n- You only want to pay for sessions that actually happen.\n- Your volume is high enough that an annual plan's lower rate pays off.\n\nIt is less suited to studies where the people you need are already your customers, users or leads. Paying $40 to $80 a session to recruit people you could email is money you do not have to spend.\n\n## A lower-cost route: Koji\n\nKoji approaches the same problem from the other side. It is an AI interviewer: participants take the interview by voice or text from a link, in 30+ languages, and Koji asks the follow-ups and writes the report. That removes the scheduling and moderation, which is where most of the hidden cost sits.\n\n### Interview your own people for a credit or three\n\nIf you already have a list, import it as a CSV and each person gets their own [personalized invite link](/docs/personalized-interview-links) (Interviews and Enterprise plans), or share one link anywhere. There is no recruiting fee. A text interview uses 1 credit and a voice interview 3 credits, at €1 per credit, and only interviews scoring 3 or more out of 5 on Koji's [quality gate](/docs/how-the-quality-gate-works) use credits. You decide whether to offer an incentive; Koji's guide to [research participant incentives](/docs/research-participant-incentives) covers what to pay.\n\n### Recruit from Koji's panel when you need strangers\n\nWhen you need people outside your own base, Koji's [panel recruitment](/docs/panel-recruitment) works on the Insights, Interviews and Enterprise plans. You describe the audience in plain language, add [screening questions](/docs/adding-screening-questions-to-your-study), and the review step shows feasibility and the panel price per completed respondent in credits before you launch. A soft launch fields around 10% of the goal first so you can check the early interviews. Credits are set aside at launch and anything unused goes back to your balance.\n\n![Koji panel recruitment wizard with audience targeting by age and employment and a supply feasibility estimate](/email/shots/shot-panel-recruit.jpg \"Describe who you want in plain words and see feasibility before you launch\")\n\nKoji's panel fits consumer and general work audiences best. For narrow specialists, such as senior security leaders in one industry, a specialist recruiter like Respondent is still the better source. You can recruit them there and send them a Koji link to run the interview.\n\n### Respondent vs Koji on cost\n\n| | Respondent | Koji |\n|---|---|---|\n| Recruiting your own customers | Not needed; you would pay the session fee | No recruiting fee, free personal or shared links |\n| Recruiting strangers | $40 consumer or $80 B2B per session, plus incentive | Panel price per completed respondent, shown before launch, on paid plans |\n| Moderation | You moderate, or use Respondent's AI moderator | Koji's AI interviewer, voice or text |\n| Analysis | Recordings and synthesis | Report with themes, findings, quotes and personas |\n| Plans | Pay as you go or $7,500+ annual commitment | Free with 10 credits, Insights €29/month, Interviews €79/month, Enterprise |\n\n![Checklist of eight questions to ask before choosing a research recruiting setup, from monthly volume to data portability](/visuals/research-recruiting-platform-checklist/research-recruiting-platform-checklist.png \"Answer these before you compare per-session prices\")\n\n## Start with the people you already have\n\nBefore you pay a recruiting fee, check whether the people you need are already in your CRM, your product or your newsletter. [Start a Koji study free](/signup) with 10 credits and no card, share the link with them, and read the first report the same day. When you need people you cannot reach, the panel is one tab away.\n","category":"Comparisons","lastModified":"2026-10-29T13:18:32.447907+00:00","metaTitle":"Respondent Pricing 2026: Fees, Incentives and Plans","metaDescription":"Respondent charges $40 per consumer and $80 per B2B session, plus incentives you set. See annual plan rates, two worked budgets and the costs most teams miss.","keywords":["respondent pricing","respondent.io pricing","respondent cost","respondent fees","respondent incentives","respondent alternative"],"aiSummary":"Respondent (respondent.io) charges a recruiting fee per completed session: $40 for consumer and $80 for B2B participants on pay as you go. Annual plans require $7,500, $20,000, $50,000 or $500,000+ commitments, with consumer rates of about $36, $34, $32 and $28 per session. Incentives are always extra, set by the researcher from $5 to $1,000; Respondent suggests $100 to $175 for remote US professionals. No fee is charged for no-shows, screen-outs or drop-outs. Worked examples: 12 B2B interviews cost about $2,160 ($180 each); 20 consumer interviews about $2,300 ($115 each), before moderation time. Koji is an AI interviewer alternative: no recruiting fee for your own participants, 1 credit per text and 3 per voice interview at EUR 1 per credit, only interviews scoring 3+ of 5 use credits, and panel recruitment on paid plans with the price per completed respondent shown before launch.","aiKeywords":["respondent","respondent.io","participant recruitment","research incentives","b2b recruiting","panel recruitment","koji"],"aiContentType":"comparison","faqItems":[{"answer":"On pay as you go, Respondent charges $40 per completed consumer session and $80 per completed B2B session, plus the incentive you set for each participant. Annual plans start at a $7,500 commitment and lower the per-session rate.","question":"How much does Respondent cost?"},{"answer":"No. Respondent says incentives are never included in its plans or pay-as-you-go pricing. You set them yourself, from $5 to $1,000 per participant, and Respondent suggests at least double a professional's hourly rate.","question":"Are incentives included in Respondent pricing?"},{"answer":"No. Respondent does not charge the recruiting fee if a participant fails the screener, does not show up or drops out mid-interview. You only pay for completed sessions.","question":"Does Respondent charge for no-shows?"},{"answer":"Respondent's examples for remote US sessions range from $100 for software engineers to $175 for cybersecurity analysts, and about double for in-person sessions. Consumer studies are typically lower, and its incentive calculator uses $75 as an example.","question":"What incentive should I offer on Respondent?"},{"answer":"If the people you need are already your customers or users, Koji lets you interview them with no recruiting fee: an AI interviewer runs voice or text interviews from a link for 1 or 3 credits at €1 each. For people outside your base, Koji's panel recruitment on paid plans shows the price per completed respondent before launch.","question":"Is there a cheaper alternative to Respondent?"}],"relatedTopics":["Respondent","participant recruitment pricing","research incentives","B2B recruiting","AI interviews"]},{"type":"blog","id":"a65b81e4-c6fa-470f-8e9c-19026273c05c","slug":"outset-vs-maze-2026","title":"Outset vs Maze (2026): Which AI-Moderated Interview Tool Fits Your Team?","url":"https://www.koji.so/blog/outset-vs-maze-2026","summary":"Comparison of Outset and Maze for AI-moderated interviews in 2026. Outset (founded 2022, $30M Series B December 2025) is built around AI interviews: text, voice and video, screen-shared usability sessions, 40+ languages, synthesis with reports and decks, partner panels such as Prolific and User Interviews, custom plans with no free tier. Maze is a product research platform (prototype tests, surveys, card sorting, tree testing) whose AI moderator launched in beta in June 2025; it has its own panel in 150+ countries and a Free plan, with the AI moderator listed on quoted plans. Koji is a self-serve alternative: voice, text and phone AI interviews in 30+ languages, free previews, a quality gate, public pricing from free (Insights EUR 29/month, Interviews EUR 79/month), but no video or prototype testing.","content":"**Short answer:** Outset is a dedicated AI interview platform: text, voice and video interviews, screen-shared usability sessions, 40+ languages and deep synthesis, sold on custom enterprise plans. Maze is a broad product research platform (prototype tests, surveys, card sorting, tree tests) that added an AI moderator in 2025 as one more study type. Choose **Outset** if AI-moderated interviews are your main research method and you have an enterprise budget. Choose **Maze** if most of your work is prototype and usability testing and you want interviews in the same tool. If you want AI interviews without a sales call, **[Koji](https://www.koji.so)** runs voice, text and phone interviews on public pricing from free.\n\n## Outset vs Maze at a glance\n\n| | Outset | Maze |\n|---|---|---|\n| What it is | A platform built around AI-moderated interviews | A product research platform with an AI moderator added |\n| Interview formats | Text, voice and video | AI-moderated interviews and a voice feedback block |\n| Usability testing | Screen share on prototypes, live sites and apps | Prototype tests, card sorts and tree tests |\n| Participants | Your own, or partners like Prolific and User Interviews | Its own panel in 150+ countries, or your own |\n| Languages | 40+, per Outset | No published list for the AI moderator |\n| Buying model | Custom plans, no free tier | Free tier; AI moderator on quoted plans |\n| Best fit | Research teams that run interviews at volume | Product and design teams that test prototypes |\n\n![Table comparing Outset and Maze on what each is, interview formats, usability testing, participants, languages, buying model and best fit](/visuals/outset-vs-maze-comparison-table/outset-vs-maze-comparison-table.png \"Outset is built around the interview; Maze added one to a testing platform\")\n\n## Why AI moderation is the comparison that matters\n\nTwo years ago, Outset and Maze barely competed. Maze was where designers tested Figma prototypes; Outset was a young AI interview startup. AI moderation changed that.\n\nThe numbers show why every research tool now wants an AI interviewer. In Maze's own 2025 Future of User Research report, **58% of teams said they use AI tools in research, up from 44% in 2024**, and **55% saw demand for research rise** ([Maze](https://maze.co/resources/user-research-report-2025/)). Analysis was the top use, at **74%** of AI users. User Interviews' State of User Research found **80% of researchers now use AI**, up 24 points in a year, while **41% view AI negatively** and **91% worry about hallucinations** ([User Interviews via UserTesting](https://www.usertesting.com/user-interviews-state-of-research-report-2026)).\n\nThat mix of heavy use and real doubt is the right lens for this comparison. You want an AI interviewer that saves time and that you can check. Participants have a say too: in one 2026 study of AI-conducted interviews, **22.7% of participants (15 of 66) said they missed talking to a person** ([arXiv](https://arxiv.org/pdf/2607.14452)). That is a minority, but it is a reason to keep the conversation easy and short.\n\n## Outset: built around the interview\n\nOutset was founded in 2022 by Aaron Cannon and Michael Hess. Everything in the product starts from an AI-moderated conversation.\n\n- **Formats.** Text, voice and video interviews. Participants can share their screen on mobile or desktop, and the AI can show images and video as prompts.\n- **Usability sessions.** Picture-in-picture screen share on Figma prototypes, live websites and mobile apps, with an agent that flags friction such as dead clicks and hesitation.\n- **Languages.** Outset says it runs 40+ languages natively.\n- **Participants.** No panel of its own. Bring your own customers, or recruit through partners such as Prolific and User Interviews.\n- **Synthesis.** Summaries, themes, quotes, highlight reels, AI-written reports and slide decks, and you can ask questions of the interviews directly ([Outset](https://outset.ai/platform/synthesis)).\n- **Backing.** A $17M Series A led by 8VC in June 2025 ([Research Live](https://www.research-live.com/article/news/series-a-funding-for-aimoderated-research-startup-outset/id/5140703)) and a $30M Series B led by Radical Ventures in December 2025, with Microsoft's M12 taking part. Named customers include Nestlé, Microsoft, WeightWatchers and HubSpot.\n\n**Pricing.** No public prices. Plans are custom, with add-ons such as white-labeling and expert project support ([Outset pricing](https://outset.ai/pricing)), and third-party listings say there is no free plan.\n\n**Limits.** G2 does not have enough reviews to give buying insight. Independent reviewers note that AI-generated themes still need a researcher to check them, and recruiting is always a separate cost.\n\n## Maze: an AI moderator inside a testing platform\n\nMaze built its name on unmoderated prototype testing. Today it covers prototype tests, surveys, card sorting, tree testing, concept tests and interview studies. Maze's own comparison pages say its panel reaches 150+ countries.\n\nThe **AI moderator** launched in beta on June 18, 2025, switched on for selected accounts first ([Maze](https://maze.co/blog/ai-moderator/)). It works in three steps: you enter your research goals and the AI drafts an editable discussion guide; it runs the interviews on its own across time zones and asks follow-up questions; then it transcribes each session, tags key moments and groups themes against your goals ([Maze](https://maze.co/features/ai-moderated-interviews)). Maze has since added visual stimulus testing and different discussion styles, and has said screen sharing for prototype and website testing is coming. We could not confirm it has shipped.\n\n**Pricing.** Maze's public pricing is in flux. It has a Free plan and quoted Organization or Enterprise plans. A Starter plan at about $99 a month appeared in many 2026 listings, but pricing trackers report Maze removed it this year, and Maze's help center says it runs pricing experiments, so plans can differ by account. Third-party listings place the AI moderator on Enterprise plans. Check with Maze before you budget.\n\n**Limits.** G2 and Software Advice reviewers mention limited survey logic, basic analytics, some Figma prototypes failing on mobile, and a restrictive free plan with key features on higher tiers.\n\n## Where they differ\n\n### Depth of the interview product\n\nOutset's interviews are the product, so it has more options: video, screen share, stimulus, a richer synthesis layer. Maze's AI moderator is newer and sits next to a dozen other study types. If interviews are 80% of your research, that difference matters. If they are 20%, Maze's breadth may matter more.\n\n### Usability testing\n\nBoth cover it in different ways. Maze has years of unmoderated prototype testing with clicks, paths and heatmaps. Outset adds AI moderation on top of a shared screen, so you hear why someone hesitated as well as where. If you test prototypes every week, Maze's unmoderated speed is hard to beat. If you want to hear the reasoning, Outset's approach fits better.\n\n![A smartphone lying on a white surface, the kind of device people use to try an app prototype in a usability test](https://images.unsplash.com/photo-1581287053822-fd7bf4f4bfec?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHw0fHxkZXNpZ25lciUyMHRlc3RpbmclMjBhcHAlMjBwcm90b3R5cGUlMjBwaG9uZXxlbnwxfDB8fHwxNzkxOTU3MDEyfDA&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [charlesdeluvio](https://unsplash.com/@charlesdeluvio?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\n### Recruiting\n\nMaze has its own panel. Outset relies on partners, which gives you choice but adds a second contract and a second bill.\n\n### Buying\n\nNeither publishes a simple price for AI-moderated interviews. Expect a sales conversation with both.\n\n## A third option: Koji\n\nIf you are comparing Outset and Maze because you want AI interviews, not a whole testing suite or an enterprise contract, look at Koji. It is an AI-native research platform you run yourself:\n\n- **Voice, text and phone interviews** in 30+ languages. Participants choose to talk or type; you can also send a call-back link or have Koji phone a list of numbers. Koji does not do video or screen-share usability tests, so it is not a replacement for Maze's prototype testing.\n- **Open and structured questions together.** AI follow-ups on open questions sit next to scales, choice, ranking and yes/no questions in one interview.\n- **Free previews.** [Take your own interview](/docs/previewing-your-interview) by voice or text before launch; previews never use credits.\n- **A quality gate.** Every interview gets a [quality score from 1 to 5](/docs/understanding-quality-scores), and only interviews scoring 3 or more use credits or enter the report.\n- **Participants your way.** Share a link, import a CSV for personal invite links (Interviews and Enterprise plans), or [recruit from Koji's panel](/docs/panel-recruitment) on paid plans.\n- **Public pricing.** Free with 10 credits, Insights at €29 a month, Interviews at €79 a month, Enterprise on request. A text interview uses 1 credit and a voice interview 3.\n\n![Koji preview completion screen confirming a test interview is saved and does not count as a response](/email/shots/shot-preview-complete.jpg \"Test your own interview before launch: previews are saved with their own analysis and never charged\")\n\nWhichever tool you shortlist, ask each vendor the same questions before you sign:\n\n![Checklist of eight questions to ask before choosing an AI interview platform, from free previews to data storage](/visuals/ai-interview-platform-vendor-checklist/ai-interview-platform-vendor-checklist.png \"Ask every AI interview vendor these eight questions before a demo turns into a contract\")\n\n## Which should you choose?\n\n**Choose Outset if:**\n\n- AI-moderated interviews are your main method and you run them at volume.\n- You need video and screen-shared usability sessions with an AI moderator.\n- You have an enterprise budget and your own participants or a partner panel.\n\n**Choose Maze if:**\n\n- Your team mostly tests prototypes, flows and information architecture.\n- You want AI interviews as one study type in a tool your designers already use.\n- You value a built-in panel over interview depth.\n\n**Choose Koji if:**\n\n- You want AI-moderated voice or text interviews without a sales cycle.\n- You want to mix open conversation with structured questions and pay only for interviews that pass a quality check.\n- You want to start this week, for free.\n\n## Try Koji free\n\nThe fastest way to judge an AI interviewer is to sit through one. [Create a free Koji account](https://www.koji.so), type your research question, and preview the interview yourself. Your 10 signup credits cover ten text interviews or three voice interviews with real participants. For more detail, see [Koji vs Outset](/blog/koji-vs-outset-2026), [Koji vs Maze](/blog/koji-vs-maze-2026), or our guide to [voice vs text interviews](/docs/voice-vs-text-interviews).\n","category":"Comparisons","lastModified":"2026-10-29T13:18:32.1504+00:00","metaTitle":"Outset vs Maze 2026: AI-Moderated Interviews Compared","metaDescription":"Outset vs Maze: interview formats, usability testing, panels, languages and pricing compared, and when a self-serve AI interview tool like Koji fits better.","keywords":["outset vs maze","maze ai moderator","outset ai","maze ai moderated interviews","outset pricing","maze pricing","ai moderated interviews"],"aiSummary":"Comparison of Outset and Maze for AI-moderated interviews in 2026. Outset (founded 2022, $30M Series B December 2025) is built around AI interviews: text, voice and video, screen-shared usability sessions, 40+ languages, synthesis with reports and decks, partner panels such as Prolific and User Interviews, custom plans with no free tier. Maze is a product research platform (prototype tests, surveys, card sorting, tree testing) whose AI moderator launched in beta in June 2025; it has its own panel in 150+ countries and a Free plan, with the AI moderator listed on quoted plans. Koji is a self-serve alternative: voice, text and phone AI interviews in 30+ languages, free previews, a quality gate, public pricing from free (Insights EUR 29/month, Interviews EUR 79/month), but no video or prototype testing.","aiKeywords":["outset","maze","outset vs maze","maze ai moderator","ai moderated interviews","usability testing","prototype testing","ai interview platforms","koji","research pricing"],"aiContentType":"comparison","faqItems":[{"answer":"Outset is built around AI-moderated interviews, with text, voice and video formats and screen-shared usability sessions. Maze is a prototype and usability testing platform that added an AI moderator in 2025 as one study type alongside surveys, card sorts and tree tests.","question":"What is the main difference between Outset and Maze?"},{"answer":"Yes. Maze launched its AI moderator in beta on June 18, 2025. It drafts a discussion guide from your goals, runs interviews with follow-up questions, then transcribes, tags and groups themes. Third-party listings place it on Enterprise plans.","question":"Does Maze have AI-moderated interviews?"},{"answer":"Outset does not publish prices. Plans are custom with optional add-ons, and third-party listings say there is no free plan. Recruiting through partners such as Prolific or User Interviews is billed separately.","question":"How much does Outset cost?"},{"answer":"Koji is self-serve. It runs AI-moderated voice, text and phone interviews in 30+ languages, starts free with 10 credits, and its paid plans are €29 and €79 a month. It does not do video or prototype testing.","question":"Is there a self-serve alternative to Outset and Maze for AI interviews?"},{"answer":"Maze, if you run frequent unmoderated prototype tests with clicks and paths. Outset, if you want an AI moderator to ask people about their reasoning while they share their screen. Koji does not run screen-share usability tests.","question":"Which is better for usability testing?"}],"relatedTopics":["ai moderated interviews","maze alternative","outset alternative","usability testing","research pricing"]},{"type":"blog","id":"3a8ba21c-dbef-4708-9873-38d54f537974","slug":"outset-vs-conveo-2026","title":"Outset vs Conveo (2026): Which AI Interview Platform Fits Your Team?","url":"https://www.koji.so/blog/outset-vs-conveo-2026","summary":"Outset and Conveo are AI-moderated interview platforms aimed at different buyers. Outset suits enterprise UX and insights teams: text, voice and video interviews plus usability tasks with screen sharing, 40+ languages, partner recruitment, and custom plans reported at around $20,000 a year. Conveo suits consumer brands and agencies: video, voice and text interviews with tone and non-verbal analysis, 50+ languages, integrated panels, and prepaid credits billed per interview minute, with an enterprise entry point listed at US$45,000. Both start with a demo. Koji is a self-serve alternative with published per-interview prices and 10 free credits, but no webcam video or screen sharing.","content":"# Outset vs Conveo (2026): Which AI Interview Platform Fits Your Team?\n\n**Short answer:** Choose **Outset** if you are a UX or insights team that wants AI-moderated interviews and usability sessions on one platform, with screen sharing on prototypes, live sites and mobile apps, and you are comfortable with a custom annual contract. Choose **Conveo** if you are a consumer brand or agency testing concepts, ads and packaging, and you want video analysis that reads tone and non-verbal cues, billed by the interview minute. Choose **[Koji](https://www.koji.so)** if voice and text are enough and you want to start this week on published, per-interview prices: 1 credit per qualified text interview, 3 per voice interview, €1 per credit, and 10 free credits.\n\nBoth Outset and Conveo replace the old \"book a moderator, wait three weeks, code the transcripts\" loop with an AI that interviews and analyses for you. They differ on who they are built for, what they analyse, and how you pay.\n\n---\n\n## Why this comparison comes up now\n\nTwo things happened in the last year. AI-moderated interviews went mainstream, and the vendors raised enough money to sell hard.\n\nMaze's 2026 Future of User Research report, based on nearly 500 responses, found that **69% of researchers use AI in at least some projects, up 19 points in a year** ([Maze](https://maze.co/blog/future-user-research-2026/)). A year earlier, **58% of teams used AI tools in research, up from 44% in 2024** ([Maze](https://maze.co/resources/user-research-report-2025/)). Budgets followed. Outset raised a **$30 million Series B led by Radical Ventures** in December 2025, with Microsoft's venture fund M12 taking part, for about $51 million in total. Conveo raised a **$50 million Series A in September 2026** ([Axios](https://axios.com/pro/enterprise-software-deals/2026/09/02/conveo-consumer-insights-dst)).\n\nThe market around them is also consolidating. At the end of September 2026, Salesforce signed an agreement to buy Listen Labs, another close rival, in a deal reported at about $2 billion ([SiliconANGLE](https://siliconangle.com/2026/10/01/salesforce-to-acquire-ai-customer-research-startup-listen-labs-in-reported-2b-deal/)), and User Interviews, one of Outset's recruitment partners, now sits under UserTesting's new parent company, Auros. Buyers who were comparing four or five independent AI interview vendors a year ago are now asking which ones will still look the same in 2027. Outset and Conveo are two of the best-funded independents left.\n\n---\n\n## Outset vs Conveo at a glance\n\n![Table comparing Outset and Conveo on main buyer, interview formats, usability testing, analysis, participants, languages and how you buy.](/visuals/outset-vs-conveo-comparison-table/outset-vs-conveo-comparison-table.png \"Same category, different buyers: UX teams lean Outset, brand and agency teams lean Conveo.\")\n\n| | Outset | Conveo |\n|---|---|---|\n| Main buyer | Enterprise UX and insights teams | Consumer brands and research agencies |\n| Interview formats | Text, voice and video | Video, voice and text |\n| Usability testing | Screen share on prototypes, live sites and mobile apps | Screen interaction analysis as a paid add-on |\n| Analysis | Themes, quotes and highlight reels | Themes and reels, plus tone and non-verbal cues |\n| Participants | Your own users, or partners such as Prolific and User Interviews | Your own list, or integrated panels such as Respondent |\n| Languages | 40+, per Outset | 50+, per Conveo |\n| How you buy | Custom plans after a demo, no free tier | Credits billed per interview minute, after a demo |\n\n---\n\n## What Outset does best\n\nOutset was built around one job: let a research team run many more interviews than its people could moderate. Its AI moderator holds conversations by **text, voice or video**, asks follow-up questions, and can switch into a usability task where the participant shares a screen on a prototype, a live website or a mobile app. Outset says it works in **more than 40 languages**.\n\nThat usability layer is its clearest edge over Conveo. A product designer can ask why someone chose a plan and, in the same session, watch them try to change it. Outset turns the results into summaries, themes, quotes and highlight reels for stakeholders. On security it states SOC 2 Type II, ISO 42001, GDPR and HIPAA, which shortens procurement at large companies.\n\nOutset has no panel of its own. You interview your own users or recruit through partners such as Prolific and User Interviews, which keeps recruitment flexible but adds a second vendor to the bill.\n\n**Pricing:** Outset's pricing page lists no figures. It describes a standard subscription, a plan with added research support, and enterprise terms. Buyer reports put the entry point at **around $20,000 a year**, sometimes described as per seat, with panel recruitment on top. We cover the details in [Outset pricing](/blog/outset-pricing-2026).\n\n---\n\n## What Conveo does best\n\nConveo started from video. Its AI moderator runs **video, voice and text** interviews with dynamic follow-ups, and its analysis goes past the transcript to **tone, facial expression and behaviour**. Each theme gets a highlight reel, and each finding links back to the source clip. Conveo says it supports **more than 50 languages** and works with more than 400 enterprises, including Google, Unilever and AB InBev.\n\nThat makes it strongest where emotion is the data: concept tests, ad tests, packaging and brand work, where a stakeholder wants to see a face react rather than read a quote. Participants can join by CSV invite, QR code or WhatsApp, or come from integrated B2C and B2B panels.\n\n**Pricing:** Conveo bills prepaid credits per **interview minute**, with an uplift for advanced video analysis such as screen interaction or diary studies. Its agency pricing page shows an example of 1,500 minutes at $5.00 a minute, and Software Advice lists an enterprise starting price of US$45,000. We worked through a full study budget in [Conveo pricing](/blog/conveo-pricing-2026).\n\n---\n\n## Where they differ, and why it matters\n\n### Usability testing vs emotional response\n\nIf your question is \"can people complete this task?\", Outset's screen sharing on prototypes and apps gives you behaviour you can watch. If your question is \"how do people feel about this ad?\", Conveo's non-verbal analysis gives you reactions you can show. Both can do some of the other, but each is tuned for one.\n\n### Seats vs minutes\n\nOutset's reported model is a subscription, scoped to your team and support needs. Conveo charges by the minute. A seat-style contract rewards teams that run research all year with the same few people. A per-minute model rewards teams with irregular, project-shaped work, but the bill depends on how long participants talk, which is hard to forecast.\n\n### Who recruits\n\nNeither runs its own panel in the way Listen Labs does. Outset leans on partners such as Prolific and User Interviews; Conveo integrates several panels, including Respondent. Either way, budget recruitment and incentives as separate lines.\n\n### How fast you can start\n\nNeither offers a self-serve signup or a published free trial. Both start with a demo. Outset's own site describes a customer whose focus groups had cost **$80,000 and taken five weeks**, which shows the comparison both vendors sell against: agency research, not other software.\n\n---\n\n## Who should choose which\n\n**Pick Outset if:**\n\n![A designer at a desk with a drawing tablet, a laptop and colour swatches, working on the kind of design a usability study tests](https://images.unsplash.com/photo-1572044162444-ad60f128bdea?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHwyfHxkZXNpZ25lciUyMHRlc3RpbmclMjBhcHAlMjBwcm90b3R5cGUlMjBsYXB0b3B8ZW58MXwwfHx8MTc5MjM4OTAxNHww&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Theme Photos](https://unsplash.com/@themephotos?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\n- Your research team runs usability work alongside interviews and wants one tool for both.\n- You need SOC 2, ISO 42001 or HIPAA paperwork for procurement.\n- Your research runs all year with a steady team, so a subscription is easy to justify.\n\n**Pick Conveo if:**\n\n- You are a consumer brand or agency testing creative, concepts or packaging.\n- Stakeholders want video reels and emotional reads, not only quotes.\n- Your work comes in projects, and a per-minute bill maps cleanly to client budgets.\n\n**Pick neither if** you need answers from your own customers this week, without a procurement cycle. That is where the buying model matters more than the feature list.\n\n---\n\n## The buying model is the real decision\n\nFor many teams, Outset vs Conveo is the wrong first question. The bigger choice is how you want to pay for AI interviews at all:\n\n![Comparison of an enterprise AI interview contract and pay per interview pricing, showing differences in commitment, trial access and who recruits.](/visuals/enterprise-contract-vs-pay-per-interview-comparison/enterprise-contract-vs-pay-per-interview-comparison.png \"Which model fits depends on how often you research and who you talk to.\")\n\n- **Enterprise contract:** an annual platform fee before any interviews, a price quoted after a demo or pilot, recruitment and moderation bundled, best for high and steady research volume.\n- **Pay per interview:** no fee until you run interviews, a published price per interview, you bring participants or buy a panel quote, best for small or irregular studies.\n\nOutset and Conveo both sit on the left. [Lyssna's 2025 synthesis survey](https://www.lyssna.com/blog/research-synthesis-challenges/) of 300 practitioners found that **60.3% say time-consuming manual work is their biggest frustration**. AI moderation fixes much of that, but a long sales cycle puts some of the delay back.\n\n---\n\n## The self-serve alternative: Koji\n\n[Koji](https://www.koji.so) is an AI-native research platform you run yourself. You type what you want to learn, an assistant drafts the research brief, and an AI interviewer holds the conversations by voice or text in 30+ languages, or by phone.\n\n- **Published, per-interview prices.** A qualified text interview uses 1 credit and a voice interview uses 3. Credits cost €1. Sign up with 10 free credits and no card, buy packs as you go, or choose Insights at €29 a month (29 credits) or Interviews at €79 a month (79 credits).\n- **You only pay for usable interviews.** Each interview gets a [quality score from 1 to 5](/docs/understanding-quality-scores), and only those scoring 3 or more use credits or enter the report.\n- **Numbers and reasons in one study.** Six question types, from open ended to ranking and yes/no, each with AI follow-ups you control.\n- **Recruit your way.** Share a link, send [personal invite links from a CSV](/docs/importing-participants-csv) on the Interviews and Enterprise plans, or [recruit from Koji's panel](/docs/panel-recruitment) on a paid plan and see the cost before launch. [Screening questions](/docs/adding-screening-questions-to-your-study) filter out people who do not fit.\n- **Findings you can check.** Reports show themes, findings, recommended actions and personas, and every finding [links to the quotes behind it](/docs/tracing-report-findings-to-evidence).\n\n![Koji report summary with a side panel of supporting participant quotes behind one finding](/email/shots/shot-evidence-panel.jpg \"Click any finding in a Koji report to see the participant quotes that back it up.\")\n\n- **Run research from your AI assistant.** Koji's [MCP connector](/docs/mcp-overview) lets you create studies and read reports from Claude and other AI tools.\n\nWhere Outset and Conveo are ahead: Koji does not record webcam video or screen shares. If you need usability sessions on a prototype, Outset is the better fit. If you need facial-expression analysis and video reels, choose Conveo.\n\n---\n\n## Start with one real question\n\nThe quickest way to compare AI interview platforms is to run the same study on two of them. With Koji you can do it today: type your question, let the assistant draft the brief, [preview the interview](/docs/previewing-your-interview) for free, and send it to ten customers. Then take the transcripts and report into your Outset or Conveo demo and see which answers you still need.\n\n[Start free with Koji](https://www.koji.so). 10 credits, no card, and you only pay for interviews that pass the quality bar.\n\nMore reading: [Koji vs Outset](/blog/koji-vs-outset-2026), [Koji vs Conveo](/blog/koji-vs-conveo-2026), [Listen Labs vs Outset](/blog/listen-labs-vs-outset-2026) and [Listen Labs vs Conveo](/blog/listen-labs-vs-conveo-2026).\n\n---\n\n*Sources: Outset pricing page and Series B announcement (December 2025); Conveo billing documentation and agency pricing page; Software Advice listing for Conveo; Axios (September 2026); SiliconANGLE (October 2026); Maze, Future of User Research Report 2026 and 2025; Lyssna, Research Synthesis Report 2025. Pricing figures are buyer reports or vendor examples, not quotes.*\n","category":"Comparisons","lastModified":"2026-10-29T13:18:31.883294+00:00","metaTitle":"Outset vs Conveo 2026: Formats, Analysis and Pricing Compared","metaDescription":"Outset pairs AI interviews with usability tests on a custom contract; Conveo reads tone and non-verbal cues on video, billed per minute. Compare both, plus a self-serve option.","keywords":["outset vs conveo","conveo vs outset","outset ai","conveo ai","ai moderated interview platforms","ai interview tools comparison"],"aiSummary":"Outset and Conveo are AI-moderated interview platforms aimed at different buyers. Outset suits enterprise UX and insights teams: text, voice and video interviews plus usability tasks with screen sharing, 40+ languages, partner recruitment, and custom plans reported at around $20,000 a year. Conveo suits consumer brands and agencies: video, voice and text interviews with tone and non-verbal analysis, 50+ languages, integrated panels, and prepaid credits billed per interview minute, with an enterprise entry point listed at US$45,000. Both start with a demo. Koji is a self-serve alternative with published per-interview prices and 10 free credits, but no webcam video or screen sharing.","aiKeywords":["Outset","Conveo","Outset vs Conveo","AI-moderated interviews","usability testing","video research","Koji"],"aiContentType":"comparison","faqItems":[{"answer":"Outset pairs AI-moderated interviews with usability testing, letting participants share their screen on prototypes, live sites and apps. Conveo is video-first and analyses tone, facial expression and behaviour, which suits concept, ad and brand testing.","question":"What is the main difference between Outset and Conveo?"},{"answer":"Neither publishes prices. Buyer reports put Outset at around $20,000 a year, sometimes per seat. Conveo bills credits per interview minute, with an example of $5.00 a minute on its agency page and an enterprise starting price of US$45,000 listed by Software Advice. The cheaper option depends on how much you research.","question":"Is Outset or Conveo cheaper?"},{"answer":"No. Outset recruits through partners such as Prolific and User Interviews or your own users. Conveo integrates several panels, including Respondent, and accepts your own list by CSV, QR code or WhatsApp.","question":"Do Outset and Conveo have their own research panels?"},{"answer":"Neither offers a self-serve signup or a published free trial. Both start with a sales demo. Koji lets you sign up with 10 free credits and no card, and preview your interview for free.","question":"Can I try Outset or Conveo for free?"},{"answer":"B2B participants often will not go on camera, which reduces the value of video analysis. Outset fits B2B teams that need usability tests. For voice or text interviews with your own customers, Koji is a self-serve option with per-interview pricing.","question":"Which is better for B2B research?"}],"relatedTopics":["AI-moderated interviews","usability testing","video research","research platform pricing","participant recruitment"]},{"type":"blog","id":"2dda6070-1569-4d4c-8ce3-3c0c8c4a86de","slug":"outset-pricing-2026","title":"Outset Pricing in 2026: How the Custom Quote Works","url":"https://www.koji.so/blog/outset-pricing-2026","summary":"Outset does not publish prices. Its pricing page describes a standard subscription, added research support and enterprise plans, with a custom quote after a demo. Buyer reports put the entry point at around $20,000 a year, in some reports per seat, with panel recruitment and add-ons such as observer seats and single sign-on on top. The quote is set by seats, research volume, support level, recruitment source and add-ons. Outset raised $51 million in total, including a $30 million Series B led by Radical Ventures in December 2025. Koji is the self-serve alternative with published prices: 1 credit per qualified text interview and 3 per voice interview at €1 per credit, billing only interviews scoring 3 or more, with 10 free credits at signup.","content":"# Outset Pricing in 2026: How the Custom Quote Works\n\n**Short answer:** Outset does not list prices. Its pricing page says each plan is built around your team, your research needs and how much support you want, and the way in is a demo. Buyer reports put the entry point at **around $20,000 a year**, described in some reports as per seat, with participant recruitment billed on top. If you want to run AI-moderated interviews this week without a quote, [Koji](https://www.koji.so) publishes its prices: 1 credit per qualified text interview, 3 per voice interview, €1 per credit, and 10 free credits to start.\n\nThis guide covers what Outset's own pricing page says, what buyers report paying, what moves the quote up or down, and how to decide whether a custom contract or a pay per interview tool fits your team.\n\n---\n\n## Why teams are pricing Outset now\n\nAI interviewing is now normal research practice. In the 2026 State of User Research report from User Interviews and UserTesting, **80% of researchers said they use AI, up 24 points in a year**. The same report found that **91% of researchers worry about hallucinations**, which is why buyers look hard at how a platform moderates and analyses. Maze's 2025 Future of User Research report found that **55% of respondents saw demand for research rise** over the previous year.\n\nOutset has grown with that demand. It raised a **$17 million Series A led by 8VC** in 2025, then a **$30 million Series B led by Radical Ventures**, with Microsoft's venture fund M12 taking part, announced in December 2025. That brought its total funding to **$51 million**. Reported customers include Microsoft, Nestlé and HubSpot. The company now describes itself as an AI-native customer experience platform, not only a research tool.\n\n---\n\n## What Outset's pricing page actually says\n\nOutset's pricing page gives no figures. It describes three things instead:\n\n- **A standard subscription** for teams that have the people to run studies end to end themselves.\n- **Added support** for busier teams, covering setup, recruitment and other research tasks.\n- **Enterprise plans** with features and terms matched to larger teams.\n\nThe only path to a number is a demo. There is no self-serve signup or published free trial, so you cannot test the platform with your own customers before a sales conversation.\n\n---\n\n## What buyers report paying\n\nBecause the page has no price, the public numbers come from buyer reports and comparison guides. Treat them as estimates:\n\n- **Entry point around $20,000 a year.** Several guides cite this figure. Some describe it as a per-seat price, others as a starting platform price.\n- **Recruitment billed separately.** Outset lets you recruit from panels or interview your own users. Panel participants add to the bill.\n- **Add-ons.** One guide lists observer seats, single sign-on and team workspaces as items that can raise the price.\n- **Unverified claims.** Some competitor blogs quote plans from about $1,999 a month. Nothing on Outset's own site supports that, so we leave it out of the maths.\n\nIf the per-seat reading is right, here is how the platform cost scales before any recruitment:\n\n| Team | Reported platform cost per year | Plus |\n|---|---|---|\n| 1 researcher | ~$20,000 | Panel recruitment, add-ons |\n| 3 researchers | ~$60,000 | Panel recruitment, add-ons |\n| 5 researchers | ~$100,000 | Panel recruitment, add-ons |\n\nIf it is a platform price instead, the first row is your floor and seats are negotiated on top. Either way, Outset is a five-figure annual commitment.\n\n---\n\n## What sets your Outset quote\n\nBased on the pricing page and buyer reports, five things move the number:\n\n![Framework of five factors that set an Outset quote: team and seats, research volume, support level, recruitment source and add-ons.](/visuals/outset-quote-pricing-factors-framework/outset-quote-pricing-factors-framework.png \"Know where you sit on each before the demo. · Source: Outset pricing page and third-party buyer reports\")\n\n1. **Team and seats.** How many people will build studies and read results.\n2. **Research volume.** How many studies and interviews you plan to run in the year.\n3. **Support level.** Whether you run studies yourself or want Outset's team to help with setup and recruitment.\n4. **Recruitment source.** Interviewing your own users costs less than buying panel participants.\n5. **Add-ons.** Observer seats, single sign-on and workspaces.\n\nGo into the demo knowing your answer to each. A vague brief tends to produce a padded quote.\n\n---\n\n## Who Outset is built for\n\nOutset is a good product for the buyer it targets. It runs **video, voice and text interviews**, supports usability tasks inside the interview, and turns results into themes and highlight reels for stakeholders. On its own site Outset describes a customer whose focus groups had cost **$80,000 and taken five weeks**, which shows the comparison it sells against: traditional agency research, not other software.\n\n![An insights team presenting research findings on a screen to colleagues in a meeting room](https://images.unsplash.com/photo-1690191795219-d88b1af86e3b?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHwxMHx8YnVzaW5lc3MlMjB0ZWFtJTIwbWVldGluZyUyMHByZXNlbnRhdGlvbiUyMHNjcmVlbnxlbnwxfDB8fHwxNzkyMDQzNDExfDA&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Walls.io](https://unsplash.com/@walls_io?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\nThat makes Outset a strong fit for:\n\n- Enterprise insights teams replacing agency projects.\n- Teams that want video responses and clip reels for executives.\n- Companies with a research budget line and a procurement process already in place.\n\nIt is a weaker fit for a product manager who needs 20 interviews before a planning meeting, or a startup that wants to test the method before committing a budget.\n\n---\n\n## Questions to ask on the Outset demo\n\nUse the same list you would use with any quote-based vendor, and ask for each answer in writing:\n\n![Checklist of questions to ask before signing an AI interview platform contract, covering billing, quality, recruitment, data and exit terms.](/visuals/ai-interview-platform-contract-questions-checklist/ai-interview-platform-contract-questions-checklist.png \"Ask Outset each of these, and get the answers in writing.\")\n\n1. What the platform fee includes: seats, studies, languages and support.\n2. The credit or session price for each audience you plan to recruit.\n3. What counts as a billable complete, including short or flagged interviews.\n4. The rate for interviewing your own customers.\n5. Whether unused credits or sessions expire, and when.\n6. The renewal increase, in writing.\n7. Export of raw transcripts and recordings if you leave.\n\n---\n\n## The pay per interview alternative\n\nIf a five-figure contract is more than your research volume justifies, [Koji](https://www.koji.so) runs the same kind of AI-moderated interview with prices on the website:\n\n- **Published, per-interview pricing.** A qualified text interview uses 1 credit and a voice interview uses 3. Credits cost €1. A 20-interview voice study with your own customers uses about 60 credits.\n- **Billing tied to quality.** Each interview gets a [quality score from 1 to 5](/docs/understanding-quality-scores). Only interviews scoring 3 or more use credits or appear in the report.\n- **No sales call.** Sign up, get 10 free credits with no card, and run a real study. Buy credit packs as you go, or pick a monthly plan: Insights at €29 (29 credits) or Interviews at €79 (79 credits). Credits stay valid for 12 months.\n- **Voice, text and phone.** [AI voice interviews](/docs/ai-voice-interviews) and text chat in 30+ languages, plus phone interviews billed by the minute.\n- **Numbers and reasons in one study.** Six question types (open ended, scale, single choice, multiple choice, ranking and yes/no), with AI follow-up probing on the answers.\n- **Reports built for you.** [Thematic analysis](/docs/thematic-analysis-guide), findings backed by quotes, and personas, generated as interviews come in. You can also [chat with your transcripts](/docs/chat-with-interview-transcripts-ai).\n- **Recruit your way.** Share a link, send [personal invite links from a CSV](/docs/importing-participants-csv), or on a paid plan use [panel recruitment](/docs/panel-recruitment) and see the per-respondent quote in credits before you launch.\n\n![Koji research report with participant, duration, themes and quality stats above a summary of findings](/email/shots/shot-report.jpg \"A Koji report: themes, quality and a written summary, built as interviews arrive.\")\n\nWhere Outset is ahead: video responses and highlight reels, and hands-on research support from its team. If those matter more than speed and cost, Outset is the better choice.\n\n---\n\n## Outset pricing vs Koji at a glance\n\n| | Outset | Koji |\n|---|---|---|\n| Published pricing | No, custom quote after a demo | Yes |\n| Reported entry cost | ~$20,000 a year | Free to start, 10 credits |\n| Billing unit | Subscription, seats and add-ons | Credits per qualified interview |\n| Interview formats | Video, voice, text | Voice, text, phone |\n| Recruitment | Panels or your own users | Your own users, CSV invites, or panel quoted per respondent |\n| Try before you buy | Demo | Self-serve, no card |\n\nMore reading: [Koji vs Outset](/blog/koji-vs-outset-2026), [Listen Labs vs Outset](/blog/listen-labs-vs-outset-2026), and [how AI-moderated interviews work](/docs/ai-moderated-interviews).\n\n---\n\n## Start with ten interviews, not a contract\n\nThe quickest way to judge AI-moderated interviews is to run some. Type your research question into Koji, let it draft the brief, preview the interview for free, then send it to ten customers. You will see the transcripts, the themes and the report before an enterprise demo would have been booked.\n\n[Start free with Koji](https://www.koji.so). 10 credits, no card, and you only pay for interviews that pass the quality bar.\n\n---\n\n*Sources: Outset pricing page and customer story (October 2026); Outset Series A and Series B announcements (2025); third-party buyer guides for reported pricing; User Interviews and UserTesting, State of User Research 2026; Maze, Future of User Research Report 2025. Outset figures are estimates. Ask Outset for a current quote.*\n","category":"Comparisons","lastModified":"2026-10-29T13:18:31.590389+00:00","metaTitle":"Outset Pricing 2026: Custom Quotes, Costs and What to Ask","metaDescription":"Outset has no public prices. Buyers report about $20K a year to start, plus recruitment and add-ons. See what sets the quote and what to ask on the demo.","keywords":["outset pricing","outset ai pricing","outset cost","outset free trial","outset ai alternative","ai interview platform pricing"],"aiSummary":"Outset does not publish prices. Its pricing page describes a standard subscription, added research support and enterprise plans, with a custom quote after a demo. Buyer reports put the entry point at around $20,000 a year, in some reports per seat, with panel recruitment and add-ons such as observer seats and single sign-on on top. The quote is set by seats, research volume, support level, recruitment source and add-ons. Outset raised $51 million in total, including a $30 million Series B led by Radical Ventures in December 2025. Koji is the self-serve alternative with published prices: 1 credit per qualified text interview and 3 per voice interview at €1 per credit, billing only interviews scoring 3 or more, with 10 free credits at signup.","aiKeywords":["outset","outset pricing","ai moderated interviews","research platform pricing","custom quote","koji"],"aiContentType":"comparison","faqItems":[{"answer":"No. Outset's pricing page describes a standard subscription, added research support and enterprise plans, but lists no figures. You get a custom quote after a demo.","question":"Does Outset publish its pricing?"},{"answer":"Buyer reports and comparison guides put the entry point at around $20,000 a year, described in some reports as per seat. Panel recruitment and add-ons such as observer seats and single sign-on are billed on top. Outset has not confirmed these figures publicly.","question":"How much does Outset cost?"},{"answer":"Outset does not offer a self-serve signup or published free trial. The path to the product is a demo with its sales team. Koji gives 10 free credits at signup with no card.","question":"Does Outset have a free trial?"},{"answer":"Five things: the number of seats, your research volume, how much support you want from Outset's team, whether you recruit from a panel or your own users, and add-ons such as observer seats and single sign-on.","question":"What affects an Outset quote?"},{"answer":"Koji publishes its prices: 1 credit per qualified text interview and 3 per voice interview, at €1 per credit, and only interviews scoring 3 or more on quality are billed. You can buy credit packs as you go or choose a plan from €29 a month.","question":"What is a cheaper alternative to Outset?"}],"relatedTopics":["Outset","AI interview platform pricing","AI moderated interviews","research procurement","enterprise research software"]},{"type":"blog","id":"a4cb08ca-43de-4b65-9e1b-c918d34d349b","slug":"multi-market-qualitative-research-agency-playbook","title":"Multi-Market Qualitative Research: A Playbook for Agencies Fielding One Guide in Many Languages","url":"https://www.koji.so/blog/multi-market-qualitative-research-agency-playbook","summary":"A playbook for research agencies running one qualitative guide across several markets and languages. It explains why moderator variance and response styles make comparisons hard, what good multi-market practice looks like, how to set up AI-moderated interviews in Koji step by step, and where local fieldwork is still the better choice.","content":"A multi-market qualitative study is one research question answered in several languages, by people who live in different places, with results you can set side by side. The hard part has never been the fieldwork itself. It is knowing whether a difference between Germany and Brazil is a real market difference or just a different moderator, a different translation or a different week. Running the same AI-moderated interview in every market removes most of that noise, and by the end of this playbook you will be able to field one guide in many languages and defend the comparison to a client.\n\nIf you run a research agency, this is the project your clients ask for most often and trust least when it comes back. Here is how to run it well, and how to set it up in Koji.\n\nUpdated October 2026: the report language is now yours to choose, so a team that reads English can set Report language to English while the quotes stay in each market's own language.\n\n## Why multi-market qualitative research matters now\n\nClients want the \"why\" in every market they sell in, and they want it faster than a round of local fieldwork partners can deliver. At the same time, the cheap way to get breadth, an online survey panel, has a data quality problem. A 2026 literature review from NORC puts fraud at an estimated 15 to 30 percent across the market research industry and as high as 45 percent on some survey platforms. NORC attributes those figures to earlier studies rather than measuring them itself, so read them as industry estimates, but the direction is not in dispute ([NORC, April 2026](https://www.norc.org/content/dam/norc-org/pdf2026/cpss-research-brief-fraud-lit-review.pdf)). If your multi-market read rests on panel open-ends from five countries, you are carrying that risk five times.\n\n![A research team on a video call with colleagues in other countries, planning a study across several markets](https://images.unsplash.com/photo-1758873269461-49cfd01504c1?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHwxfHx0ZWFtJTIwdmlkZW8lMjBjYWxsJTIwbGFwdG9wJTIwaW50ZXJuYXRpb25hbHxlbnwxfDB8fHwxNzkyMDQ3MDEwfDA&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Vitaly Gariev](https://unsplash.com/@silverkblack?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\nLanguage matters to the buyer too. CSA Research surveyed 8,709 consumers in 29 countries in 2020 and found that 76 percent prefer to buy products with information in their own language ([CSA Research, 2020](https://csa-research.com/blogs-events/csa-in-the-media/press-releases/consumers-prefer-their-own-language)). People also answer research questions better in the language they think in. An interview run in a second language flattens the words people reach for.\n\nThen there is the comparability problem, which is older than AI. Harzing's study of 26 countries showed that response styles differ by country, with traits such as power distance and collectivism linked to more agreement and more extreme ratings ([Harzing, 2006, summarised on her site](https://harzing.com/blog/2017/02/what-if-fully-agree-doesnt-mean-the-same-thing-across-cultures)). Clarke found the same for extreme responding in a four-country application, and warned that unadjusted cross-cultural comparisons can lead marketing researchers to wrong conclusions ([Clarke, 2001](https://www.emerald.com/insight/content/doi/10.1108/02651330110396488/full/html)). Add a different human moderator in every country and you have stacked a second source of variation on top of the first.\n\nFinally, the depth argument has real evidence behind it now. In a randomized experiment on the Verasight panel with 3,160 respondents, AI-moderated interviews produced 4.8 times as many words in free-response questions as standard written open-ends. The same report found completion fell from 99.4 percent to 40.5 percent, with drop-off concentrated among people less comfortable with AI tools ([Verasight, August 2026](https://verasight.io/reports/ai-moderated-interviews-vs-survey-open-ends)). Both halves matter, and we come back to the second one below.\n\nThe cost of doing nothing is the usual one. You either deliver a thin, translated-survey read that the client has already seen from three other suppliers, or you quote weeks of fieldwork and lose the pitch to someone faster.\n\n## What good looks like\n\nExperienced multi-market teams agree on a few habits. None of them are new, but most projects skip at least two.\n\n**Fix the objectives, not the wording.** Quirks' guidance on international qualitative work is to be specific about what you need to learn and loose about how each market phrases it ([Quirks](https://www.quirks.com/articles/overcoming-the-obstacles-to-conducting-international-qualitative-research)). Literal translation of a question can change its meaning. Write each question as an objective (\"find out how they decided, in their own words\") and let the interviewer carry the local phrasing.\n\n**Hold the interviewer constant.** Most of the noise in a classic multi-country project is moderator variance: pace, warmth, how hard they probe, how much they lead. If one interviewer logic runs in every market, a difference you see is far more likely to belong to the market.\n\n**Pair open conversation with a few identical numbers.** A single-choice question on the decision, and a 1 to 10 rating of the concept, asked the same way everywhere, let you compare markets on something countable. Then read the open answers to understand the numbers, and remember Harzing: do not compare raw ratings between countries as if scale use were equal. Compare the reasons behind the ratings, and compare each market against its own baseline.\n\n**Size each market on its own.** A common rule of thumb is twenty to thirty good interviews per market to see the main themes. Set a response limit per market so a fast-responding country does not crowd out a slow one.\n\n**Check the participant, not just the answer.** With panel recruitment in particular, read the first interviews in each market before you let fieldwork run on. Short, generic or off-topic answers are visible in the first handful.\n\n**Read the originals.** Whatever tool codes the themes, read a sample of verbatims in the original language, or have a native speaker do it. A theme label in English is a summary of what someone said, not the thing itself.\n\nCommon mistakes:\n\n- Translating the guide word for word and calling it standardised.\n- Comparing average ratings across countries without allowing for how each culture uses a scale.\n- Letting one market run with a different screener because the local partner \"knows the audience\".\n- Merging all markets into one set of themes before looking at each market alone.\n- Treating a machine translation of consent or privacy text as final.\n\n![Checklist for a multi-market qualitative study: objectives, shared questions, translation review, previews, market caps and reading original quotes.](/visuals/multi-market-qualitative-guide-checklist/multi-market-qualitative-guide-checklist.png \"Run this list before you open fieldwork in a second market.\")\n\n## How to set up a multi-market study in Koji\n\nKoji runs the same AI interviewer in each language you enable, so the interview logic stays the same while the language changes. Text interviews work on every plan, including Free. Voice interviews need the Interviews plan or Enterprise. Panel recruitment needs a paid plan, and sharing a study with client or team workspaces is available from the Insights plan.\n\n![Process with six steps for setting up a multi-market qualitative study in Koji, from the agency template to language review and a shared report.](/visuals/multi-market-qualitative-study-setup-steps/multi-market-qualitative-study-setup-steps.png \"Six steps from one guide to a comparable read across markets.\")\n\n1. **Start from the agency interview template.** In the dashboard's template gallery, open the In-Depth Interview template for agencies. It gives you a rapport opening, behaviour-based questions and probing rules written for client work. Put your client's category and decision into the research brief, write each question as an objective, and add guardrails for anything the interviewer should not discuss, such as the sponsor or pricing.\n2. **Add structured questions for the comparable bits.** Use a single-choice question for country or city, one for the decision made, and a 1 to 10 scale for the concept or brand, each with an open follow-up. These numbers line up across markets in the report.\n3. **Enable the languages and review the text.** Participants pick their language on the welcome page. The study's own text, such as the title, description and button, can be machine translated into each language in the Texts and Translations editor, and each field is marked as machine translated or reviewed once a person has edited it. Consent text is treated more strictly: a machine translation of it is not shown to participants, who get the source-language version until a person approves the translation. Have a native speaker check that text.\n4. **Preview in every language.** Run a preview interview in each language. Previews are saved with their own analysis but are not counted or charged, so use them to catch awkward phrasing before fieldwork.\n5. **Choose voice or text, invite or recruit, and cap each market.** Voice suits markets where people prefer to talk. Text suits busy professionals and sensitive topics. Send a custom link per market, or use panel recruitment, where the wizard shows a feasibility estimate for your audience, so check it for every market before you commit. Set a response limit for each market, then read the first handful of conversations and their quality scores before the rest come in.\n6. **Read the report and share it.** Open the report, compare findings by market using the country question, and publish it or share it with your client's workspace.\n\n### Make the invitation feel local\n\nThe welcome page is the first thing a participant sees in their own language. Spend ten minutes here. It shows the estimated time, the privacy promise and the language choice.\n\n![Koji translations editor with German interview text marked as reviewed or machine translated.](/email/shots/shot-translations.jpg \"German study text with a status on each field: reviewed by a person, or machine translated.\")\n\nExample questions, written for a beverage client testing a new concept across markets:\n\n- \"Think about the last time you bought a drink like this. Walk me through it from the moment you decided you wanted one.\" (open, follow-ups on)\n- \"Where did you buy it?\" (single choice, with the channels that matter in each market)\n- \"Looking at this concept, how likely are you to pick it over what you buy now?\" (1 to 10), followed by \"What is behind that number?\"\n- \"What would a friend in your city say if you showed them this?\" (open)\n- \"Is there anything about this that would not work where you live?\" (open)\n- \"Which country do you live in?\" (single choice, used to read results by market)\n\nEach market costs credits per completed interview: 1 credit for a text interview and 3 for a voice interview. Twenty text interviews in each of five markets is 100 credits. The same plan in voice is 300 credits. Conversations that score below 3 out of 5 on quality are not charged, on every plan ([how the quality gate works](/docs/how-the-quality-gate-works)).\n\n## What you get\n\nFor each study you get per-interview analysis, a quality score, and a report with themes, high-impact findings and the supporting quotes behind each one. The report is written in the report language you pick (the study's language by default, otherwise English), so if your team reads English, set Report language to English on the Report page or in the Context tab. Supporting quotes stay in the participant's original words, so a reviewer can check the claim against what was actually said.\n\n![Koji report summary with a highlighted claim and a side panel of supporting interview quotes.](/email/shots/shot-evidence-panel.jpg \"Every finding opens to the quotes behind it.\")\n\nHow agencies use it the next day:\n\n- **In the debrief:** open the report with the client and click through from a finding to the quotes. The argument shifts from \"do you believe this\" to \"do you see the same thing\".\n- **In the deck:** use the country question to show the same theme in each market, with a native-language quote next to it.\n- **In the proposal for wave two:** the first wave shows which markets need more interviews and which are already saturated. Report history keeps each version if you add interviews later.\n- **In the data room:** export responses to CSV on any plan if you want to do your own cut.\n\n## The same playbook elsewhere\n\nThe same one-guide, many-voices approach works wherever a comparison has to hold up. Brand teams run it across regions before a rebrand. HR teams run it across offices to see whether an engagement result is global or local. Customer success teams interview accounts in each region before a renewal cycle. If your client is an in-house team, the neighbouring playbooks are written for them: see [pivot decision research for founders](/blog/pivot-decision-research-founders) for a decision-first structure, and [no-decision loss interviews for GTM teams](/blog/no-decision-loss-interviews-gtm-playbook) for another playbook built on asking the same questions of many buyers. All Koji playbooks for agencies are listed on the [use-case library page for agencies](/use-cases/agencies).\n\n## Why Koji for multi-market qualitative work\n\nThe table sets Koji against the two ways most agencies cover several markets today. Local moderators and fieldwork partners are still the right answer in many cases, and a translated survey is still the right way to count things.\n\n![Table comparing local moderators, translated surveys and Koji for multi-market research on depth, cost, scale, consistency, candour and data handling.](/visuals/multi-market-qualitative-fieldwork-comparison/multi-market-qualitative-fieldwork-comparison.png \"How the three approaches compare on the criteria a multi-market buyer weighs.\")\n\nReasons that hold up in the product:\n\n- **One interviewer logic in every language.** The same interviewer and probing rules run in each language you enable, so moderator variance stops being a variable in your comparison. The language list in the app has 30+ languages.\n- **Structured questions next to open conversation.** Choice and scale questions sit inside the interview, so you get comparable numbers and the reasons behind them from the same person.\n- **Evidence you can audit.** Quality scores per interview, and every finding links back to its supporting quotes.\n- **Control over what gets said.** Guardrails keep the interviewer off topics you name, and you can preview every language before launch.\n- **Data handling.** Koji states EU data residency. If your client needs control of the model keys, Bring Your Own Key is part of the Enterprise plan ([bring your own key](/docs/bring-your-own-key)).\n\nFor a wider view of how this compares with a large agency, see [Koji vs Kantar](/blog/koji-vs-kantar-2026), and for the tools market see [best multilingual research tools](/blog/best-multilingual-research-tools-2026).\n\n## Questions you might have about AI interviews\n\n**Will people in every market talk to an AI interviewer?** Some will not. The Verasight experiment found that completion dropped from 99.4 percent to 40.5 percent when open-ends were replaced with AI-moderated interviews, and that those who finished were not a perfect mirror of those who left. Plan for it: give participants a choice of voice or text, keep the welcome page honest about length and privacy, and compare your completion by market. A market with much lower completion is a finding in itself. Do not read a completed sample as the whole market.\n\n**Can the interview handle local idiom and nuance?** The interviewer conducts the whole conversation in the language the participant chooses. That does not make it a replacement for local knowledge. Have a native speaker review the preview in each language, and read the original-language quotes before you sign off a theme.\n\n**What about quality and fake respondents?** Every conversation gets a quality score from 1 to 5, and you can read individual interviews and their scores in the report. Short or off-topic conversations show up as low scores, and anything below 3 is not charged. This is a check on the conversation, not a guarantee about who is on the other end, so check the first interviews from any panel yourself. The deeper reading on this is in [survey fraud and respondent quality](/docs/survey-fraud-respondent-quality).\n\n## When Koji may not be the right fit\n\nKoji is not the right tool when the project needs an in-person observation, a physical product to handle, a group dynamic you want to watch, or a language outside the 28 in the app. It is also not a substitute for a local expert who knows what a market's taboos are. If the client requires a named senior moderator in the room, or a market where you cannot recruit through a link or the panel, use local fieldwork for that market and use Koji for the rest. Many agencies end up with exactly that hybrid.\n\n## Metrics to track\n\n- Completion rate by market, and where in the interview people leave\n- Average quality score by market, and the share scoring below 3\n- Average interview length by market, as a check on depth\n- Share of each market's themes that also appear in the other markets\n- Time from brief to first report\n- Client questions in the debrief that you answered by opening a quote\n\n## Frequently asked questions\n\n**How many interviews do I need per market?** A common rule of thumb is twenty to thirty good interviews per market to see the main themes. Add more if a market has several distinct segments you need to read separately.\n\n**Can I run voice interviews in every market?** Voice interviews are available on the Interviews and Enterprise plans, and the language list has 30+ languages. Text interviews work on every plan, so you can mix: voice where people like to talk, text elsewhere.\n\n**How do I compare markets fairly?** Ask the same structured questions everywhere, read each market on its own first, and compare reasons rather than raw ratings, because scale use differs by country.\n\n**Does the report translate quotes?** No. Themes and findings are written in the report language you choose (the study's language by default, otherwise English), and supporting quotes stay in the participant's own words so you can check them against the claim.\n\n**Can I use Koji for part of a project and local fieldwork for the rest?** Yes. Many agencies run most markets in Koji and keep local moderators for the markets or methods that need a person in the room.\n\n## Start your first multi-market study\n\nStart from the agencies In-Depth Interview template in your Koji dashboard, and run a preview in two languages before you commit to five. For more about how agencies use Koji, see the [agencies page](/for/agencies) and the [user research for agencies guide](/docs/user-research-for-agencies). For the underlying detail on languages, see the [multi-language research guide](/docs/multilingual-research-guide).\n","category":"Use Cases","lastModified":"2026-10-29T13:18:31.322997+00:00","metaTitle":"Multi-Market Qualitative Research: Agency Playbook","metaDescription":"How agencies field one qualitative guide in many languages, keep markets comparable and run it with AI-moderated interviews in Koji.","keywords":["multi-market qualitative research","multi-country qualitative research","international qualitative research","multilingual interviews","research agency playbook","AI-moderated interviews"],"aiSummary":"A playbook for research agencies running one qualitative guide across several markets and languages. It explains why moderator variance and response styles make comparisons hard, what good multi-market practice looks like, how to set up AI-moderated interviews in Koji step by step, and where local fieldwork is still the better choice.","aiKeywords":["multi-market qualitative research","multi-country research","multilingual AI interviews","research agencies","qualitative fieldwork","cross-cultural research"],"aiContentType":"guide","faqItems":[{"answer":"A common rule of thumb is twenty to thirty good interviews per market to see the main themes. Add more if a market has several distinct segments you need to read separately.","question":"How many interviews do I need per market?"},{"answer":"Voice interviews are available on the Interviews and Enterprise plans, and the language list has 30+ languages. Text interviews work on every plan, so you can mix voice where people like to talk with text elsewhere.","question":"Can I run voice interviews in every market?"},{"answer":"Ask the same structured questions everywhere, read each market on its own first, and compare the reasons behind ratings rather than raw ratings, because scale use differs by country.","question":"How do I compare markets fairly?"},{"answer":"No. Themes and findings are written in the report language you choose (the study's language by default, otherwise English), and supporting quotes stay in the participant's own words so you can check them against the claim.","question":"Does the report translate participant quotes?"},{"answer":"Yes. Many agencies run most markets in Koji and keep local moderators for the markets or methods that need a person in the room.","question":"Can I use Koji for part of a project and local fieldwork for the rest?"}],"relatedTopics":["multi-market qualitative research","research agencies","multilingual research","AI-moderated interviews"]},{"type":"blog","id":"06d7894e-d643-48ea-8fc0-a6fcbb08d2fa","slug":"maze-alternatives-2026","title":"8 Best Maze Alternatives for User Research in 2026 (Ranked & Reviewed)","url":"https://www.koji.so/blog/maze-alternatives-2026","summary":"Maze is a strong unmoderated prototype-test tool but lacks AI-moderated voice interviews, automatic thematic analysis, and qualitative depth. Koji is the best Maze alternative in 2026 for founders, PMs, researchers, and agencies: AI-moderated voice + text interviews, six structured question types, automatic thematic analysis, and interviews as low as €1 per qualified interview (€3 per qualified voice interview). Start free with 10 credits, no card, then pay as you go with no subscription.","content":"# 8 Best Maze Alternatives for User Research in 2026 (Ranked & Reviewed)\n\nIf you've used Maze for usability testing or prototype validation, you've probably hit the same wall every other product team eventually hits: it's great for unmoderated card sorts and prototype clicks, but it can't actually *talk* to your users. No moderated interviews. No voice. No real qualitative depth. And the qualitative depth is exactly where modern teams need to invest in 2026.\n\nThis guide ranks the 8 best Maze alternatives for user research in 2026, with a clear-eyed look at what each tool does well, where it falls short, and which type of team should pick it.\n\n## TL;DR — The Best Maze Alternative in 2026\n\nFor most product teams, founders, and researchers, **Koji is the strongest Maze alternative** in 2026. It replaces Maze's prototype-only model with AI-moderated voice and text interviews, automatic thematic analysis, six structured question types, and one-click reports — at a price point startups and mid-market teams can actually afford.\n\nIf you specifically need pixel-perfect prototype testing, stick with a hybrid (Koji for interviews + Lyssna or UXtweak for prototype tasks). If you need an enterprise panel with 1M+ recruits, UserTesting is the legacy fit.\n\n## Why Teams Leave Maze\n\nMaze does some things well — fast unmoderated tests, clean reporting on click paths, lightweight surveys. But the limitations have become hard to ignore as research expectations grow.\n\nThe most cited Maze problems in 2026:\n\n- **No moderated interviews or voice.** Maze can't run a live or AI-moderated conversation with a user. You see clicks; you don't hear stories.\n- **Shallow qualitative analysis.** Open-ended question reports are basic. There's no automatic thematic clustering or quote extraction.\n- **No native participant panel included in mid-tier plans.** External recruitment costs add up fast.\n- **Prototype crashes on mobile.** A persistent complaint in 2026 reviews — mobile testing is unreliable.\n- **Limited conditional logic.** You can't build a real adaptive survey flow.\n- **Reporting isn't editable.** Pre-defined reports can't be combined or restructured into a deliverable.\n\nMeanwhile, the research landscape has shifted hard. **78% of UX and product teams now use AI in their research workflows** — more than double the 34% adoption rate in 2024 — and **88% of researchers identify AI-assisted analysis as the top trend impacting research in 2026**. Maze's unmoderated-only model puts you on the wrong side of that shift.\n\nThe teams that win in 2026 don't run static clickthroughs. They run two-way AI-moderated interviews that adapt in real time, then synthesize themes automatically. That's the category every Maze alternative below is competing in.\n\n## The 8 Best Maze Alternatives in 2026\n\n### 1. Koji — Best AI-Native Maze Alternative Overall\n\n**Best for:** Founders, PMs, UX researchers, and agencies who want fast, deep qualitative insight without paying for legacy enterprise contracts.\n\n[Koji](https://www.koji.so) is an AI-native customer research platform that does what Maze *cannot*: it runs full two-way conversations with your users — in voice or text — using an AI moderator that probes deeper, follows up adaptively, and stays neutral. Then it synthesizes the conversations into themes, quotes, and an editable report automatically.\n\n**Why Koji beats Maze:**\n\n- **Two-way AI-moderated interviews** (voice or text) in 30+ languages, not just click-tracking.\n- **Six structured question types** — open-ended, scale, single choice, multiple choice, ranking, yes/no — so you can run survey *and* interview logic in one study. See the [structured questions guide](/docs/study-questions-guide) for details.\n- **Automatic thematic analysis** — quotes, sentiment, theme clusters generated within minutes of the last interview closing.\n- **Customizable AI consultants** that act as domain experts (PMM, UX lead, founder coach) when interviewing your participants.\n- **Priced per interview, not per seat.** Interviews start as low as €1 per qualified interview, and €3 per qualified voice interview. A conversation that scores below 3 out of 5 is free and never charged, so you only pay for the ones that produced something. Start with pay as you go. No subscription needed. Volume pricing and plans are there when you want them.\n- **No annual lock-in.** Cancel monthly.\n- **MCP integration** — query your research directly from Cursor, Claude, or any MCP client. The API and MCP connector are available on every plan, including a free account; outbound webhooks are not available yet.\n- **Built-in recruitment.** You do not need an audience of your own: interview your own users, or describe the people you want and Koji recruits them, with a live per respondent quote in credits you approve before launch.\n\n**Where Maze wins:** If your entire workflow is prototype click-testing on a Figma export, Maze's UX for that one specific task is more polished. But that workflow is shrinking — 4.5x more insightful responses come from AI-moderated interviews than from static surveys or prototype clickthroughs.\n\n**Pricing:** As low as €1 per qualified interview, and €3 per qualified voice interview. Start free with 10 credits, no card. Then pay as you go. No subscription needed.\n\n---\n\n### 2. Lyssna (formerly UsabilityHub)\n\n**Best for:** Quick unmoderated tests with a built-in panel.\n\nLyssna covers a lot of the same surface area as Maze — first-click tests, preference tests, five-second tests, surveys — and includes a participant panel. The qualitative depth is still limited though, and like Maze, you don't get true two-way interviews.\n\n**Strengths:** Built-in panel, clean UI, fair pricing for usability tasks.\n**Limitations:** No voice or AI-moderated interviews. Analysis is mostly counts and percentages.\n\n**Pricing:** Starts around $89/month for Basic.\n\n---\n\n### 3. UserTesting\n\n**Best for:** Enterprises with budget who need a large participant pool and human-moderated sessions.\n\nUserTesting is the legacy heavyweight: 1M+ participant panel, demographic filters, moderated and unmoderated tests, video recordings, Figma/Miro integrations. It's the safest \"enterprise\" choice — and the most expensive.\n\n**Strengths:** Massive panel, mature platform, audit-friendly compliance.\n**Limitations:** Enterprise pricing (often $20K+/seat/year), slow to deploy, analysis is still largely manual unless you bolt on their AI add-ons. Best fit for organizations where research is a dedicated function, not a PM side-quest.\n\n---\n\n### 4. UXtweak\n\n**Best for:** Teams needing advanced mobile testing and card sorting on a budget.\n\nUXtweak earns a real spot here because of mobile testing reliability — exactly where Maze breaks. Card sorting, tree testing, prototype testing, and session recording are all solid. Pricing scales reasonably.\n\n**Strengths:** Strong mobile, broad UX-method coverage, EU-based (good for GDPR-sensitive teams).\n**Limitations:** No AI-moderated voice interviews. Reporting feels old-school.\n\n---\n\n### 5. Optimal Workshop\n\n**Best for:** Information architecture work — card sorting, tree testing, first-click testing.\n\nOptimal serves Netflix, LEGO, and Apple, and has SOC 2 compliance with SSO. If your research focus is IA and navigation, it's a strong specialist tool. It's not trying to be a full discovery platform though.\n\n**Strengths:** Best-in-class IA methods, enterprise compliance.\n**Limitations:** Narrow scope, no two-way interviews, expensive for what it covers.\n\n---\n\n### 6. Lookback\n\n**Best for:** Live moderated 1:1 sessions where a human researcher runs the call.\n\nLookback gives you human-moderated remote interviews with screen sharing and recording. If you specifically want the researcher in the chair, it's a clean tool. The catch: it doesn't scale — every interview requires a researcher's time. Compare that to Koji running hundreds of AI-moderated voice interviews in parallel.\n\n**Strengths:** Solid live-session UX, good for usability deep-dives.\n**Limitations:** No AI moderation, no automatic synthesis, slow.\n\n---\n\n### 7. Userlytics\n\n**Best for:** Mid-market teams wanting both moderated and unmoderated testing with a panel.\n\nUserlytics is a solid generalist — moderated, unmoderated, mobile, prototype testing, and a participant network. They've added AI summarization, but it sits on top of a traditional video-task workflow rather than replacing it.\n\n**Strengths:** Flexible methods, decent pricing for what's included.\n**Limitations:** AI features are bolt-on rather than native. Reports are video-clip heavy, theme-light.\n\n---\n\n### 8. Dovetail (with a research tool of your choice)\n\n**Best for:** Teams who already do interviews elsewhere and need a research repository.\n\nDovetail isn't a recruitment or interview platform — it's the repository where you store and tag transcripts. So it's only a Maze \"alternative\" if you pair it with something that actually runs the studies (like Koji). On its own, it doesn't replace Maze.\n\n**Strengths:** Tagging, repository search, AI summaries on uploaded transcripts.\n**Limitations:** You still need a separate tool to recruit and run interviews. See [Koji vs Dovetail](/docs/koji-vs-dovetail) for the full comparison.\n\n---\n\n## Maze Alternatives Compared at a Glance\n\n| Tool | AI-Moderated Voice | Auto Thematic Analysis | Built-in Recruitment | Starting Price | Best Fit |\n|---|---|---|---|---|---|\n| **Koji** | ✅ | ✅ | ✅ | Free, then from €1 per qualified interview | Founders, PMs, researchers, agencies |\n| Lyssna | ❌ | Partial | ✅ | $89/mo | Quick usability tasks |\n| UserTesting | ❌ (human-moderated only) | Partial | ✅ | ~$20K/yr | Enterprise research teams |\n| UXtweak | ❌ | ❌ | Partial | $80/mo | Mobile testing, IA |\n| Optimal Workshop | ❌ | ❌ | Partial | ~$208/mo | Information architecture |\n| Lookback | ❌ (human-moderated) | ❌ | ❌ | $25/mo | Live 1:1 sessions |\n| Userlytics | Partial | Partial | ✅ | $49/test | Mid-market generalists |\n| Dovetail | ❌ | ✅ | ❌ | $39/mo | Research repository only |\n\n## How to Pick the Right Maze Alternative\n\nThree questions clarify the decision fast:\n\n**1. Do you need actual conversations with users — or just click data?**\nIf conversations, you need AI-moderated interviews. Koji is the only tool above that runs them at scale on a startup budget.\n\n**2. How much qualitative analysis do you want to do manually?**\nIf \"none,\" you need automatic thematic analysis. Koji and (to a lesser extent) Dovetail are the options. Everyone else hands you transcripts and counts.\n\n**3. What's your budget reality?**\nFounders and small teams: Koji or Lyssna. Mid-market: Koji or Userlytics. Enterprise with procurement cycles: UserTesting or Koji's Enterprise plan.\n\nThe single biggest 2026 trend is that AI-moderated platforms compress 4–6 week qualitative research cycles to under 24 hours. If your team is still treating \"qualitative research\" as a multi-week project, the gap is going to keep widening.\n\n## Stop Settling for Click Data — Start Talking to Your Users\n\nMaze was built for a 2020-era research workflow where prototype clicks were the deepest insight a small team could afford. That era is over. **AI-moderated interviews now deliver 4.5x more insightful responses than traditional surveys**, in any language, at roughly one-third the cost of legacy research.\n\n[Koji](https://www.koji.so) is the modern replacement: real conversations, automatic themes, transparent pricing, no annual contract. You can run your first AI-moderated study today — voice or text — and have a publishable insights report by tonight.\n\n**[Try Koji free →](https://www.koji.so)**\n\n---\n\n## Frequently Asked Questions\n","category":"Research","lastModified":"2026-10-29T13:18:31.053873+00:00","metaTitle":"8 Best Maze Alternatives for User Research in 2026 (Ranked)","metaDescription":"The 8 best Maze alternatives in 2026 ranked. Compare Koji, Lyssna, UserTesting, UXtweak, Lookback and more on AI interviews, pricing, and qualitative depth.","keywords":["maze alternatives","maze alternatives 2026","alternatives to maze","maze competitors","best maze alternatives","ai user research tool","ai moderated interview platform"],"aiSummary":"Maze is a strong unmoderated prototype-test tool but lacks AI-moderated voice interviews, automatic thematic analysis, and qualitative depth. Koji is the best Maze alternative in 2026 for founders, PMs, researchers, and agencies: AI-moderated voice + text interviews, six structured question types, automatic thematic analysis, and interviews as low as €1 per qualified interview (€3 per qualified voice interview). Start free with 10 credits, no card, then pay as you go with no subscription.","aiKeywords":["maze alternatives","alternatives to maze","ai moderated interview platforms","ai user research","user research tools 2026","qualitative research platform"],"aiContentType":"listicle","faqItems":[{"answer":"Yes for most teams. Koji runs interviews as low as €1 per qualified interview and €3 per qualified voice interview, starting free with 10 credits and no card, then pay as you go with no subscription, while Maze's Starter plan is $99/month and limits unmoderated tests. For two-way AI-moderated voice interviews, Maze doesn't offer them at all. Koji is the only AI-native option in that price range.","question":"Is Koji really cheaper than Maze?"},{"answer":"Koji runs two-way AI-moderated voice and text interviews, generates automatic thematic analysis with quotes and sentiment, supports six structured question types in one study, and offers a customizable AI consultant. Maze is limited to unmoderated prototype clicks, surveys, and basic usability tasks.","question":"What can Koji do that Maze can't?"},{"answer":"Yes. Some teams use Maze for prototype click-tracking on early-stage Figma exports and Koji for deeper qualitative interviews and concept testing. But most teams in 2026 consolidate into Koji because it covers both qualitative interviews and structured surveys in one platform.","question":"Can I use Koji and Maze together?"},{"answer":"Only if your specific need is prototype click-testing with a built-in panel (Lyssna) or enterprise-scale moderated sessions with a 1M+ participant pool (UserTesting). For most product teams in 2026, Koji wins on price, speed, AI-moderated voice interviews, and automatic thematic analysis.","question":"Is Lyssna or UserTesting a better Maze alternative than Koji?"},{"answer":"About an afternoon. You can sign up for Koji free, build your first AI-moderated study in under 30 minutes, and run your first interviews the same day. Most teams report having their first publishable insights report within 24 hours of switching.","question":"How long does it take to switch from Maze to Koji?"}],"relatedTopics":["koji-vs-maze-2026","best-ai-customer-interview-tools-2026","best-user-research-tools-for-startups-2026","ai-moderated-interview-platforms-2026","best-ux-research-repository-tools-2026"]},{"type":"blog","id":"58135ba4-984b-46e2-b26f-04a722f1cf46","slug":"marvin-pricing-2026","title":"Marvin Pricing in 2026: Free, Starter, Pro and Enterprise Explained","url":"https://www.koji.so/blog/marvin-pricing-2026","summary":"Marvin (HeyMarvin) has four plans in 2026: Free, Starter, Pro and Enterprise. Free costs nothing, needs no card, and includes 5 file uploads a month and limited AI-moderated interviews (up to 5). Starter, Pro and Enterprise are quote-only; review sites last listed Essentials from $3,000 a year and Standard from $6,000 a year, each for a five-user minimum. Marvin is a research repository and analysis platform with AI tagging, Ask AI search, AI-moderated interviews in 40+ languages and Live Intercept (September 2026), with SOC 2 Type II, HIPAA and GDPR; G2 rating 4.8 from about 133 reviews. Koji is an interview-first alternative with public pricing: free with 10 credits, Insights EUR 29/month, Interviews EUR 79/month, 1 credit per text and 3 per voice interview, only interviews scoring 3+ of 5 use credits, AI voice and text interviews in 30+ languages with automatic reports.","content":"Marvin (HeyMarvin) has **four plans: Free, Starter, Pro and Enterprise**. Only the Free plan has a public price: it costs nothing, needs no credit card and includes 5 file uploads a month plus its AI Moderated Interviewer. **Starter, Pro and Enterprise are quote-only.** Review sites last listed Marvin's paid tiers at about **$3,000 a year (Essentials) and $6,000 a year (Standard), each with a five-user minimum**, roughly $50 and $100 per user per month. Treat those figures as a floor, not a quote.\n\nBelow is what each plan includes, what the historical numbers tell you about a real quote, how Marvin's AI interviews are counted, and when an interview-first tool with public pricing is the better fit.\n\n## Marvin plans and pricing in 2026\n\n| | Free | Starter | Pro | Enterprise |\n|---|---|---|---|---|\n| Price | $0, no card | Contact sales | Contact sales | Contact sales |\n| Uploads | 5 files a month | More | More | Custom |\n| AI Moderated Interviewer | Included, limited | Included | Included | Included |\n| What paid plans add | | More seats, integrations, AI features and admin controls | | |\n\nSource: [Marvin's pricing page](https://heymarvin.com/pricing) as indexed in October 2026. Marvin does not publish per-plan seat counts, hours or storage limits. A third-party guide from [Enterpret](https://www.enterpret.com/guides/the-6-best-heymarvin-alternatives-in-2026) also describes paid plans as quote-based as of September 2026.\n\n### What third parties report\n\nBefore Marvin moved to quote-only pricing, review sites listed:\n\n- **Essentials:** from $3,000 a year for 5 users ([TrustRadius](https://www.trustradius.com/products/heymarvin/pricing))\n- **Standard:** from $6,000 a year for 5 users, with a five-user minimum ([G2](https://www.g2.com/products/marvin-marvin/pricing), TrustRadius)\n- **Enterprise:** custom\n\nThose plan names no longer match the pricing page, so use them only to anchor your budget. A five-person team should expect to start in the low thousands of dollars a year, and more for enterprise security, more seats or heavy interview volume. Competitor Looppanel also says Marvin does not offer month-to-month billing, so plan for an annual contract.\n\n### How Marvin counts AI interviews\n\nMarvin's [help center](https://help.heymarvin.com/en/articles/11402590-how-do-i-use-ai-moderated-interviewer) says Free users can run up to 5 AI interviews, while paid customers currently get unlimited AI interviews, and that the AI interviewer is open to all plans \"for a limited period\". Its homepage also offers 75 free interviews when you request a demo. Both look like launch offers, so ask how AI interviews will be priced at renewal before you build a research program on them.\n\n## What Marvin does well\n\nMarvin started as a research repository and analysis tool and has grown into a broader platform. Its strengths:\n\n- **One home for qualitative data.** You can invite Marvin to Zoom, Microsoft Teams and Google Meet calls, or upload audio, video, Word, Google Docs, spreadsheets, slides, PDFs and text files.\n- **AI analysis and search.** Automatic transcription, AI tagging and themes, and Ask AI search across past research. In January 2026 Marvin launched [agentic Ask AI](https://www.businesswire.com/news/home/20260127667920/en/HeyMarvin-Launches-Industrys-First-Agentic-AI-Search-for-Customer-Research), a multi-agent search over your research.\n- **AI-moderated interviews** in more than 40 languages, and since September 2026, [Live Intercept](https://www.financialcontent.com/article/bizwire-2026-9-16-marvin-launches-live-intercept-bringing-voice-to-voice-ai-moderated-research-inside-the-digital-customer-experience), a voice AI interviewer embedded in a website or product.\n- **Security.** Marvin lists SOC 2 Type II, HIPAA and GDPR compliance, with US and EU data centers.\n\nReviewers are positive: Marvin holds 4.8 out of 5 from about 133 reviews on [G2](https://www.g2.com/products/heymarvin/reviews). The common complaints are that the AI sometimes misreads a question or makes up an answer, transcription and translation are uneven in some languages, there is a learning curve across many tabs, quantitative and survey analysis is limited, and the price feels high next to alternatives.\n\n## Why a repository is worth paying for, and when it is not\n\nA repository earns its price when research gets used. Marvin's own [State of Modern Research 2026](https://www.businesswire.com/news/home/20260604383986/en/HeyMarvin-Reveals-94-of-Leaders-Say-Research-Should-Drive-Decisions-but-Only-27-Consistently-Use-It) survey (vendor research, 309 respondents) found 94% of leaders say research should drive decisions, but only 27% cite it in nearly every major decision. If your team already has years of recordings, notes and survey files scattered across drives, a searchable home for them is valuable.\n\nAI is now part of most research workflows, which is why every tool in this category leans on it. In User Interviews' [State of User Research 2025](https://www.userinterviews.com/state-of-user-research-report), 80% of researchers said they use AI, up 24 points in a year. Maze's [Future of User Research 2025](https://maze.co/resources/user-research-report-2025/) put the share of product professionals using AI tools at 58%, up from 44% the year before, and its [2026 report](https://maze.co/resources/user-research-report/) found 69% use AI in at least some projects.\n\nBut a repository only helps once there is research to store. If your problem is that you are not hearing from enough customers in the first place, you are paying for shelves before you have anything to put on them.\n\n![Table comparing Marvin and Koji on main job, interview modes, languages, research library, published pricing and billing](/visuals/marvin-vs-koji-comparison-table/marvin-vs-koji-comparison-table.png \"Repository first or interviews first: pick the one that matches your gap\")\n\n## Who Marvin is right for\n\nMarvin fits best when:\n\n![A research team sorting notes on a wall, the kind of shared backlog a research repository is meant to organize](https://images.unsplash.com/photo-1568219557405-376e23e4f7cf?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHwyfHxyZXNlYXJjaCUyMHRlYW0lMjBzdGlja3klMjBub3RlcyUyMHdhbGx8ZW58MXwwfHx8MTc5MjIxNjIxNHww&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Hugo Rocha](https://unsplash.com/@hugorrocha?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\n- You run a research team with a backlog of recordings and documents and want them searchable in one place.\n- Several teams need to search past research, and the repository is the main thing you are buying.\n- You need HIPAA compliance or EU data hosting and are ready for an annual, sales-led contract.\n- You want AI interviews as one part of a broader research operation.\n\nIt fits less well if you are a founder, PM or small team who mainly needs to run interviews now, if you want to see the price before talking to sales, or if you need month-to-month billing.\n\n## Marvin vs Koji\n\nKoji is built the other way round: interviews first. You describe what you want to learn, Koji drafts a research brief with questions, and you share a link. Participants talk to Koji's AI interviewer by voice or text, in 30+ languages, and it asks follow-up questions based on each answer. You can [preview the interview](/docs/ai-moderated-interviews) yourself before anyone else sees it.\n\n![Koji voice interview in progress with an AI orb asking a follow-up question and a microphone button](/email/shots/shot-voice-interview.jpg \"A Koji voice interview: the AI asks a follow-up based on what the participant just said\")\n\nEvery interview gets a quality score from 1 to 5. Only interviews scoring 3 or more use credits or go into the report ([how the quality gate works](/docs/how-the-quality-gate-works)), so you never pay for someone who clicked through. Koji then builds the report: themes, findings backed by quotes you can click through to, personas and a summary, updated as new interviews arrive ([generating research reports](/docs/generating-research-reports)).\n\nYou can reach your own customers with a shared link or [personalized invite links](/docs/personalized-interview-links), screen them with [screening questions](/docs/adding-screening-questions-to-your-study), or recruit from Koji's research panel on paid plans ([panel recruitment](/docs/panel-recruitment)). On the Interviews and Enterprise plans you can also query studies from Claude and other AI assistants through Koji's [MCP connector](/docs/mcp-overview).\n\n### Pricing side by side\n\n| | Marvin | Koji |\n|---|---|---|\n| Free plan | 5 uploads a month, limited AI interviews | 10 free credits on signup, no card |\n| Paid plans | Starter, Pro, Enterprise: quote only | Insights €29/month (29 credits), Interviews €79/month (79 credits), Enterprise |\n| Historical list price | $3,000 to $6,000+ a year for 5 users | Not applicable: prices are public |\n| Billing | Annual, via sales | Monthly or annual, self-serve |\n| What you pay for | Seats and plan features | Completed interviews: 1 credit text, 3 voice, €1 per extra credit |\n| Low-quality interviews | Not specified | Free: only interviews scoring 3+ use credits |\n\nOn Koji's Interviews plan, €79 a month covers 79 text interviews or 26 voice interviews with the reports included.\n\n## How to decide\n\n![Checklist of questions to ask before buying interview analysis software, from who runs the calls to cost per completed interview](/visuals/interview-analysis-software-buying-checklist/interview-analysis-software-buying-checklist.png \"Answer these before the sales call so the quote matches your real use\")\n\nChoose Marvin if your biggest gap is organizing research you already have, across many teams and file types, and you are buying an annual platform. Choose Koji if your biggest gap is talking to enough customers, quickly, at a price you can see today. Many teams start with interviews and add a repository later, once there is a steady flow of research worth storing.\n\n## Run your first study free\n\n[Start a Koji study free](/signup) with 10 credits and no card. Type one research question, preview the interview, and share the link with a handful of customers. You will have a report to read before a sales call with any vendor would even be booked.\n","category":"Comparisons","lastModified":"2026-10-29T13:18:30.777083+00:00","metaTitle":"Marvin Pricing 2026: Free, Starter, Pro and Enterprise","metaDescription":"Marvin (HeyMarvin) has a free plan with 5 uploads a month; Starter, Pro and Enterprise are quote-only. See past list prices, AI interview limits and alternatives.","keywords":["marvin pricing","heymarvin pricing","marvin cost","marvin free plan","marvin ai interviewer","marvin alternative"],"aiSummary":"Marvin (HeyMarvin) has four plans in 2026: Free, Starter, Pro and Enterprise. Free costs nothing, needs no card, and includes 5 file uploads a month and limited AI-moderated interviews (up to 5). Starter, Pro and Enterprise are quote-only; review sites last listed Essentials from $3,000 a year and Standard from $6,000 a year, each for a five-user minimum. Marvin is a research repository and analysis platform with AI tagging, Ask AI search, AI-moderated interviews in 40+ languages and Live Intercept (September 2026), with SOC 2 Type II, HIPAA and GDPR; G2 rating 4.8 from about 133 reviews. Koji is an interview-first alternative with public pricing: free with 10 credits, Insights EUR 29/month, Interviews EUR 79/month, 1 credit per text and 3 per voice interview, only interviews scoring 3+ of 5 use credits, AI voice and text interviews in 30+ languages with automatic reports.","aiKeywords":["marvin","heymarvin","marvin pricing","research repository","ai moderated interviews","qualitative research software","koji"],"aiContentType":"comparison","faqItems":[{"answer":"Marvin's Free plan costs nothing. Starter, Pro and Enterprise are quote-only. Review sites last listed paid tiers at about $3,000 a year (Essentials) and $6,000 a year (Standard) for a minimum of five users, roughly $50 and $100 per user per month.","question":"How much does Marvin cost?"},{"answer":"Yes. Marvin's Free plan needs no credit card and includes 5 file uploads a month and its AI Moderated Interviewer, with up to 5 AI interviews according to its help center.","question":"Does Marvin have a free plan?"},{"answer":"Marvin's help center says paid customers currently get unlimited AI interviews and that the AI interviewer is open to all plans for a limited period. Ask how AI interviews will be priced at renewal.","question":"Does Marvin charge per AI interview?"},{"answer":"Marvin's paid plans are sold through sales, and a competitor reports there is no month-to-month option. Plan for an annual contract and confirm the terms in your quote.","question":"Does Marvin offer monthly billing?"},{"answer":"Koji publishes all of its self-serve prices: free with 10 credits, Insights at €29 a month and Interviews at €79 a month. Its AI interviewer runs voice and text interviews and writes the report, at 1 credit per text and 3 per voice interview, and only interviews scoring 3 or more out of 5 use credits.","question":"What is a Marvin alternative with public pricing?"}],"relatedTopics":["Marvin","HeyMarvin pricing","research repository","AI-moderated interviews","research tool pricing"]},{"type":"blog","id":"12c8ea39-5b82-4e93-9587-7e0d243c4c65","slug":"lyssna-vs-optimal-workshop-2026","title":"Lyssna vs Optimal Workshop (2026): Which UX Testing Tool Fits Your Team?","url":"https://www.koji.so/blog/lyssna-vs-optimal-workshop-2026","summary":"Lyssna and Optimal Workshop are unmoderated UX testing tools with overlapping methods. Optimal Workshop (now branded Optimal) specialises in information architecture with OptimalSort card sorting, Treejack tree testing and Chalkmark first-click testing, plus surveys, prototype and live site testing, Qualitative Insights (formerly Reframer) and an AI Interviews tool in 13 languages. Its Starter plan is $199 a month billed annually with 5 launched studies a year, unlimited seats and responses; there is no free plan, only a 7-day trial. Lyssna (formerly UsabilityHub) focuses on fast design tests including five-second and preference tests, alongside card sorting, tree testing, first-click, prototype tests and surveys; it has a free plan, a Growth plan at about $165 a month billed annually or $199 monthly, and a 690,000+ person panel priced at about 1 credit (roughly US$1) per test minute per participant. Five users find about 85% of usability problems, but card sorts need about 15 users to reach a 0.90 correlation (Nielsen Norman Group). Maze's 2026 report found 69% of research teams use AI and 21% are experimenting with AI-moderated interviews. Koji is an AI interviewer that adds the why behind test results, with voice and text interviews in 30+ languages, six structured question types, 10 free credits at signup and text interviews at 1 credit and voice at 3 credits (EUR 1 per credit).","content":"## The short answer\n\n**Optimal Workshop is the specialist for information architecture research; Lyssna is the broader, cheaper-to-start tool for quick unmoderated tests.** Both run card sorts, tree tests, first-click tests, surveys and prototype tests. Optimal goes deeper on the IA methods and sells an annual plan priced by studies a year. Lyssna adds five-second and preference tests, has a free plan, bills monthly if you want, and comes with its own participant panel.\n\nPick **Optimal Workshop** if you are restructuring navigation or a large site and need the strongest card sorting and tree testing analysis. Pick **Lyssna** if you run lots of small design tests and want to start free. Neither will tell you _why_ customers behave the way they do in their own words at scale; for that you need interviews, which we cover at the end.\n\n## Lyssna vs Optimal Workshop at a glance\n\n|                | Lyssna                                                                                                | Optimal Workshop                                                                                               |\n| -------------- | ----------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------- |\n| Best known for | Fast unmoderated design tests                                                                         | Card sorting, tree testing and first-click                                                                     |\n| Methods | Five-second, preference, first-click, card sort, tree test, prototype, surveys | Card sort, tree test, first-click, surveys, prototype, qualitative analysis, AI interviews |\n| Free option    | Free plan                                                                                             | 7-day trial, no free plan                                                                                      |\n| Billing        | Monthly or annual                                                                                     | Annual only                                                                                                    |\n| Plan limits    | Studies per month                                                                                     | Launched studies per year                                                                                      |\n| Participants   | Own panel of 690,000+, priced per test minute                                                         | Recruitment priced in credits                                                                                  |\n\n![Table comparing Lyssna, Optimal Workshop and Koji on what each is best at, methods, free option, billing, limits and participants](/visuals/lyssna-vs-optimal-workshop-vs-koji-comparison-table/lyssna-vs-optimal-workshop-vs-koji-comparison-table.png \"Two testing tools and an AI interviewer, compared on what you would actually use them for\")\n\n## What Optimal Workshop does best\n\nOptimal Workshop (now branded Optimal) built its name on three tools: **OptimalSort** for card sorting, **Treejack** for tree testing and **Chalkmark** for first-click testing with heatmaps. If you are deciding how to label and group a navigation menu, these are still the tools many IA specialists reach for. It also has surveys, a usability testing tool for Figma prototypes and live websites, and **Qualitative Insights** (formerly Reframer) for tagging and analysing notes.\n\nIn the past year it added AI-simplified survey questions and an **AI Interviews** tool that finds themes and links each insight to video, in 13 languages ([Optimal](https://www.optimalworkshop.com/blog/optimal-interviews-launch)).\n\n**Pricing.** The Starter plan is **$199 a month, billed annually**, with **5 launched studies a year**, unlimited seats and unlimited responses per study. You can buy extra study bundles. Enterprise is priced on request. There is no free plan, only a 7-day trial ([Optimal](https://www.optimalworkshop.com/pricing)). Recruitment is paid in credits; the per-credit price isn't published.\n\n**The catch.** Five studies a year is tight for a team that tests every sprint, and reviewers describe a steeper learning curve than lighter tools. It is built for planned IA projects, not quick checks.\n\n## What Lyssna does best\n\nLyssna, formerly UsabilityHub, is the quick-test tool. Its **five-second tests** and **preference tests** answer the questions designers ask every week: what do people remember from this page, and which of these two versions do they prefer? It covers card sorting, tree testing, first-click, prototype tests and surveys too, plus live website testing and moderated interviews with recordings on paid plans.\n\nIts own **panel of 690,000+ people** with 395+ targeting options is a big part of the appeal ([Lyssna](https://www.lyssna.com/features/research-panel/)). Unmoderated panel responses cost about **1 credit per test minute per participant**, with a credit worth roughly US$1 and cheaper in bulk ([Lyssna help](https://help.lyssna.com/en/articles/4953377-plan-and-pricing-overview)).\n\nRecent additions include AI follow-up questions, AI summaries, a synthesis feature and an MCP server in beta for querying research from AI assistants ([Lyssna](https://www.lyssna.com/blog/2025-lyssna-year-in-review/)). As of mid 2026, Lyssna does not run AI-moderated interviews.\n\n**Pricing.** There is a **Free** plan with limited self-recruited responses. **Growth** is about **$165 a month billed annually** (Lyssna's comparison page says \"from $166\") or **$199 billed monthly**, with five seats and a monthly study allowance. Enterprise is custom. Panel responses are billed separately on every plan. Check [Lyssna's pricing page](https://www.lyssna.com/pricing/) for current limits.\n\n**The catch.** Studies-per-month limits on Growth can bite, and the IA analysis is lighter than Optimal's.\n\n## How many participants each method needs\n\nWhichever tool you use, sample size drives your cost, and it differs a lot by method:\n\n- **Usability testing:** Nielsen and Landauer's model says **five users find about 85% of usability problems** in a qualitative test ([Nielsen Norman Group](https://www.nngroup.com/articles/why-you-only-need-to-test-with-5-users/)).\n- **Card sorting:** results from 5 users correlate only **0.75** with the final structure; you need about **15 users to reach 0.90** (Tullis and Wood, via [Nielsen Norman Group](https://www.nngroup.com/articles/card-sorting-how-many-users-to-test/)).\n- **Tree testing:** NN/g treats task success **below 40% as poor** and **above 90% as excellent**, which means you need enough participants per task to tell those apart ([Nielsen Norman Group](https://www.nngroup.com/articles/interpreting-tree-test-results/)).\n\nThat is why panel pricing matters as much as the plan price. A card sort with 15 recruited people on Lyssna costs a known amount per minute; on Optimal, ask for the credit price before you plan the study. Our guides to [card sorting](/docs/card-sorting-guide), [tree testing](/docs/tree-testing-guide) and [first-click testing](/docs/first-click-testing-guide) cover how to set each one up.\n\nIf you're not sure which method fits your question, start here:\n\n![Listicle matching six UX research methods to the question each answers, from card sorting and tree testing to AI interviews](/visuals/ux-research-methods-which-question-each-answers/ux-research-methods-which-question-each-answers.png \"Pick the method from the question, then pick the tool\")\n\n## What both tools leave out\n\nLyssna and Optimal are built to measure _what_ people do with a design: where they click, where they get lost, which version they choose. They are less suited to the question that usually comes next: _why_.\n\n![A designer arranging sticky notes on a wall, sorting the questions a click test left unanswered](https://images.unsplash.com/photo-1580934174026-8142803ebb5b?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHwxfHxkZXNpZ25lciUyMHN0aWNreSUyMG5vdGVzJTIwd2FsbCUyMHBsYW5uaW5nfGVufDF8MHx8fDE3OTIzMDI2MTR8MA&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Paper Textures](https://unsplash.com/@inthemakingstudio?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\nMaze's 2026 research report found **69% of research teams use AI in at least some projects**, up from 50% a year earlier, and **21% are experimenting with AI-moderated interviews** ([Maze](https://maze.co/resources/user-research-report/)). The reason is simple. A tree test might show you that a third of people looked for billing under Account. It can't tell you what they expected to find there, or what they tried before they gave up.\n\n## Which one should you pick?\n\n**Pick Optimal Workshop if** you are doing serious IA work (a site restructure, a new navigation, a large content migration), you can plan around a yearly study allowance, and you want the deepest card sort and tree test analysis.\n\n**Pick Lyssna if** you test designs every week, want five-second and preference tests, like the option to pay monthly, and want a large built-in panel with per-minute pricing.\n\n**Add interviews if** your tests keep raising \"why\" questions that a click map can't answer.\n\n## Where Koji fits: the \"why\" behind the clicks\n\nKoji is an AI interviewer, not a testing tool. It doesn't run card sorts or tree tests. What it does is talk to people: each participant has a voice or text conversation with an AI that asks your questions and follows up when an answer is vague. Pair it with either tool above. Run the tree test, then interview the people who failed the task.\n\n![Koji question editor showing open ended, 1 to 10 scale and single choice questions with follow-up settings](/email/shots/shot-question-types.jpg \"Open questions, scales and choices sit in the same AI interview\")\n\n- **Structured and open questions together.** Six question types (open ended, scale, single choice, multiple choice, ranking and yes or no) in one conversation, so you get charts and quotes from the same study. See [AI interviews vs surveys](/docs/ai-interviews-vs-surveys).\n- **Recruit or bring your own.** Share a link with your users, or use [panel recruitment](/docs/panel-recruitment) on a paid plan, with the cost per respondent shown before you launch.\n- **30+ languages**, voice or text, with no moderator to schedule.\n- **A quality gate.** Each interview is scored 1 to 5, and only interviews scoring 3 or more use credits or enter the report. See [how the quality gate works](/docs/how-the-quality-gate-works).\n- **Reports with evidence.** Themes and findings, each [traced to the quotes behind it](/docs/tracing-report-findings-to-evidence).\n\nPricing is public. You get 10 free credits at signup with no credit card. A text interview costs 1 credit and a voice interview 3 credits, at €1 a credit. Insights is €29 a month with 29 credits, Interviews is €79 a month with 79 credits, and Enterprise is custom.\n\n## The bottom line\n\nOptimal Workshop is the IA specialist with an annual, studies-capped plan. Lyssna is the quick-test generalist with a free plan, monthly billing and a big panel. Most teams can decide in five minutes: big IA project, Optimal; weekly design checks, Lyssna.\n\nWhen the clicks raise questions, [try Koji free](https://www.koji.so) and ask the people behind them. For more comparisons, see [Koji vs Lyssna](/blog/koji-vs-lyssna-2026), [Koji vs Optimal Workshop](/blog/koji-vs-optimal-workshop-2026), [Maze vs Lyssna](/blog/maze-vs-lyssna-2026) and the [best card sorting tools](/blog/best-card-sorting-tools-2026).\n","category":"Comparisons","lastModified":"2026-10-29T13:18:30.489359+00:00","metaTitle":"Lyssna vs Optimal Workshop (2026): Features and Pricing","metaDescription":"Lyssna vs Optimal Workshop: card sorting, tree testing, quick design tests, panels and pricing compared, plus how to get the why behind clicks.","keywords":["lyssna vs optimal workshop","optimal workshop vs lyssna","lyssna pricing","optimal workshop pricing","card sorting tools","tree testing tools","lyssna alternative","optimal workshop alternative"],"aiSummary":"Lyssna and Optimal Workshop are unmoderated UX testing tools with overlapping methods. Optimal Workshop (now branded Optimal) specialises in information architecture with OptimalSort card sorting, Treejack tree testing and Chalkmark first-click testing, plus surveys, prototype and live site testing, Qualitative Insights (formerly Reframer) and an AI Interviews tool in 13 languages. Its Starter plan is $199 a month billed annually with 5 launched studies a year, unlimited seats and responses; there is no free plan, only a 7-day trial. Lyssna (formerly UsabilityHub) focuses on fast design tests including five-second and preference tests, alongside card sorting, tree testing, first-click, prototype tests and surveys; it has a free plan, a Growth plan at about $165 a month billed annually or $199 monthly, and a 690,000+ person panel priced at about 1 credit (roughly US$1) per test minute per participant. Five users find about 85% of usability problems, but card sorts need about 15 users to reach a 0.90 correlation (Nielsen Norman Group). Maze's 2026 report found 69% of research teams use AI and 21% are experimenting with AI-moderated interviews. Koji is an AI interviewer that adds the why behind test results, with voice and text interviews in 30+ languages, six structured question types, 10 free credits at signup and text interviews at 1 credit and voice at 3 credits (EUR 1 per credit).","aiKeywords":["Lyssna","Optimal Workshop","card sorting","tree testing","unmoderated usability testing","UX research tools"],"aiContentType":"comparison","faqItems":[{"answer":"Optimal Workshop specialises in information architecture research, with deep card sorting, tree testing and first-click analysis, sold on an annual plan limited by studies per year. Lyssna is a broader quick-test tool that adds five-second and preference tests, offers a free plan and monthly billing, and includes its own participant panel. Both also run surveys and prototype tests.","question":"What is the difference between Lyssna and Optimal Workshop?"},{"answer":"Optimal Workshop's Starter plan is $199 a month, billed annually, with 5 launched studies a year, unlimited seats and unlimited responses per study. Extra study bundles and an Enterprise plan are priced on request. There is no free plan, only a 7-day trial, and recruitment is paid in credits whose price isn't published.","question":"How much does Optimal Workshop cost?"},{"answer":"Lyssna has a free plan with limited self-recruited responses. Its Growth plan is about $165 a month billed annually, or $199 billed monthly, and Enterprise is custom. Panel responses are billed separately at about 1 credit per test minute per participant, with a credit worth roughly US$1.","question":"How much does Lyssna cost?"},{"answer":"Optimal Workshop is the specialist: OptimalSort and Treejack are built for information architecture work and offer deeper analysis. Lyssna covers both methods well enough for occasional studies. For reliable card sort results, plan on about 15 participants, which gets a 0.90 correlation with the final structure according to Nielsen Norman Group.","question":"Which is better for card sorting and tree testing?"},{"answer":"As of mid 2026, Lyssna runs moderated interviews with recordings but not AI-moderated interviews. Optimal Workshop launched an AI Interviews tool that finds themes and links insights to video. Koji is built around AI-moderated voice and text interviews, in 30+ languages, with 10 free credits at signup and no credit card required.","question":"Do Lyssna or Optimal Workshop run AI-moderated interviews?"}],"relatedTopics":["Lyssna","Optimal Workshop","card sorting","tree testing","UX testing tools"]},{"type":"blog","id":"d2e6ba95-a2be-418d-b8fa-d6db2a33c51f","slug":"looppanel-pricing-2026","title":"Looppanel Pricing in 2026: Pro, Enterprise and the Real Cost per Interview","url":"https://www.koji.so/blog/looppanel-pricing-2026","summary":"Looppanel has two plans in 2026. Pro is listed at $4,200 a year (about $350 a month billed annually; third parties report about $395 month to month) with 360 transcription hours, 120 uploads and 5 editors, extra editors $75 a month up to 10. Enterprise is a custom quote adding 10 editors, custom hours, SSO, PII redaction and auto data deletion. Older $30, $300 and $1,000 monthly tiers are discontinued. Looppanel records and analyzes Zoom, Meet and Teams calls with transcripts in 17 languages, but does not recruit or moderate, so recruiting and researcher time sit outside the price. Koji is an AI interviewer alternative: it runs voice and text interviews in 30+ languages and writes the report, free to start with 10 credits, Insights EUR 29/month, Interviews EUR 79/month, 1 credit per text and 3 per voice interview, and only interviews scoring 3+ of 5 use credits.","content":"Looppanel's current pricing has two plans. **Pro is listed at $4,200 a year** (about $350 a month, billed annually) with 360 transcription hours, 120 uploads and 5 editors. **Enterprise is a custom quote** that adds more hours, 10 editors and security extras such as SSO, PII redaction and automatic data deletion. Third-party trackers also report a month-to-month Pro price of about $395. There is no permanent free plan, and the old $30-a-month solo tier is no longer on the pricing page.\n\nThat is the short answer. The longer one is about what you are paying for. Looppanel analyzes interviews you have already run. It does not recruit participants or moderate interviews, so the subscription is only one part of what a study really costs. This guide covers each plan, what the limits mean in practice, and when an AI interviewer like Koji is the better buy.\n\n## Looppanel plans and prices in 2026\n\n| | Pro | Enterprise |\n|---|---|---|\n| Price | $4,200/year (about $350/month billed annually) | Custom quote |\n| Monthly billing | About $395/month (reported by third parties) | Custom |\n| Transcription | 360 hours a year (30 a month) | Custom |\n| Uploads | 120 a year | Custom |\n| Editors | 5 included, extra editors at $75/month up to 10 | 10 included |\n| Security | SOC 2 Type II, GDPR, encryption, role-based access | Adds custom SSO, PII redaction, auto data deletion |\n\nSources: Looppanel's pricing and security pages as indexed in October 2026, plus pricing trackers [xpay](https://www.xpay.sh/saas-pricing/looppanel-software/) and [ColdIQ](https://coldiq.com/tools/looppanel). Looppanel's own comparison table lists Pro editors in two different ways, so confirm the seat count with sales before you sign.\n\n### What changed from older Looppanel pricing\n\nIf you have seen Looppanel pricing quoted at $30 a month, that is out of date. Aggregators such as [Toolify](https://www.toolify.ai/tool/looppanel) and [TrustRadius](https://www.trustradius.com/products/looppanel/pricing) still list earlier tiers: a Starter or Solo plan at about $30 a month with 10 transcription hours, a Teams plan at around $300 to $350 a month and a Business plan at $1,000 a month. The current page starts at Pro, so a solo researcher now pays the same entry price as a five-person team.\n\n### Is there a free trial?\n\nSources disagree. ColdIQ reports a one-month free trial on both plans, G2 and TrustRadius mention a trial without saying how long it lasts, and other trackers say a trial is only available on request. Some [G2 reviewers](https://www.g2.com/products/looppanel/reviews) say the trial was too limited to explore the product properly. Ask for the trial terms in writing before you book a demo.\n\n## What Looppanel does well\n\nLooppanel is a focused analysis tool for researchers who run their own calls. It joins Zoom, Google Meet and Microsoft Teams calls to record them, or takes uploaded recordings. It then produces transcripts in 17 languages, automatic notes mapped to your discussion guide, tags and themes, video clips and a searchable repository.\n\nReviewers like how quickly it goes from a recording to a first synthesis. On G2 it holds 4.6 out of 5 from 26 reviews, and several reviewers call it a cheaper alternative to Dovetail. The complaints are mostly about customization and search, plus the trial limits mentioned above.\n\nThat speed matters. In Lyssna's [2025 Research Synthesis Report](https://www.lyssna.com/reports/research-synthesis/), 65.3% of researchers said synthesis takes them one to five days per project, and 59% named reading through the data as their biggest time sink. A tool that cuts that reading time pays for itself quickly if you run calls every week.\n\n## What Looppanel pricing does not cover\n\nThe subscription covers analysis. Three costs sit outside it:\n\n1. **Recruiting.** Looppanel's own blog says you \"cannot recruit users or build participant profiles\" in the product. You still need a panel, a recruiting service or your own customer list, plus incentives.\n2. **Moderation time.** Every hour Looppanel transcribes is an hour someone on your team spent running the call, plus scheduling, no-shows and prep.\n3. **Seats beyond five.** Extra editors are $75 a month each, which adds $900 a year per person on top of the $4,200.\n\nSo the real question is not \"is $4,200 a lot?\" It is how many interviews your team can run in a year, and what each one costs once you add the people.\n\n![Comparison of a recording analysis tool and an AI interviewer: who runs the call, who recruits, and when you get results](/visuals/recording-analysis-tool-vs-ai-interviewer-comparison/recording-analysis-tool-vs-ai-interviewer-comparison.png \"Decide which job you are buying software for before you compare prices\")\n\n### The cost per interview, worked out\n\nTake Pro at $4,200 a year with 360 transcription hours. If your interviews run 45 minutes, that is up to 480 interviews a year, or about $8.75 of software per interview. Few teams use every hour, though. A team running 10 interviews a month (120 a year) pays $35 per interview in software alone, before recruiting, incentives or researcher time.\n\nThe bottleneck is usually people, not software. In Maze's [Future of User Research Report 2025](https://maze.co/resources/user-research-report-2025/), 63% of product professionals named time and bandwidth as their biggest research challenge. Looppanel speeds up the analysis of each call, but somebody still has to hold the call.\n\n## Who Looppanel is right for\n\nLooppanel is a good fit when:\n\n![A researcher on a video call at a laptop, running the kind of weekly customer interview Looppanel is built to analyze](https://images.unsplash.com/photo-1713946598521-3a196fed3643?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHw1fHxyZXNlYXJjaGVyJTIwdmlkZW8lMjBjYWxsJTIwbGFwdG9wJTIwaGVhZHBob25lc3xlbnwxfDB8fHwxNzkyMjE2MjEyfDA&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Vitaly Gariev](https://unsplash.com/@silverkblack?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\n- You have researchers who moderate interviews every week and want faster notes and synthesis.\n- Your interviews already happen in Zoom, Meet or Teams, and you want them recorded and searchable in one place.\n- You need SOC 2 Type II and GDPR compliance, and can pay for Enterprise if you need SSO or PII redaction.\n- Five editors covers your team, or you are ready to budget for more.\n\nIt is a weaker fit if you are a solo founder or PM (the entry price is now a team price), if you want surveys or quantitative work alongside interviews, or if you do not have time to run the calls yourself.\n\n## Looppanel vs an AI interviewer\n\nThe industry is moving toward AI at both ends of the interview. In User Interviews' [State of User Research 2025](https://www.userinterviews.com/state-of-user-research-report), 80% of researchers said they use AI in their work, up 24 points in a year. The same survey found 91% worry about accuracy and hallucinations. That worry is fair, and it is why traceability (every finding linked back to what a participant actually said) matters more than any feature list.\n\nLooppanel puts AI on the analysis side. An AI interviewer like Koji puts it on both sides: it runs the interview and then analyzes it.\n\n### How Koji works\n\nWith Koji you describe what you want to learn, and Koji drafts a research brief with interview questions. You can [preview the interview](/docs/ai-moderated-interviews) yourself, then share a link. Participants talk to an AI interviewer by voice or text, in 30+ languages, whenever suits them. The interviewer asks follow-up questions based on what each person says, so you get the depth of a moderated call without booking one.\n\n![Koji research report with participant, duration, themes and quality stats above a summary of findings](/email/shots/shot-report.jpg \"A Koji report: themes, quality stats and a written summary, built as interviews come in\")\n\nEach interview gets a quality score from 1 to 5, and only interviews scoring 3 or more use credits or go into the report ([how the quality gate works](/docs/how-the-quality-gate-works)). Koji then writes the report: themes, findings backed by supporting quotes, personas and a summary ([generating research reports](/docs/generating-research-reports)).\n\nYou can bring your own participants with a link or [personalized invite links](/docs/personalized-interview-links), add [screening questions](/docs/adding-screening-questions-to-your-study), or recruit from Koji's research panel on paid plans ([panel recruitment](/docs/panel-recruitment)).\n\n### Koji pricing next to Looppanel\n\n| | Looppanel Pro | Koji |\n|---|---|---|\n| Entry price | $4,200/year | Free to start with 10 credits, no card |\n| Paid plans | Pro or custom Enterprise | Insights €29/month (29 credits), Interviews €79/month (79 credits), Enterprise |\n| What you pay for | Transcription hours, uploads, editors | Completed interviews: 1 credit for text, 3 for voice, €1 per extra credit |\n| Who moderates | Your team | Koji's AI interviewer |\n| Recruiting | Not included | Your own list, links, or Koji's panel on paid plans |\n| Low-quality sessions | Count toward your hours | Free: only interviews scoring 3+ use credits |\n\nOn the Interviews plan, €79 a month covers 79 text interviews or 26 voice interviews, analyzed and reported, with no moderator time.\n\n## How to choose\n\n![Checklist of questions to ask before buying interview analysis software, from who runs the calls to cost per completed interview](/visuals/interview-analysis-software-buying-checklist/interview-analysis-software-buying-checklist.png \"Run through these before a demo so the pricing conversation is about your real volume\")\n\nPick Looppanel if your researchers already run plenty of calls and analysis is where the time goes. Pick an AI interviewer if the calls themselves are the bottleneck, if you need answers in days rather than weeks, or if you want to hear from far more people than your team can schedule. Plenty of teams use both: human-moderated sessions for the hardest conversations, AI interviews for breadth.\n\n## Try Koji free\n\nIf your problem is too few interviews rather than slow analysis, [start a Koji study free](/signup) with 10 credits and no card. Write one research question, preview the interview, and share the link. Your first report can be ready the same day the answers come in.\n","category":"Comparisons","lastModified":"2026-10-29T13:18:30.157881+00:00","metaTitle":"Looppanel Pricing 2026: Pro, Enterprise and Limits","metaDescription":"Looppanel Pro is listed at $4,200 a year with 360 transcription hours and 5 editors; Enterprise is custom. What the limits mean and what each interview really costs.","keywords":["looppanel pricing","looppanel cost","looppanel plans","looppanel free trial","looppanel pro","looppanel alternative"],"aiSummary":"Looppanel has two plans in 2026. Pro is listed at $4,200 a year (about $350 a month billed annually; third parties report about $395 month to month) with 360 transcription hours, 120 uploads and 5 editors, extra editors $75 a month up to 10. Enterprise is a custom quote adding 10 editors, custom hours, SSO, PII redaction and auto data deletion. Older $30, $300 and $1,000 monthly tiers are discontinued. Looppanel records and analyzes Zoom, Meet and Teams calls with transcripts in 17 languages, but does not recruit or moderate, so recruiting and researcher time sit outside the price. Koji is an AI interviewer alternative: it runs voice and text interviews in 30+ languages and writes the report, free to start with 10 credits, Insights EUR 29/month, Interviews EUR 79/month, 1 credit per text and 3 per voice interview, and only interviews scoring 3+ of 5 use credits.","aiKeywords":["looppanel","looppanel pricing","interview analysis software","ai transcription","research repository","ai interviewer","koji"],"aiContentType":"comparison","faqItems":[{"answer":"Looppanel Pro is listed at $4,200 a year, about $350 a month billed annually, with 360 transcription hours, 120 uploads and 5 editors. Third-party trackers report about $395 a month on monthly billing. Enterprise is a custom quote.","question":"How much does Looppanel cost?"},{"answer":"No. The current pricing page has no permanent free plan. Sources disagree on the trial: some report a one-month free trial, others say a trial is only available on request, so ask Looppanel for the terms.","question":"Does Looppanel have a free plan?"},{"answer":"Pro includes 5 editors, and you can add more at $75 a month each up to 10. Enterprise includes 10 editors. Looppanel's comparison table describes Pro seats in two different ways, so confirm the count with sales.","question":"How many editors does Looppanel Pro include?"},{"answer":"No. Looppanel records, transcribes and analyzes calls your team runs in Zoom, Google Meet or Microsoft Teams, or recordings you upload. Recruiting, incentives and moderation are separate costs.","question":"Does Looppanel recruit participants or run interviews?"},{"answer":"Koji starts free with 10 credits and paid plans cost €29 or €79 a month. Instead of analyzing calls you ran yourself, Koji's AI interviewer runs voice and text interviews and writes the report. A text interview uses 1 credit and a voice interview 3, and only interviews scoring 3 or more out of 5 use credits.","question":"What is a cheaper alternative to Looppanel?"}],"relatedTopics":["Looppanel","interview analysis software","research tool pricing","AI interviews","user research software"]},{"type":"blog","id":"3070367d-6650-49f7-8310-704a500f88ab","slug":"listen-labs-vs-outset-2026","title":"Listen Labs vs Outset (2026): Which AI Interview Platform Should You Buy?","url":"https://www.koji.so/blog/listen-labs-vs-outset-2026","summary":"Listen Labs and Outset are both enterprise-priced AI-moderated interview platforms at roughly $20K/year entry. Listen Labs bundles a 30M-person recruitment panel across 100+ languages; Outset is bring-your-own-participants with text, voice, and video in 40+ languages plus usage-based billing. Neither offers self-serve access or structured quantitative question types. Koji offers the same AI-moderated voice and chat interview workflow with six structured question types, automatic thematic analysis, and quality-gated credits, self-serve as low as EUR 1 per qualified interview and EUR 3 per qualified voice interview, pay as you go with no contract.","content":"\n# Listen Labs vs Outset (2026): Which AI Interview Platform Should You Buy?\n\n**Short answer:** Choose **Listen Labs** if you need recruitment and moderation bundled into one enterprise contract and your panel spend is already six figures. Choose **Outset** if you bring your own participants, want text, voice, and video moderation in 40+ languages, and prefer usage-based billing on top of a seat. Choose **[Koji](https://www.koji.so)** if you want the same AI-moderated interview workflow without a $20,000 entry ticket: self-serve, starting free with 10 credits, then as low as EUR 1 per qualified interview and EUR 3 per qualified voice interview, pay as you go with no subscription, with six structured question types in a single study.\n\nBoth Listen Labs and Outset are legitimate, well-funded platforms. Neither is a bad product. But they are both sold the same way — through an enterprise motion with a five-figure floor — and that shapes who can actually use them. This comparison is written for the buyer who has both tabs open and wants specifics rather than category marketing.\n\n---\n\n## Why this comparison matters in 2026\n\nAI-moderated interviewing stopped being experimental. Roughly **72% of insights buyers now use generative AI in at least one stage of a research project, up from 23% in 2023**, and adoption of AI-moderated interviews specifically runs at about **81% among dedicated research teams, 73% among UX teams (up from 38% in 2024), and 67% among product management teams**.\n\nThe economics explain the speed. Industry benchmarks put **AI-moderated qualitative interviews at roughly $8-$15 per completed interview versus $150-$300 for a human-moderated equivalent**. When a method gets 10-20x cheaper per unit, the constraint moves from \"can we afford to talk to customers\" to \"which platform do we buy.\" That is exactly the question Listen Labs and Outset are competing over — inside a market research industry projected at about **$150 billion in 2026**, where qualitative is growing around 7.2% CAGR and AI-native methods are the only segment growing double digits.\n\n---\n\n## Listen Labs vs Outset: side-by-side\n\n| | **Listen Labs** | **Outset** | **Koji** |\n|---|---|---|---|\n| Funding to date | $69M (Series B, Jan 2026) | $30M Series B (Radical Ventures, M12) | Bootstrapped, self-serve |\n| Interview modalities | Voice, video, text | Text, voice, video | Voice + chat |\n| Languages | 100+ | 40+ | 40+ |\n| Built-in participant panel | Yes — ~30M via Listen Atlas | No — bring your own | Bring your own, recruit from your product, or built-in panel recruitment quoted in credits |\n| Structured question types | Primarily open-ended | Primarily open-ended | 6 types in one study |\n| Automatic thematic analysis | Yes | Yes | Yes |\n| Highlight reels / reports | Yes | Yes | One-click reports + transcript chat |\n| Pricing model | ~$20K annual base + $300-400/session panel cost | ~$20K/seat/year + usage-based billing | From EUR 1 per qualified interview, EUR 3 voice; pay as you go, no subscription |\n| Free tier | No | No | Yes — 10 credits at signup |\n| Self-serve signup | No — sales-led | No — sales-led | Yes |\n| Annual contract | Typically required | Typically required | No: pay as you go, no subscription |\n| Quality-gated billing | No | No | Yes — only interviews scoring 3+ consume credits |\n| MCP / AI-client integration | No | No | Yes (Claude, Cursor) |\n\n---\n\n## What Listen Labs actually does well\n\nListen Labs is the most-funded name in the category, having raised **$69 million across three rounds, closing a Series B in January 2026**. Its core bet is *end-to-end*: study co-design, recruitment, moderation, and analysis inside one contract.\n\n**Its genuine strengths:**\n\n- **Recruitment is not your problem.** Listen Atlas provides a verified participant network of roughly 30 million people, with a recruitment ops layer that scopes hard-to-reach audiences for you. If you have spent three weeks trying to find 12 anesthesiologists who switched EHR vendors, this is worth real money.\n- **Language breadth.** 100+ languages for transcription and translation is the widest in the category — meaningful if you run simultaneous multi-market studies.\n- **Concurrency at scale.** The platform is built to run thousands of voice and video interviews in parallel, then compile them into reports, highlight reels, and presentation-ready decks.\n- **Executive familiarity.** Listen Labs has marketed aggressively. If your CMO has already heard the name, internal selling gets shorter.\n\n**The trade-off:** cost and access. Third-party buyer guides put entry at roughly **$20,000 annual base plus $300-400 per session in panel costs**, with services scoped on top. A modest 30-interview study can therefore land in the $9,000-$12,000 range in panel fees *alone*, before the platform fee. There is no free tier and no self-serve path.\n\n## What Outset actually does well\n\nOutset raised a **$30M Series B led by Radical Ventures with participation from M12 (Microsoft's venture arm)**, alongside 8VC and Y Combinator, after reporting more than **8x business growth**. In 2026 it repositioned beyond research toward \"AI-native customer experience management.\"\n\n**Its genuine strengths:**\n\n- **Three modalities, one study.** Text, voice, and video moderation across **40+ languages**, with adaptive personas that probe deeper as the conversation develops.\n- **Usage-based billing on top of seats.** If your volume is spiky, paying for questions asked during live interviews is more honest than a flat per-session panel fee.\n- **Highlight reels.** Customizable clip reels are genuinely effective for moving stakeholders who will never read a report.\n- **Security posture.** SOC 2, GDPR adherence, Azure hosting, and a \"never trained on your data\" policy — the things that clear a security review.\n- **Recruitment flexibility.** No included panel, but an integration with User Interviews if you need one. If you already have customers to talk to, you are not paying for a panel you will not use.\n\n**The trade-off:** Outset is bring-your-own-participants. Reported pricing sits around **$20,000 per seat per year** with usage billing layered on, so a three-person team is a $60K conversation before a single interview runs. Pricing is not published, and there is no standard free trial.\n\n---\n\n## The honest head-to-head\n\n**If recruitment is your bottleneck, Listen Labs wins.** The bundled 30M panel and recruitment ops layer solve the single hardest, least-automatable part of qualitative research. Outset simply does not do this natively.\n\n**If you already have customers, Outset wins on cost efficiency.** You are not subsidising a panel you do not need, and per-question usage billing scales down as well as up. For B2B teams with an existing customer base, paying Listen Labs' session fees to interview *your own users* makes no sense.\n\n**On moderation depth, they are close to even.** Both run adaptive AI moderators that follow up in real time. Both auto-synthesize themes with supporting quotes. Reviewers do not consistently separate them on interview quality, and you should not expect a decisive difference here.\n\n**On languages, Listen Labs wins** (100+ vs 40+) — but only matters if you genuinely run beyond 40 languages, which most teams do not.\n\n**On procurement, they are equally slow.** Both are sales-led, contract-first, and priced in the same ~$20K-and-up neighbourhood. Neither lets you try the product on a Tuesday afternoon and have findings by Thursday.\n\n---\n\n## The gap both platforms leave open\n\nHere is what the Listen Labs vs Outset framing hides: **both answer the same question — \"how do we serve an enterprise insights team with a budget?\"** Neither answers \"how does a founder, PM, or three-person research team run rigorous AI-moderated interviews next week?\"\n\nThat gap is where [Koji](https://www.koji.so) sits, and it is not a smaller version of the same thing — it is a different set of design decisions:\n\n- **[AI-moderated voice and chat interviews](/docs/ai-moderated-interviews)** that adapt in real time, powered by an ElevenLabs voice stack, running 24/7 with no moderator bias and no scheduling. See [AI voice interviews](/docs/ai-voice-interviews) and [voice vs text interviews](/docs/voice-vs-text-interviews).\n- **Six structured question types in a single study** — open-ended, scale, single choice, multiple choice, ranking, and yes/no. This is the concrete capability gap. Listen Labs and Outset are built around open-ended conversation; Koji lets you capture the qualitative \"why\" *and* quantifiable signal in one instrument, so you are not running a separate survey to get a number you can chart. See [open-ended questions](/docs/open-ended-questions-ai-interviews) and [multiple-choice questions](/docs/multiple-choice-questions-ai-interviews).\n- **[Automatic thematic analysis](/docs/thematic-analysis-guide)** with [AI auto-tagging](/docs/ai-auto-tagging-customer-interviews) across every transcript, plus [chat with your transcripts](/docs/chat-with-interview-transcripts-ai) — ask a question across 100 interviews, get a cited answer in seconds.\n- **Customizable AI consultants** you tune to your domain, brand voice, and research goals.\n- **Quality-gated credits.** Only conversations scoring 3+ on the quality rubric consume credits. Drop-offs and junk responses do not bill you — a protection neither competitor offers.\n- **Transparent pricing.** Interviews start as low as EUR 1 per qualified interview, EUR 3 for voice. Start with pay as you go: no subscription needed, and you pay only for the interviews your study actually uses. Ten free credits at signup, no annual contract, no sales call.\n- **Recruitment, either way.** You do not need your own audience: describe the one you want (market, demographics, role, screening) and approve a live per-respondent quote in credits before launch. If you do have people, [recruit from your own product](/docs/recruiting-from-your-product) or import your list. Either way you can run [GDPR-compliant research](/docs/gdpr-compliant-ai-user-research) on EU-transparent pricing.\n\n**Do the arithmetic.** A 30-voice-interview study on Koji starts at about EUR 90, at as low as EUR 3 per qualified voice interview. The same 30 sessions sourced through Listen Labs' panel, at reported $300-400 per session, is $9,000-$12,000 in panel fees before the ~$20K platform base. That is not a 20% difference. It is two different categories of decision.\n\n---\n\n## How to choose: a decision framework\n\n**Buy Listen Labs if** you need hard-to-reach participants recruited for you, you run studies in more than 40 languages, you have a six-figure research budget already approved, and a 60-90 day procurement cycle is normal at your company.\n\n**Buy Outset if** you already have participants, you want video moderation alongside voice and text, you need SOC 2 documentation for a security review, and per-seat plus usage billing fits how your team works.\n\n**Buy Koji if** you want to start this week, you want quantitative and qualitative signal in one study rather than two tools, you would rather not pay for a panel you do not need, and you want to see whether AI-moderated interviews work for your team before committing a budget line to them. For a wider field, see our [best AI interview software guide](/docs/best-ai-interview-software-2026) and [Listen Labs alternatives](/docs/listen-labs-alternatives-2026).\n\n---\n\n## Run your first AI-moderated study today\n\nYou do not need to pick between two $20,000 contracts to find out whether AI interviews work for your product.\n\n[Koji](https://www.koji.so) gives you 10 free credits at signup — enough to run real AI-moderated interviews with your own customers, get automatic thematic analysis, and generate a one-click report. From research question to synthesized insight in hours, not weeks. No research expertise required, no sales call, no annual commitment.\n\nIf AI-moderated research turns out to be central to how you build, you will know it from your own data instead of a vendor deck. [Start free with Koji](https://www.koji.so).\n\n---\n\n## Frequently asked questions\n","category":"Comparisons","lastModified":"2026-10-29T13:18:29.891775+00:00","metaTitle":"Listen Labs vs Outset (2026): AI Interview Platforms Compared","metaDescription":"Listen Labs vs Outset compared on pricing, recruitment, languages, and moderation. Listen Labs bundles a 30M panel at ~$20K/yr; Outset is BYO participants at ~$20K/seat. Plus the self-serve alternative both leave open.","keywords":["listen labs vs outset","listen labs vs outset ai","outset vs listen labs","ai interview platform comparison","listen labs pricing","outset pricing","ai moderated interview platforms","best ai research platform 2026"],"aiSummary":"Listen Labs and Outset are both enterprise-priced AI-moderated interview platforms at roughly $20K/year entry. Listen Labs bundles a 30M-person recruitment panel across 100+ languages; Outset is bring-your-own-participants with text, voice, and video in 40+ languages plus usage-based billing. Neither offers self-serve access or structured quantitative question types. Koji offers the same AI-moderated voice and chat interview workflow with six structured question types, automatic thematic analysis, and quality-gated credits, self-serve as low as EUR 1 per qualified interview and EUR 3 per qualified voice interview, pay as you go with no contract.","aiKeywords":["listen labs","outset","ai moderated interviews","ai interview platform","customer research platform","qualitative research automation","research platform pricing","koji"],"aiContentType":"comparison","faqItems":[{"answer":"They are priced in the same neighbourhood but bill differently. Buyer guides put Listen Labs at roughly a $20,000 annual base plus $300-400 per session in panel costs, while Outset is reported at roughly $20,000 per seat per year with usage-based billing on top and no included panel. If you already have your own participants, Outset is usually cheaper because you are not paying session-level panel fees. If recruitment is your bottleneck, Listen Labs can be cheaper overall than buying a panel separately. Koji is self-serve: interviews as low as EUR 1 per qualified interview and EUR 3 per qualified voice interview. Start free with 10 credits, no card, then pay as you go with no subscription needed.","question":"Is Listen Labs or Outset cheaper?"},{"answer":"No. Outset is bring-your-own-participants, with an integration to User Interviews if you need external recruitment. Listen Labs includes Listen Atlas, a verified network of roughly 30 million participants, plus a recruitment ops layer. Koji supports bringing your own list, recruiting from inside your product, or built-in panel recruitment: you describe the audience and approve a per-respondent quote paid in credits.","question":"Does Outset include a participant panel?"},{"answer":"Listen Labs supports 100+ languages for transcription and translation, the widest coverage in the category. Outset supports 40+ languages across text, voice, and video moderation. Koji supports 30+ languages for AI-moderated voice and chat interviews. For most teams the difference is academic — the gap only matters if you genuinely run studies in more than 40 markets.","question":"Which platform supports more languages?"},{"answer":"Neither offers standard self-serve signup or a public free tier. Both are sales-led with custom quotes and typically annual contracts, so evaluation runs through a demo and procurement rather than hands-on trial. Koji offers 10 free credits at signup with no sales call, so you can run real interviews with your own customers before committing.","question":"Can I try Listen Labs or Outset before buying?"},{"answer":"Listen Labs and Outset are both built primarily around open-ended AI-moderated conversation. If you need scaled or choice-based data you generally run a separate survey. Koji supports six structured question types in a single study — open-ended, scale, single choice, multiple choice, ranking, and yes/no — so you capture the qualitative why and quantifiable signal in one instrument.","question":"Do these platforms support quantitative questions as well as open-ended ones?"},{"answer":"Listen Labs has raised $69 million across three rounds, closing a Series B in January 2026. Outset raised a $30 million Series B led by Radical Ventures with participation from M12, Microsofts venture arm, alongside 8VC and Y Combinator, after reporting more than 8x business growth.","question":"How much did Listen Labs and Outset raise?"}],"relatedTopics":["AI interview platforms","competitor comparison","research platform pricing","AI moderated interviews","participant recruitment"]},{"type":"blog","id":"624c9eef-f7a9-4b2c-b917-f01976520512","slug":"listen-labs-vs-conveo-2026","title":"Listen Labs vs Conveo (2026): AI Interview Platforms After the Salesforce Deal","url":"https://www.koji.so/blog/listen-labs-vs-conveo-2026","summary":"Listen Labs and Conveo are AI-moderated interview platforms aimed at different enterprise buyers. Salesforce signed a definitive agreement to acquire Listen Labs on September 29, 2026, with closing expected in Salesforce's fiscal Q4 2027; a price of about $2 billion was reported but not confirmed. Listen Labs, based in San Francisco, raised about $96M to $100M, including a $69M Series B at about a $500M valuation, runs video and voice interviews with emotion analysis and says its panel layer reaches 30M+ people in 45+ countries; customers include Microsoft, Anthropic and Sweetgreen. Conveo, founded in Antwerp in April 2024 and backed by Y Combinator, raised a $50M Series A in September 2026 (about $55.8M total), runs video, voice and text interviews with non-verbal analysis, focuses on brand, concept and ad testing, and names Unilever, Orange, Sanofi and Google as customers. Neither publishes prices: Listen Labs is estimated at about a $20K annual base plus per-interview credits, and Capterra lists Conveo from US$45,000. Koji is a self-serve alternative with 10 free credits, text interviews at 1 credit and voice at 3 credits (EUR 1 per credit), 28 interview languages, panel recruitment on paid plans, a quality gate where only interviews scoring 3 or more of 5 use credits, and findings traced to quotes.","content":"## The short answer\n\n**Listen Labs and Conveo both run AI-moderated interviews at scale, but they sell to different buyers, and one of them is about to become part of Salesforce.** On September 29, 2026, Salesforce signed a definitive agreement to acquire Listen Labs, with the deal expected to close in Salesforce's fiscal fourth quarter of 2027, subject to regulatory clearance ([Salesforce](https://www.salesforce.com/news/stories/salesforce-signs-definitive-agreement-to-acquire-listen-labs/)). Four weeks earlier, Conveo raised a $50M Series A to expand in the US ([Pulse 2.0](https://pulse2.com/conveo-raises-50-million-series-a-to-expand-ai-consumer-intelligence-platform/)).\n\nPick **Listen Labs** if you want a large built-in panel, video interviews and a vendor that will soon sit inside the Salesforce stack. Pick **Conveo** if you run consumer and brand research, test concepts and ads, and want a European vendor with EU hosting. Pick **neither** if the people you most need to hear from are your own customers and you want to start this week without a sales call. Both are quote-based enterprise tools.\n\n## Where each company stands in October 2026\n\n|                   | Listen Labs                                                               | Conveo                                                           |\n| ----------------- | ------------------------------------------------------------------------- | ---------------------------------------------------------------- |\n| Based in          | San Francisco                                                             | Antwerp, Belgium                                                 |\n| Founded           | Platform launched April 2025                                              | April 2024, Y Combinator S24                                     |\n| Funding           | About $96M to $100M, including a $69M Series B at about a $500M valuation | About $55.8M, including a $50M Series A in September 2026        |\n| Ownership         | Being acquired by Salesforce (deal not yet closed)                        | Independent                                                      |\n| Interview formats | Video and voice, with emotion analysis                                    | Video, voice and text, with analysis of tone and non-verbal cues |\n| Participants      | Its own panel layer, which it says reaches 30M+ people in 45+ countries   | Panel partners, or your own list                                 |\n| Named customers   | Microsoft, Anthropic, Sweetgreen, Perplexity                              | Unilever, Orange, Sanofi, Google                                 |\n| How you buy       | Demo and pilot, credits priced by audience                                | Contact sales, credits by interview minutes                      |\n\nSources: Listen Labs funding and customers from [SiliconANGLE](https://siliconangle.com/2026/10/01/salesforce-to-acquire-ai-customer-research-startup-listen-labs-in-reported-2b-deal/) and [eWeek](https://www.eweek.com/fr/news/listen-labs-ai-startup-microsoft-customers/); Conveo founding and seed from [Tech.eu](https://tech.eu/2025/03/06/y-combinator-backed-conveo-raises-53m-seed-to-develop-ai-powered-research-coworker/), Series A and customer count from [Implicator](https://www.implicator.ai/conveo-lands-50-million-for-always-on-consumer-intelligence/). Panel and analysis claims are each company's own.\n\nSet against Koji, the self-serve option covered at the end of this post, the practical differences look like this:\n\n![Table comparing Listen Labs, Conveo and Koji on where they are based, status, interview formats, participants, analysis and how you buy](/visuals/listen-labs-vs-conveo-vs-koji-comparison-table/listen-labs-vs-conveo-vs-koji-comparison-table.png \"Three AI interview platforms side by side, from enterprise contracts to self-serve\")\n\n## What the Salesforce deal changes for Listen Labs buyers\n\nBusiness Insider first reported the talks on September 9 and put the price under discussion at about $2 billion. Neither company has confirmed a figure ([Channel Insider](https://www.channelinsider.com/channel-business/news-salesforce-listen-labs-acquisition-ai-customer-research/)). Salesforce says Listen Labs will add customer understanding to Marketing Cloud, Service Cloud and its wider AI portfolio. Before the deal, Listen Labs had signed a term sheet for a $125M Series C at a $1.5B valuation, which never closed ([Dealroom](https://dealroom.co/news/143721-listen-labs-targets-125m-series-c-at-1-5b-valuation/)).\n\n![A team in a meeting room talking through what a vendor's acquisition could mean for their research contract](https://images.unsplash.com/photo-1758518726324-62bef7c815b0?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHw2fHxidXNpbmVzcyUyMHRlYW0lMjBtZWV0aW5nJTIwcm9vbSUyMGxhcHRvcHN8ZW58MXwwfHx8MTc5MjMwMjYxMnww&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Vitaly Gariev](https://unsplash.com/@silverkblack?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\nFor a team that already runs Salesforce, this could be good news: interview findings that land next to the CRM record. For everyone else, it raises the questions any acquisition raises. Will the product stay sold on its own? Will pricing move toward Salesforce bundles? Will the roadmap follow Salesforce's priorities instead of researchers'? None of that is known yet, and anyone who tells you otherwise is guessing. If you sign now, ask for price protection and a clean data export clause in writing.\n\n## How the two products differ\n\n### Listen Labs: panel scale and fast turnaround\n\nListen Labs is built around AI agents that design a study, recruit people, interview them and write up the results, with the company claiming results in under 24 hours. It reports more than one million AI-led interviews run on the platform. Its pitch rests on reach: a panel layer it calls Listen Atlas, which it says matches across several panel partners, plus fraud checks that look at video, voice, content and device signals. It also offers simulated customer responses, often called digital twins, alongside real interviews.\n\nIts emotion analysis reads tone, word choice and facial expressions, and ties each label to a timestamp and quote. That traceability matters. An emotion score without the moment behind it is hard to defend in a meeting.\n\n### Conveo: brand and concept research, built in Europe\n\nConveo started in Antwerp in April 2024 and went through Y Combinator that summer. Its seed round of $5.3M closed in March 2025 with YC, 6 Degrees Capital and Pitchdrive among the investors. It runs asynchronous AI-moderated interviews, mostly video, with adaptive follow-up questions, and analyses speech, tone and non-verbal cues together.\n\nWhere Listen Labs leans toward tech and product buyers, Conveo's customer list leans toward consumer goods and brand teams: Unilever, Orange and Sanofi. Concept and ad testing are core use cases, and it can generate new concepts from earlier results for the next round of testing. It says more than 400 enterprises use it, including more than 50 from the Fortune 500. Since the Series A it has been moving toward StoryLines, a product for ongoing research instead of one-off studies.\n\n## What each one costs\n\nNeither company publishes a price list, so treat every number here as an estimate.\n\n**Listen Labs** is credit-based, with credits priced by how hard the audience is to reach. Third-party estimates put a typical contract at about a $20K annual base plus roughly $300 to $400 per completed interview. Listen Labs' own pricing article says most enterprise customers pay $85 to $150 per interview ([Listen Labs](https://listenlabs.ai/articles/ai-customer-research-pricing-2026)). Companies with more than 100 employees go through a demo and a pilot.\n\n**Conveo** sells credits based on interview minutes, with the per-minute rate falling as volume grows. Capterra lists a starting price of US$45,000 ([Capterra](https://www.capterra.co.uk/software/1077727/Conveo)).\n\nEither way, you are looking at a five-figure commitment and a procurement cycle before the first interview. Our [user research tool pricing breakdown](/blog/user-research-tool-pricing-2026) puts these numbers next to the rest of the market.\n\n![Checklist of questions to ask before signing an AI interview platform contract, covering billing, quality, recruitment, data and exit terms.](/visuals/ai-interview-platform-contract-questions-checklist/ai-interview-platform-contract-questions-checklist.png \"Get these answered in writing before you sign with either vendor\")\n\n## Why this category is getting so much money\n\nThe funding follows the way researchers already work. Qualtrics' 2026 market research trends report, a survey of more than 3,000 researchers, found that **53% use AI regularly** and nine in ten have tried it. It also found that **78% expect AI agents to run more than half of research projects by 2028** ([Research Live](https://www.research-live.com/article/news/most-researchers-using-ai-regularly-finds-qualtrics-study/id/5144776)). On the UX side, User Interviews' State of User Research 2025 found that **80% of researchers used AI in their work**, up 24 points in a year, while **91% worry about accuracy and hallucinations** ([User Interviews](https://www.userinterviews.com/state-of-user-research-report)).\n\nThat last number is the one to keep in mind when you evaluate any of these tools. The question is no longer whether AI can run an interview. It is whether you can check what it concluded. Ask every vendor to show you how a finding traces back to the quotes behind it. We cover the evidence on this in [are AI-moderated interviews reliable?](/blog/are-ai-moderated-interviews-reliable-2026)\n\n## Which one should you pick?\n\n**Pick Listen Labs if** you need consumer or prospect audiences at volume, want video and emotion analysis, and either already run Salesforce or are comfortable with the acquisition's unknowns.\n\n**Pick Conveo if** your work is brand, packaging, concept or ad testing for consumer markets, you want a vendor based in the EU, and your budget starts in the tens of thousands.\n\n**Pick neither if** most of your questions start with \"our customers\": why they churned, why trials went quiet, why a feature isn't used. Both platforms are built around buying audiences and scoping studies. That is the right model for market sizing, and an expensive one for weekly conversations with people who already use your product.\n\n## The self-serve option: Koji\n\nKoji runs the same kind of AI-moderated interview without the contract. You describe what you want to learn, Koji drafts the research brief, and you can [preview the interview yourself](/docs/previewing-your-interview) for free before anyone else sees it.\n\n![Koji home screen asking \"What do you want to research?\" with a question typed and a Start study button](/email/shots/shot-brief-from-one-message.jpg \"A study starts from one research question, not a scoping call\")\n\nWhat you get:\n\n- **Voice, text and phone interviews** in 30+ languages, with the AI asking follow-up questions when an answer is thin. See [how AI interviewers work](/docs/how-ai-interviewers-work).\n- **Structured questions inside the conversation.** Six question types (open ended, scale, single choice, multiple choice, ranking and yes or no) sit in the same interview, so one study gives you quotes and charts.\n- **Your own customers or a recruited audience.** Share one interview link with your customer list (on Interviews and Enterprise, a CSV import gives each person their own link), or use [panel recruitment](/docs/panel-recruitment) on a paid plan, with the cost per respondent shown before you launch. Add [screening questions](/docs/adding-screening-questions-to-your-study) to keep the wrong people out.\n- **A quality gate.** Every interview gets a score from 1 to 5, and only interviews scoring 3 or more use credits or enter the report. See [how the quality gate works](/docs/how-the-quality-gate-works).\n- **Reports you can check.** Themes, findings and personas, each [traced to the quotes behind it](/docs/tracing-report-findings-to-evidence).\n\nPricing is public. You start with 10 free credits and no credit card. A text interview costs 1 credit and a voice interview 3 credits, at €1 a credit. Insights is €29 a month with 29 credits, Interviews is €79 a month with 79 credits, and Enterprise is custom. Koji runs in the EU, with its app hosted in Paris. For the regulatory side, read [the EU AI Act and user research](/docs/eu-ai-act-user-research-compliance).\n\n## The bottom line\n\nListen Labs and Conveo are both serious products, and the money behind them shows how fast AI interviews moved from experiment to budget line. Listen Labs brings panel scale and, soon, Salesforce. Conveo brings brand and concept research with a European base. Both price research as a project you plan and procure.\n\nIf you want to talk to your customers this week instead, [try Koji free](https://www.koji.so). Write one research question, preview the interview, and send the link. Comparing more of the field? See [Koji vs Listen Labs](/blog/koji-vs-listen-labs-2026), [Koji vs Conveo](/blog/koji-vs-conveo-2026) and the full [guide to AI customer interview tools](/blog/best-ai-customer-interview-tools-2026).\n","category":"Comparisons","lastModified":"2026-10-29T13:18:29.61251+00:00","metaTitle":"Listen Labs vs Conveo (2026): Features, Pricing, Verdict","metaDescription":"Salesforce is acquiring Listen Labs and Conveo raised a $50M Series A. Compare panels, interview formats, customers and pricing.","keywords":["listen labs vs conveo","conveo vs listen labs","listen labs salesforce acquisition","listen labs pricing","conveo pricing","ai interview platforms","ai moderated interviews"],"aiSummary":"Listen Labs and Conveo are AI-moderated interview platforms aimed at different enterprise buyers. Salesforce signed a definitive agreement to acquire Listen Labs on September 29, 2026, with closing expected in Salesforce's fiscal Q4 2027; a price of about $2 billion was reported but not confirmed. Listen Labs, based in San Francisco, raised about $96M to $100M, including a $69M Series B at about a $500M valuation, runs video and voice interviews with emotion analysis and says its panel layer reaches 30M+ people in 45+ countries; customers include Microsoft, Anthropic and Sweetgreen. Conveo, founded in Antwerp in April 2024 and backed by Y Combinator, raised a $50M Series A in September 2026 (about $55.8M total), runs video, voice and text interviews with non-verbal analysis, focuses on brand, concept and ad testing, and names Unilever, Orange, Sanofi and Google as customers. Neither publishes prices: Listen Labs is estimated at about a $20K annual base plus per-interview credits, and Capterra lists Conveo from US$45,000. Koji is a self-serve alternative with 10 free credits, text interviews at 1 credit and voice at 3 credits (EUR 1 per credit), 28 interview languages, panel recruitment on paid plans, a quality gate where only interviews scoring 3 or more of 5 use credits, and findings traced to quotes.","aiKeywords":["Listen Labs","Conveo","Salesforce acquisition","AI moderated interviews","AI market research","platform comparison"],"aiContentType":"comparison","faqItems":[{"answer":"Listen Labs is a San Francisco AI interview platform built around panel scale, video and voice interviews, emotion analysis and fast turnaround, with customers such as Microsoft and Sweetgreen. Conveo is an Antwerp-based platform focused on consumer brand, concept and ad testing, with video, voice and text interviews, non-verbal analysis and customers such as Unilever and Sanofi. Both are quote-based enterprise tools.","question":"What is the difference between Listen Labs and Conveo?"},{"answer":"Yes. Salesforce signed a definitive agreement to acquire Listen Labs on September 29, 2026. The deal is expected to close in Salesforce's fiscal fourth quarter of 2027, subject to regulatory clearance. Terms were not disclosed, though about $2 billion was reported during the talks.","question":"Is Salesforce buying Listen Labs?"},{"answer":"Listen Labs does not publish a price list. Third-party estimates put a typical contract at about a $20K annual base plus roughly $300 to $400 per completed interview, while Listen Labs' own pricing article says most enterprise customers pay $85 to $150 per interview. Credits are priced by how hard the audience is to reach.","question":"How much does Listen Labs cost?"},{"answer":"Conveo does not publish prices. It sells credits based on interview minutes, with lower per-minute rates at higher volume, and Capterra lists a starting price of US$45,000.","question":"How much does Conveo cost?"},{"answer":"Koji is self-serve. You start with 10 free credits and no credit card, a text interview costs 1 credit and a voice interview 3 credits at EUR 1 per credit, and paid plans start at EUR 29 a month. It runs AI-moderated voice, text and phone interviews in 30+ languages, with panel recruitment on paid plans and reports whose findings trace back to quotes.","question":"Is there a self-serve alternative to Listen Labs and Conveo?"}],"relatedTopics":["Listen Labs","Conveo","AI interview platforms","AI moderated interviews","research pricing"]},{"type":"blog","id":"bdd85f95-2ea1-482d-a247-b537181eb66e","slug":"listen-labs-pricing-2026","title":"Listen Labs Pricing in 2026: What It Costs and What You Get","url":"https://www.koji.so/blog/listen-labs-pricing-2026","summary":"Listen Labs does not publish pricing. Larger companies go through a demo and pilot, and third-party sources (Contrary Research and buyer guides) report an annual platform base of about $20,000 plus roughly $300 to $400 per completed interview, with credit cost rising for hard-to-reach audiences. At that rate a 30-interview year costs about $29,000 to $32,000 and a 100-interview year about $50,000 to $60,000. Listen Labs suits enterprises that need managed recruitment from its 30M-person network. Koji is the self-serve alternative with published prices: 1 credit per qualified text interview and 3 per voice interview at €1 per credit, only interviews scoring 3 or more billed, 10 free credits at signup.","content":"# Listen Labs Pricing in 2026: What It Costs and What You Get\n\n**Short answer:** Listen Labs does not publish a price list. Larger companies are sent to a demo and a pilot, and the quote depends on your audience and volume. The most widely cited estimate, from Contrary Research and several buyer guides, is an annual platform base of about **$20,000 plus roughly $300 to $400 per completed interview**, with hard-to-reach audiences costing more. If you want AI-moderated interviews without a five-figure contract, [Koji](https://www.koji.so) publishes its prices: a qualified text interview uses 1 credit and a qualified voice interview uses 3, credits cost €1, and you start with 10 free credits and no card.\n\nThis guide explains how Listen Labs charges, what a realistic study costs at different sizes, which questions to ask on the sales call, and when a self-serve tool is the better buy.\n\n---\n\n## Why Listen Labs pricing matters right now\n\nAI-moderated interviews have gone mainstream fast. In the 2026 State of User Research report from User Interviews and UserTesting, **80% of researchers said they use AI, up 24 points in a year**. Maze's 2025 Future of User Research report found **58% of respondents using AI tools** and named **time and bandwidth as the top challenge for 63%** of teams. Research demand is rising while headcount is not, so more teams are pricing out platforms that run interviews for them.\n\n![A researcher at a laptop taking notes, the kind of busy insights team now weighing AI interview platforms](https://images.unsplash.com/photo-1520607162513-77705c0f0d4a?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHwzfHxyZXNlYXJjaGVyJTIwbGFwdG9wJTIwbm90ZXMlMjBvZmZpY2V8ZW58MXwwfHx8MTc5MjA0MzQxMHww&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Tyler Franta](https://unsplash.com/@tfrants?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\nListen Labs is the best-funded name in that market. In January 2026 it announced a **$69 million Series B led by Ribbit Capital**, with Sequoia, Conviction and Pear VC participating. The company says it has interviewed more than one million people and works with enterprises such as Microsoft, Sweetgreen, Perplexity and Robinhood. That profile tells you how it sells: to insights teams with an annual budget, through a sales process.\n\n---\n\n## How Listen Labs charges\n\nListen Labs' own articles describe the model in general terms rather than with a price list. From those pages and from third-party reports, the structure looks like this:\n\n- **A platform subscription.** Reported by Contrary Research and buyer guides at around $20,000 a year. Listen Labs has not confirmed this figure publicly, so treat it as an estimate.\n- **Credits spent per participant.** Listen Labs says credit cost scales with audience difficulty, so a general-population study uses fewer credits than a niche B2B or healthcare audience. Third-party estimates put the all-in cost at about $300 to $400 per completed interview.\n- **A demo and pilot for larger companies.** Listen Labs directs companies with 100 or more employees to a guided pilot instead of a self-serve price.\n- **Optional bring-your-own participants.** Listen Labs says teams can interview their own customers at a reduced cost, though it does not publish that rate.\n\nTwo features shape the bill. **Listen Atlas** is the company's participant network, which it describes as 30 million verified respondents across 45 or more countries. **Quality Guard** monitors interviews for fraud, low-effort answers and mismatched profiles, and one review reports that you are charged only for completes that pass it.\n\n---\n\n## What a Listen Labs study costs at different sizes\n\nUsing the reported figures ($20,000 base, $300 to $400 per completed interview), here is what different volumes would cost over a year. These are estimates built on third-party data, not quotes.\n\n| Interviews in the year | Platform base | Per-interview cost | Estimated total |\n|---|---|---|---|\n| 10 | ~$20,000 | $3,000 to $4,000 | ~$23,000 to $24,000 |\n| 30 | ~$20,000 | $9,000 to $12,000 | ~$29,000 to $32,000 |\n| 100 | ~$20,000 | $30,000 to $40,000 | ~$50,000 to $60,000 |\n| 300 | ~$20,000 | $90,000 to $120,000 | ~$110,000 to $140,000 |\n\nThe pattern is simple. At low volume, the platform base is most of what you pay, so a team that runs two or three small studies a year pays a high price per insight. At high volume, the per-interview cost takes over, and audience difficulty becomes the variable that matters most.\n\n---\n\n## What you get for the money\n\nListen Labs is a full-service platform, and the price buys real things:\n\n- **Recruitment handled for you.** A 30-million-person network and a recruitment operations team for hard audiences, such as clinicians or enterprise IT buyers. If recruiting is what slows your studies down, this is the part worth paying for.\n- **Voice and video interviews at scale**, with analysis, highlight reels and reports produced from the transcripts.\n- **Fraud and quality screening** through Quality Guard, plus a cap that limits panel participants to three studies a month.\n- **Language coverage** that Listen Labs puts at 50 or more languages for its emotion analysis.\n- **Enterprise buying support:** security reviews, contracts and onboarding.\n\nWhat the price does not buy is speed of purchase. There is no public rate card, so you cannot try a real study with your own customers this afternoon and decide by Friday.\n\n---\n\n## Questions to ask on the sales call\n\nBecause the price is negotiated, the quote you get depends on what you ask. Take this list into the call:\n\n![Checklist of questions to ask before signing an AI interview platform contract, covering billing, quality, recruitment, data and exit terms.](/visuals/ai-interview-platform-contract-questions-checklist/ai-interview-platform-contract-questions-checklist.png \"Get these answered in writing before you compare quotes.\")\n\n1. **What is the platform fee, and what does it include?** Seats, studies, languages and support levels.\n2. **How are credits priced per audience?** Ask for the rate for your two most likely audiences, not a blended average.\n3. **What counts as a billable complete?** Ask whether a short, low-effort or flagged interview uses credits.\n4. **What does it cost to interview our own customers?** Bring-your-own pricing is often where small teams save the most.\n5. **Do unused credits expire, and when?**\n6. **What is the renewal increase?** Get any escalation clause in writing.\n7. **Can we export raw transcripts and recordings if we leave?**\n\n---\n\n## When Listen Labs is worth it\n\nListen Labs makes sense when all of these are true:\n\n- You need hard-to-reach participants recruited for you, every study.\n- You run research in many markets and languages at once.\n- You have an annual research budget approved, and a procurement cycle of several weeks is normal.\n- You want one vendor to handle recruitment, moderation, analysis and reporting.\n\nIf that describes your team, the platform fee is a reasonable price for a managed workflow.\n\n## When a self-serve tool is the better buy\n\nMany teams that look at Listen Labs are not in that position. A product manager who needs 15 customer interviews before a roadmap review, or a founder testing pricing with existing users, pays mostly for a platform base they will barely use.\n\n[Koji](https://www.koji.so) is built for that buyer, and its prices are public:\n\n- **Pay per qualified interview.** A text interview uses 1 credit and a voice interview uses 3. Credits cost €1, so a 30-interview voice study with your own customers uses about 90 credits.\n- **Only good interviews are billed.** Every interview gets a [quality score from 1 to 5](/docs/understanding-quality-scores), and only those scoring 3 or more use credits or enter the report.\n- **No contract to start.** You get 10 free credits at signup with no card. After that you can buy credit packs as you go or choose a monthly plan: Insights at €29 a month (29 credits) or Interviews at €79 a month (79 credits). Credits stay valid for 12 months.\n- **Recruitment when you need it.** Interview your own customers through a link, [a CSV import with personal invite links](/docs/importing-participants-csv), or [recruit from your product](/docs/recruiting-from-your-product). On a paid plan you can also use [panel recruitment](/docs/panel-recruitment): describe your audience, see the quote per respondent in credits, then launch.\n- **Structured and open questions in one study.** Six question types (open ended, scale, single choice, multiple choice, ranking and yes/no) so you get a number to chart and the reason behind it.\n- **[AI voice interviews](/docs/ai-voice-interviews) and text interviews** in 30+ languages, with [screening questions](/docs/adding-screening-questions-to-your-study) and a free [interview preview](/docs/previewing-your-interview) before you launch.\n\n![Koji quality rating card scoring an interview Excellent with relevance, depth, coverage and completion scores](/email/shots/shot-rating-details.jpg \"Each interview is scored, and only those scoring 3 or more use credits.\")\n\nThe trade-off is honest: Koji's panel is quoted per study rather than bundled into a managed recruitment service, and there is no dedicated recruitment team scoping a sub-1% incidence audience for you. If that service is what you need, Listen Labs is built for it.\n\n---\n\n## Listen Labs pricing vs Koji at a glance\n\n| | Listen Labs | Koji |\n|---|---|---|\n| Published pricing | No, quote after a demo | Yes |\n| Reported entry cost | ~$20,000 a year platform base | Free to start, 10 credits |\n| Per interview | ~$300 to $400 per complete (reported, with panel) | 1 credit text, 3 credits voice, €1 per credit (own participants) |\n| Billing on quality | Reported: only completes that pass Quality Guard | Only interviews scoring 3 or more |\n| Recruitment | 30M network plus recruitment ops | Own participants, in-product, CSV, or panel quoted per respondent |\n| Trial | Pilot for larger companies | Self-serve, no card |\n\n![Comparison of an enterprise AI interview contract and pay per interview pricing, showing differences in commitment, trial access and who recruits.](/visuals/enterprise-contract-vs-pay-per-interview-comparison/enterprise-contract-vs-pay-per-interview-comparison.png \"Which model fits depends on volume and who you interview.\")\n\nFor a broader view, see our [Koji vs Listen Labs comparison](/blog/koji-vs-listen-labs-2026), the [Listen Labs alternatives guide](/docs/listen-labs-alternatives-2026), and [how AI-moderated interviews work](/docs/ai-moderated-interviews).\n\n---\n\n## Try AI interviews before you sign anything\n\nYou do not need a $20,000 contract to find out whether AI-moderated interviews work for your team. Write a research question, let Koji draft the brief, preview it for free, and send the link to ten customers. You will have transcripts, themes and a report before a pilot would have been scheduled.\n\n[Start free with Koji](https://www.koji.so). You get 10 credits, no card, and you only pay for interviews that pass the quality bar.\n\n---\n\n*Sources: Listen Labs Series B announcement (January 14, 2026); Listen Labs articles on pricing, Listen Atlas and Quality Guard; Contrary Research and third-party buyer guides for reported pricing; User Interviews and UserTesting, State of User Research 2026; Maze, Future of User Research Report 2025. Listen Labs pricing figures are third-party estimates as of October 2026. Ask Listen Labs for a current quote.*\n","category":"Comparisons","lastModified":"2026-10-29T13:18:29.327439+00:00","metaTitle":"Listen Labs Pricing 2026: Costs, Credits and What to Ask","metaDescription":"Listen Labs has no public price list. Reported pricing is about $20K a year plus $300 to $400 per interview. See study costs by size and what to ask sales.","keywords":["listen labs pricing","listen labs cost","listen labs price per interview","listen labs free trial","ai interview platform pricing","listen labs alternative"],"aiSummary":"Listen Labs does not publish pricing. Larger companies go through a demo and pilot, and third-party sources (Contrary Research and buyer guides) report an annual platform base of about $20,000 plus roughly $300 to $400 per completed interview, with credit cost rising for hard-to-reach audiences. At that rate a 30-interview year costs about $29,000 to $32,000 and a 100-interview year about $50,000 to $60,000. Listen Labs suits enterprises that need managed recruitment from its 30M-person network. Koji is the self-serve alternative with published prices: 1 credit per qualified text interview and 3 per voice interview at €1 per credit, only interviews scoring 3 or more billed, 10 free credits at signup.","aiKeywords":["listen labs","listen labs pricing","ai moderated interviews","research platform pricing","listen atlas","quality guard","koji"],"aiContentType":"comparison","faqItems":[{"answer":"No. Listen Labs does not publish a rate card. Larger companies, those with 100 or more employees, are sent to a demo and a guided pilot, and the quote depends on audience difficulty and volume. Third-party figures from Contrary Research and buyer guides are the main public reference.","question":"Does Listen Labs publish its pricing?"},{"answer":"Third-party estimates put it at roughly $300 to $400 per completed interview on top of a platform base of about $20,000 a year. Listen Labs says credit cost scales with audience difficulty, so niche B2B or healthcare audiences cost more than general-population studies.","question":"How much does Listen Labs cost per interview?"},{"answer":"Listen Labs evaluates larger customers through a guided pilot or demo rather than an open self-serve trial. Koji gives 10 free credits at signup with no card, enough to run real interviews with your own customers.","question":"Is there a Listen Labs free trial?"},{"answer":"Usually not at low volume. With a platform base of around $20,000, a team running ten interviews a year pays over $2,000 per interview. Listen Labs fits teams that need managed recruitment of hard audiences at high volume.","question":"Is Listen Labs worth it for small teams?"},{"answer":"Koji publishes its prices: a qualified text interview uses 1 credit and a voice interview 3 credits, at €1 per credit, and only interviews scoring 3 or more on quality are billed. Plans start at €29 a month, or you can buy credit packs as you go.","question":"What is a cheaper alternative to Listen Labs?"}],"relatedTopics":["Listen Labs","AI interview platform pricing","AI moderated interviews","research procurement","participant recruitment"]},{"type":"blog","id":"bc9c61b4-a003-4166-bebd-9ee8a5201c60","slug":"koji-vs-voiceform-2026","title":"Koji vs Voiceform (2026): Voice Surveys or AI Voice Interviews?","url":"https://www.koji.so/blog/koji-vs-voiceform-2026","summary":"Koji vs Voiceform (2026): Voiceform is a survey builder with voice and video answers, 14 question types, AI probing of 1 to 5 follow-ups after voice answers, automatic transcription, themes and sentiment, and Qualtrics and Decipher embeds; its published Lite plan includes 200 answers and 40 AI follow-ups a month, with larger plans by quote. Koji is an AI interviewer that holds live voice, text and phone conversations in 30+ languages, follows up on every question at a depth you set, mixes six structured question types, scores each interview 1 to 5, and builds reports where every finding links to its quotes, on public pricing: free with 10 credits, Insights €29/month, Interviews €79/month. Research shows spoken answers contain more words and topics than typed ones.","content":"**Short answer:** Voiceform is a survey builder where people can answer out loud. You design a form with voice, video and classic question types, respondents record their answers, and AI probing can add a few follow-up questions after a voice answer. Koji is an AI interviewer. It holds a real spoken or typed conversation, follows up on each answer the way a researcher would, and turns every conversation into a report you can trace back to the quotes. If you want to add voice answers to a survey you already run, Voiceform is a sensible choice. If you want the depth of an interview without scheduling one, Koji is built for that.\n\n## Koji vs Voiceform at a glance\n\n| | Koji | Voiceform |\n|---|---|---|\n| What it is | AI interviewer that holds a conversation | Survey builder with voice and video answers |\n| How people answer | Talk with an AI voice, type in a chat, or take a phone call | Record voice or video, or use form fields |\n| Follow-up questions | Part of every question, with depth you set | AI probing after voice answers, capped at 1 to 5 |\n| Question types | Six, inside one conversation | 14 form types, including NPS, matrix and file upload |\n| Languages | 28 for the full interview | Stated as 50+ on its site |\n| Report | Findings, personas and quotes behind every claim | Transcripts, themes, sentiment and charts |\n| Buying | Self-serve, prices on the website | One published plan with monthly limits, larger plans by quote |\n\n![Table comparing Koji and Voiceform on what each tool is, how people answer, follow-ups, question types, languages, reports and buying](/visuals/koji-vs-voiceform-feature-comparison/koji-vs-voiceform-feature-comparison.png \"Same goal, richer answers, but one is a survey and the other is a conversation\")\n\n## Why voice answers say more\n\nTyped open-ended questions are where surveys go to die. People write five words and move on. Speaking is different. In a randomised experiment with 1,001 smartphone respondents, Höhne, Gavras and Claassen found that \"voice answers have a higher number of words and a higher number of topics than their text counterparts\", with no difference in sentiment ([Social Science Computer Review, 2024](https://journals.sagepub.com/doi/10.1177/08944393231160961)).\n\n![A person speaking into a smartphone to leave a voice answer instead of typing one](https://images.unsplash.com/photo-1758876201450-cf77ab8b95bc?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHwzfHx3b21hbiUyMHRhbGtpbmclMjBzbWFydHBob25lJTIwdm9pY2UlMjBtZXNzYWdlfGVufDF8MHx8fDE3OTE3ODQyMTF8MA&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Vitaly Gariev](https://unsplash.com/@silverkblack?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\nVoice is not free of costs, and it is worth knowing them before you pick a tool. Revilla and colleagues found voice recording produced longer, more elaborate answers but also substantially higher nonresponse on Android phones ([Social Science Computer Review, 2020](https://journals.sagepub.com/doi/10.1177/0894439318810715)). And in a study by Lenzner and Höhne, only about 40% of respondents were willing to answer by voice at all ([International Journal of Market Research, 2022](https://journals.sagepub.com/doi/full/10.1177/14707853221084213)). Meanwhile classic channels keep shrinking: Pew Research Center reported a typical telephone survey response rate of 6% in 2018, down from 36% in 1997 ([Pew Research Center, 2019](https://www.pewresearch.org/short-reads/2019/02/27/response-rates-in-telephone-surveys-have-resumed-their-decline/)).\n\nThe lesson for tool choice: voice gets richer answers from the people who use it, so give people the choice of voice or text, and make the experience feel worth their time.\n\n## What Voiceform does\n\nVoiceform describes itself as \"conversational surveys and forms using voice, video, and text\" ([CB Insights profile](https://www.cbinsights.com/company/voiceform)). It was founded around 2021 by Philip Brook and Artem Alekhin and has recently moved toward AI-moderated research ([Voiceform blog](https://voiceform.com/blog/what-is-an-ai-moderated-interview)).\n\nWhat you get:\n\n- **A wide question library.** Voice response, audio question, video recording, NPS, rating, matrix, multiple choice, checkbox, dropdown, yes/no, date, email, phone and file upload ([Voiceform docs](https://docs.voiceform.com/docs/adding-questions)).\n- **AI probing.** After a voice answer, AI can generate follow-up questions. You set a probing goal and cap it at 1 to 5 follow-ups, or let the AI decide when an answer is enough ([Voiceform docs](https://docs.voiceform.com/docs/voice-response-question)).\n- **Automatic processing.** Responses are transcribed, translated and themed, with sentiment and charts ([voiceform.com](https://voiceform.com/)).\n- **Logic and integrations.** Skip logic, branching and piping, plus Zapier, webhooks and widgets you can embed in Qualtrics and Forsta Decipher surveys ([Voiceform docs](https://docs.voiceform.com/docs/qualtrics)).\n\n### Voiceform's pricing\n\nVoiceform's pricing page lists one self-serve plan, Lite, available as a free trial. It includes 200 answers a month, 40 AI follow-ups a month, a recording limit of 2 minutes 30 seconds per question and one seat. Larger plans have custom limits and are priced by quote ([Voiceform pricing](https://voiceform.com/pricing)). Older price lists still circulate on third-party sites, so check the live page before you budget.\n\n### Voiceform's trade-offs\n\n- **It is still a survey.** The respondent records an answer to a question on a screen. AI probing adds follow-ups after voice answers, but there is no back-and-forth conversation with a voice that listens and replies.\n- **Follow-ups are metered.** On the Lite plan, 40 AI follow-ups a month covers a small study at most.\n- **Mixed reviews on speed.** Voiceform rates 4.7 out of 5 from 48 reviews on G2, with praise for ease of use and audio answers. Some reviewers describe slow load times and unreliable embeds ([G2](https://www.g2.com/products/voiceform/reviews)).\n- **Language support is unclear.** The site says 50+ languages, while one of its own landing pages says transcription covers 13+.\n\n## What Koji does\n\nKoji does not ask people to record answers into a form. It interviews them. An AI interviewer opens the conversation, asks your questions, listens to each answer and follows up on what the person actually said, then moves on when it has what it needs.\n\n- **Three ways to take part.** Participants pick voice or text on the welcome screen. Voice is a live conversation with an AI voice, text is a chat, and phone interviews work through a call-back link or outbound calls you schedule. See [AI voice interviews](/docs/ai-voice-interviews) and [voice vs text interviews](/docs/voice-vs-text-interviews).\n- **Follow-ups on every question.** You set how deep the AI probes per question, with no monthly follow-up allowance to ration. See [probing and follow-up questions](/docs/probing-and-follow-up-questions).\n- **Six structured question types** inside the conversation: open ended, scale, single choice, multiple choice, ranking and yes/no, each charted in the report. See the [structured questions guide](/docs/structured-questions-guide).\n- **30+ languages**, chosen per study. The whole interview, follow-ups included, runs in the participant's language, and all answers land in one report.\n- **Screening and previews.** [Screening questions](/docs/adding-screening-questions-to-your-study) turn away people who do not qualify, and you can [preview the interview](/docs/previewing-your-interview) by voice or text for free.\n- **Recruitment built in.** Share a link, give each person a [personal invite link](/docs/personalized-interview-links) from a CSV, or [recruit from Koji's panel](/docs/panel-recruitment).\n\n![Koji voice interview in progress with an AI orb asking a follow-up question and a microphone button](/email/shots/shot-voice-interview.jpg \"A live voice interview: the AI asks a follow-up based on what the participant just said\")\n\n### Reports with the evidence attached\n\nEvery interview gets a quality score from 1 to 5, and only interviews scoring 3 or more enter the report or use credits ([how the quality gate works](/docs/how-the-quality-gate-works)). The report opens with what you learned, then findings, recommendations and personas. Click any claim and the quotes behind it open beside the report ([tracing report findings to evidence](/docs/tracing-report-findings-to-evidence)).\n\n### What Koji costs\n\nKoji's prices are on its website. You start free with 10 credits and no card. Insights is €29 a month with 29 credits and Interviews is €79 a month with 79 credits, with Enterprise on contract and extra credits at €1. A text interview uses 1 credit, a voice interview 3 and a phone interview 0.6 credits a minute. Follow-ups are included in the interview, and interviews scoring under 3 are free.\n\n## Voice survey or AI voice interview?\n\nThe real choice is not between two brands. It is between two formats.\n\n| Voice survey | AI voice interview |\n|---|---|\n| Respondent records answers to fixed questions | Respondent talks with an AI that listens and replies |\n| Follow-ups, if any, come after the recording | Follow-ups come in the flow of the conversation |\n| Good for adding colour to a survey you already run | Good for learning why, at the depth of an interview |\n| Many form question types for quant-heavy studies | Fewer question types, more depth on each one |\n| You analyse transcripts and themes | You get a report with findings tied to quotes |\n\n![Comparison of a voice survey and an AI voice interview, from how people answer to what you get at the end](/visuals/voice-survey-vs-ai-voice-interview/voice-survey-vs-ai-voice-interview.png \"Choose the format first, then the tool\")\n\n## Which one should you pick?\n\n### Pick Voiceform if\n\n- You already run surveys and want voice or video answers on a few open questions.\n- You need form features like matrix questions, file uploads or date pickers.\n- You want to embed voice questions inside an existing Qualtrics or Decipher survey.\n\n### Pick Koji if\n\n- You want people to have a conversation, not fill in a form.\n- You need follow-ups on every answer without a monthly cap.\n- You want phone interviews for people who will not sit at a screen.\n- You want a finished report where every finding links to what people said.\n- You want to know a study's cost before you launch it.\n\n## The bottom line\n\nVoiceform adds voice to the survey. Koji replaces the survey with an interview. Both get you richer answers than a text box, and the research above says that is worth doing. The difference shows up in the follow-ups and the report: Koji asks the next question the way a researcher would, on every answer, and hands you findings you can check against the quotes.\n\n## Hear your customers explain themselves\n\n[Start free on Koji](/signup) with 10 credits. Write your research question, preview the voice interview yourself, and send the link to a few customers today. If you are moving from a form, the [survey to AI interview guide](/docs/convert-survey-to-ai-interview) shows how to turn your existing questions into a conversation, and the [AI voice surveys guide](/docs/ai-voice-surveys-complete-guide) covers when voice works best.\n","category":"Comparisons","lastModified":"2026-10-29T13:18:29.072088+00:00","metaTitle":"Koji vs Voiceform 2026: Voice Surveys or AI Voice Interviews?","metaDescription":"Voiceform adds voice answers to surveys. Koji runs AI voice, text and phone interviews with follow-ups on every answer. Compare features, pricing and fit.","keywords":["koji vs voiceform","voiceform alternative","voiceform pricing","voice survey software","ai voice interviews","voice survey vs interview","ai probing"],"aiSummary":"Koji vs Voiceform (2026): Voiceform is a survey builder with voice and video answers, 14 question types, AI probing of 1 to 5 follow-ups after voice answers, automatic transcription, themes and sentiment, and Qualtrics and Decipher embeds; its published Lite plan includes 200 answers and 40 AI follow-ups a month, with larger plans by quote. Koji is an AI interviewer that holds live voice, text and phone conversations in 30+ languages, follows up on every question at a depth you set, mixes six structured question types, scores each interview 1 to 5, and builds reports where every finding links to its quotes, on public pricing: free with 10 credits, Insights €29/month, Interviews €79/month. Research shows spoken answers contain more words and topics than typed ones.","aiKeywords":["voiceform","voiceform alternative","voice survey","ai voice interviews","phone interviews","ai follow-up questions","voice vs text answers","survey software","research reports","conversational research"],"aiContentType":"comparison","faqItems":[{"answer":"Voiceform is a survey builder where respondents can answer by voice, video or standard form fields. It transcribes, translates and themes the answers, and its AI probing feature can ask 1 to 5 follow-up questions after a voice answer.","question":"What is Voiceform?"},{"answer":"In a voice survey, people record answers to fixed questions on a screen. In an AI voice interview, they talk with an AI that listens, replies and follows up in the flow of the conversation. Koji runs AI voice interviews; Voiceform is a voice survey tool with AI follow-ups added.","question":"What is the difference between a voice survey and an AI voice interview?"},{"answer":"Voiceform's pricing page lists a Lite plan, available as a free trial, with 200 answers and 40 AI follow-ups a month and one seat. Larger plans are priced by quote. Koji starts free with 10 credits, then Insights at €29 a month or Interviews at €79 a month, with follow-ups included in every interview.","question":"How much does Voiceform cost?"},{"answer":"They are usually richer. A 2024 experiment with 1,001 smartphone respondents found voice answers contain more words and more topics than typed ones. But fewer people are willing to answer by voice, so it helps to let participants choose voice or text, as Koji does.","question":"Are voice answers better than typed answers?"},{"answer":"Yes, if you want interview depth rather than survey answers. Koji follows up on every question, offers voice, text and phone interviews in 30+ languages, and ends with a report where each finding links to the quotes behind it. If you mainly need voice questions inside an existing survey, Voiceform fits that job.","question":"Is Koji a good Voiceform alternative?"}],"relatedTopics":["voice survey software","voiceform alternative","ai voice interviews","voice vs text answers","conversational surveys"]},{"type":"blog","id":"7d20b5fa-75cd-460f-a1f5-8c7ec7ff4cae","slug":"koji-vs-tellet-2026","title":"Koji vs Tellet (2026): AI-Moderated Interview Platforms Compared","url":"https://www.koji.so/blog/koji-vs-tellet-2026","summary":"Koji vs Tellet (2026): Tellet is an Amsterdam-based AI interview platform founded by a market research veteran, offering voice, video, photo and text responses with AI follow-ups, a design and analysis co-pilot, and panel partners; it does not publish pricing and sells to agencies, consultancies and brands. Koji is a self-serve AI-native research platform with voice, text and phone interviews in 30+ languages, six structured question types, screening questions, a built-in panel, a 1 to 5 quality score per interview, and reports where every finding links to supporting quotes. Koji pricing is public: free with 10 credits, Insights €29/month, Interviews €79/month, €1 per extra credit; text interviews use 1 credit and voice 3.","content":"**Short answer:** Tellet and Koji both run AI-moderated research interviews and turn them into themes and reports. Tellet comes from market research: it was founded in Amsterdam by a market research veteran, sells to agencies and brands, and does not publish its prices. Koji is built for product, research and customer teams who want to start a study today on public, per-interview pricing, with voice, text and phone interviews, six structured question types and a report where every finding links back to the people who said it. If you run agency fieldwork and want a vendor relationship, Tellet is worth a demo. If you want to launch this week and know what it will cost before you start, Koji is the faster route.\n\n## Koji vs Tellet at a glance\n\n| | Koji | Tellet |\n|---|---|---|\n| Interview formats | Voice, text and phone calls | Voice, video, photo and text responses |\n| Languages | 28, chosen per study | Not consistently stated (sources say 30 to 64) |\n| Structured questions | Six types: open ended, scale, single choice, multiple choice, ranking, yes/no | Qualitative and quantitative questions |\n| Recruitment | Your own list, links, CSV invite links, or Koji's built-in panel | Your own participants or panel partners |\n| Analysis | Themes, personas, findings with linked evidence, quality score per interview | Themes, patterns, recommendations and quotes |\n| Pricing | Published on the website; free to start, pay per interview | Not published; demo or free trial |\n| Best fit | Product, UX, CX and founder teams; agencies wanting self-serve | Market research agencies, consultancies and brands |\n\n![Table comparing Koji and Tellet on interview formats, languages, question types, recruitment, analysis and how each is bought](/visuals/koji-vs-tellet-ai-interview-platform-comparison/koji-vs-tellet-ai-interview-platform-comparison.png \"The differences that decide which platform fits your team\")\n\nThe table hides one thing worth saying out loud: on most features these two products are closer than their marketing suggests. The real split is in who they are built for and how you buy them.\n\n## Why AI-moderated interviews became a category\n\nTwo numbers explain why a dozen platforms now do what Tellet and Koji do. In User Interviews' State of User Research 2025, the share of researchers using AI jumped 24 points in a year to reach 80% ([User Interviews, 2025](https://www.userinterviews.com/state-of-user-research-report)). Qualtrics' 2025 Market Research Trends report, which surveyed 3,198 researchers across 14 countries, found 89% already use AI tools regularly or experimentally ([Qualtrics, 2024](https://www.qualtrics.com/articles/news/ai-to-drive-massive-changes-to-market-research-in-2025-qualtrics-report-says/)). Maze's 2026 Future of User Research report puts it at 69% of research participants using AI in at least some projects ([Maze, 2026](https://maze.co/resources/user-research-report/)).\n\nThe trust gap matters just as much. The same User Interviews report found 91% of researchers worry about the accuracy of AI output. That is why the comparison below spends as much time on evidence and quality as on features. An AI moderator that asks good follow-ups is only half the job. You also need to check what it concluded.\n\n## What Tellet does\n\nTellet describes itself as a way to \"use AI to conduct and analyse consumer research interviews\" ([tellet.ai](https://tellet.ai/)). The company is based in Amsterdam and was founded around 2022 by Merik te Grotenhuis, who spent roughly 25 years in market research and sold his previous firm, CG Research, to Norstat in 2021. Tellet raised a €400K pre-seed round from angel investors ([Silicon Canals](https://siliconcanals.com/tellet-raises-400k/)).\n\nWhat you get, based on Tellet's own pages:\n\n- **Rich response formats.** Participants can answer by voice, video, photo or text, and the AI asks follow-up questions as they go ([Tellet solution page](https://tellet.ai/solution)).\n- **A design co-pilot** that sharpens your questions and suggests new ones, with support for both qualitative and quantitative questions.\n- **An analysis co-pilot** that surfaces themes, patterns, opportunities and recommendations, with automatic transcription. Its reports include tensions, decision drivers and quotes.\n- **Recruitment** through your own participants or Tellet's network of panel partners ([Insight Platforms](https://www.insightplatforms.com/platforms/tellet/)).\n- **Security claims** of GDPR compliance, AES-256 encryption and models that are never trained on your data. Tellet's site describes ISO 27001 as in progress rather than certified, at the time of writing.\n\n### Who Tellet is built for\n\nTellet's home page says it is \"trusted by the world's leading agencies, consultancies and brands\", and research agencies such as Blauw have written publicly about using it as a moderator ([Blauw](https://www.blauw.com/en/blog/employ-tellet-ai-as-moderator)). That heritage shows. Photo and video answers suit consumer and shopper studies, and an agency running a client project is used to booking a demo and agreeing terms.\n\n![A market research team talking through a client project around a meeting table](https://images.unsplash.com/photo-1603202662747-00e33e7d1468?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHw5fHxtYXJrZXQlMjByZXNlYXJjaCUyMHRlYW0lMjBtZWV0aW5nJTIwb2ZmaWNlfGVufDF8MHx8fDE3OTE3ODQyMTB8MA&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Cherrydeck](https://unsplash.com/@cherrydeck?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\n### Tellet's trade-offs\n\n- **No public pricing.** Tellet's site points visitors to a demo, an email or a free trial. You cannot work out what a 200-interview study costs before you talk to them.\n- **Few public reviews.** At the time of writing we could not find a Tellet listing with reviews on G2 or Capterra, so there is little independent user feedback to read.\n- **Exports are basic.** Tellet's guide lists CSV and TXT exports of transcripts and metadata, with PowerPoint export marked as coming soon ([Tellet guide](https://tellet.ai/guides/analyze-your-study/)).\n- **Language count is unclear.** Tellet's own content and third parties give figures from 30 to 64. Check the current number with them if you research in many markets.\n\n## What Koji does\n\nKoji is an AI-native research platform. You describe what you want to learn, Koji drafts the research brief and interview guide, an AI interviewer runs the conversations, and a report builds itself as interviews come in.\n\n- **Voice, text and phone interviews.** Participants choose voice or text on the welcome screen, and phone interviews work through a call-back link or outbound calls you schedule. See [AI voice interviews](/docs/ai-voice-interviews) and [voice vs text interviews](/docs/voice-vs-text-interviews).\n- **30+ languages.** You pick which languages a study offers, participants choose theirs, and the whole interview, follow-ups included, runs in that language. Every answer lands in one report.\n- **Six structured question types** in the same conversation as open-ended probing: open ended, scale, single choice, multiple choice, ranking and yes/no. Each gets the right chart in the report. See the [structured questions guide](/docs/structured-questions-guide).\n- **Follow-ups you control.** You set how deep the AI probes on each question, and [guardrails](/docs/setting-interview-guardrails) keep it away from topics it must not touch.\n- **Screening before the interview.** [Screening questions](/docs/adding-screening-questions-to-your-study) run for every participant, however they arrived, and turn away anyone who does not qualify.\n- **Recruitment built in.** Share a link, import a CSV to give each person a [personal invite link](/docs/personalized-interview-links), or [recruit from Koji's panel](/docs/panel-recruitment) and see the cost before you launch.\n- **Free previews.** Take your own interview by text or voice before launch. [Previews](/docs/previewing-your-interview) never use credits and never reach the report.\n\n### Reports you can check\n\nThis is where Koji answers the 91% who worry about AI accuracy. Every interview gets a quality score from 1 to 5, and only interviews scoring 3 or more enter the report or use credits ([how the quality gate works](/docs/how-the-quality-gate-works)). The report opens with what you learned, then lists findings, recommendations and personas. Click any number or claim and a panel opens with every participant behind it, in their own words. See [tracing report findings to evidence](/docs/tracing-report-findings-to-evidence).\n\n![Koji report summary with a side panel of supporting participant quotes behind one finding](/email/shots/shot-evidence-panel.jpg \"Click a claim in the report and the quotes behind it open beside it\")\n\n### What Koji costs\n\nKoji's pricing is on its website. You start free with 10 credits and no card. Paid plans are Insights at €29 a month (29 credits) and Interviews at €79 a month (79 credits), with Enterprise on contract. Extra credits cost €1. A text interview uses 1 credit, a voice interview 3, and a phone interview 0.6 credits a minute. Because of the quality gate, a low-effort interview costs nothing.\n\n## Head to head: where each one wins\n\n### Choose Tellet if\n\n- You are a market research agency or consultancy and want photo and video answers for consumer work.\n- You prefer a sales-led relationship with onboarding and agreed terms.\n- Your participants come from panel partners you already use with Tellet.\n\n### Choose Koji if\n\n- You want to set up and launch a study today without a demo.\n- You need to know the cost of a study before you start it.\n- You want qualitative depth and structured data (ratings, rankings, choices) from the same conversation.\n- You need phone interviews for people who will not sit at a laptop.\n- Your stakeholders will ask \"how many people said that?\" and you want the answer one click away.\n- You research across many markets and want one report for all languages.\n\n## Questions to ask any AI interview vendor\n\nWhichever platform you shortlist, run the same checks. These are the questions that separate a good demo from a tool that holds up on a real study:\n\n1. Can I take the interview myself before launch, for free?\n2. How deep does the AI probe, and can I set that per question?\n3. Can I screen people out before the interview starts?\n4. Are low-effort interviews scored, filtered and left uncharged?\n5. Can I trace every finding back to the quotes behind it?\n6. What will a 100-interview study cost, in writing, before I start?\n7. Which languages work for voice, not only text?\n8. Where is the data stored, and is it used to train models?\n\n![Checklist of eight questions to ask before choosing an AI interview platform, from free previews to data storage](/visuals/ai-interview-platform-vendor-checklist/ai-interview-platform-vendor-checklist.png \"Run every vendor through the same eight questions before you sign\")\n\nKoji answers yes to the first five out of the box and publishes its prices, which covers the sixth.\n\n## The bottom line\n\nTellet is a credible AI moderator with real market research roots, a strong fit for agencies that work in photo and video and are happy to buy through a demo. Koji covers the same core job, AI interviews and automatic analysis, for teams that want to move on their own schedule: voice, text and phone interviews in 30+ languages, structured questions alongside open ones, a quality gate on every interview, and a report you can audit line by line, on pricing you can read before you sign up.\n\n## Try Koji on your next study\n\nThe quickest way to compare is to run something real. [Start free on Koji](/signup) with 10 credits, describe what you want to learn in one message, preview the interview yourself, then send the link to a handful of customers. You will have a report to judge before a demo call would have been booked. For the details, start with [how AI interviewers work](/docs/how-ai-interviewers-work) and [AI-moderated interviews](/docs/ai-moderated-interviews).\n","category":"Comparisons","lastModified":"2026-10-29T13:18:28.511218+00:00","metaTitle":"Koji vs Tellet 2026: AI Interview Platforms Compared","metaDescription":"Tellet and Koji both run AI-moderated interviews. Compare formats, languages, question types, recruitment, analysis and pricing to see which fits your team.","keywords":["koji vs tellet","tellet alternative","tellet pricing","tellet ai","ai moderated interviews","ai interview platform","ai research interviews"],"aiSummary":"Koji vs Tellet (2026): Tellet is an Amsterdam-based AI interview platform founded by a market research veteran, offering voice, video, photo and text responses with AI follow-ups, a design and analysis co-pilot, and panel partners; it does not publish pricing and sells to agencies, consultancies and brands. Koji is a self-serve AI-native research platform with voice, text and phone interviews in 30+ languages, six structured question types, screening questions, a built-in panel, a 1 to 5 quality score per interview, and reports where every finding links to supporting quotes. Koji pricing is public: free with 10 credits, Insights €29/month, Interviews €79/month, €1 per extra credit; text interviews use 1 credit and voice 3.","aiKeywords":["tellet","tellet alternative","ai moderated interviews","ai voice interviews","ai interview platform","market research","qualitative research at scale","panel recruitment","structured questions","research reports"],"aiContentType":"comparison","faqItems":[{"answer":"Tellet is an Amsterdam-based AI research interview platform. Its AI moderator runs interviews where participants answer by voice, video, photo or text, asks follow-up questions, and an analysis co-pilot turns the answers into themes, patterns and recommendations. It mainly serves market research agencies, consultancies and brands.","question":"What is Tellet?"},{"answer":"Tellet does not publish its pricing. Its website points you to a demo, an email or a free trial. Koji publishes its prices: free to start with 10 credits, then Insights at €29 a month or Interviews at €79 a month, with extra credits at €1.","question":"How much does Tellet cost?"},{"answer":"Yes, if you want to launch AI-moderated interviews yourself without a sales process. Koji runs voice, text and phone interviews in 30+ languages, mixes six structured question types with open-ended probing, and builds a report where every finding links to the quotes behind it.","question":"Is Koji a good Tellet alternative?"},{"answer":"No. Koji focuses on voice, text and phone interviews. If photo or video answers are central to your study, for example shopper or packaging research, Tellet's formats may suit you better.","question":"Does Koji support video answers like Tellet?"},{"answer":"Koji. Product teams usually want to start a study the same day, preview it for free, recruit from their own users or a panel, and get a report they can check against the source quotes. Koji is built around that loop on public pricing.","question":"Which is better for product teams?"}],"relatedTopics":["ai moderated interviews","tellet alternative","voice interviews","market research platforms","research reports"]},{"type":"blog","id":"d0131a61-930f-473d-8e0f-ccdf29fe0874","slug":"koji-vs-qualzy-2026","title":"Koji vs Qualzy (2026): Async Qual Communities or AI Interviews?","url":"https://www.koji.so/blog/koji-vs-qualzy-2026","summary":"A comparison of Qualzy, an asynchronous qualitative platform for online communities, diary studies and homework tasks, and Koji, which runs AI-moderated voice, text and phone interviews in 30+ languages. It covers method, who moderates, participant effort, languages and pricing models, and suggests running AI interviews for breadth and a community or diary for behaviour over time.","content":"**Short answer:** Qualzy and Koji both help research teams collect qualitative data online, but they run very different kinds of study. Qualzy is an asynchronous qual platform: you build communities, diary studies and homework tasks from more than 15 activity types, participants respond over days or weeks, and AI transcribes and summarises each response as it arrives. Koji runs AI-moderated interviews: one conversation per participant, by voice, text or phone, where the AI asks follow-up questions in real time and a report builds as interviews finish. Pick Qualzy if your method is a moderated community or a multi-week diary with tasks like card sorts and mood boards. Pick Koji if you need many one-to-one interviews quickly, with probing on every answer, a quality score on every conversation and public pricing from free.\n\n## Koji vs Qualzy at a glance\n\n| | Qualzy | Koji |\n|---|---|---|\n| Method | Async communities, diaries and tasks | AI-moderated one-to-one interviews |\n| Who moderates | Your researchers, with AI follow-up prompts | The AI interviewer, with follow-ups in real time |\n| Activity types | 15+, including video tasks, card sorts and mood boards | Open questions plus scales, choice, ranking and yes/no |\n| Participant time | Repeated entries over days or weeks | One conversation, usually minutes |\n| Languages | AI translation in 30+ languages, per Qualzy | Interviews in 30+ languages |\n| Pricing | Per project or by participant slots | Public pricing; credits per completed interview |\n| Best fit | Agencies running communities and diary studies | Teams that need many interviews, fast |\n\n![Table comparing Qualzy and Koji on method, moderation, activity types, participant time, languages, pricing and best fit](/visuals/koji-vs-qualzy-comparison-table/koji-vs-qualzy-comparison-table.png \"Qualzy runs communities and diaries over weeks, Koji runs one AI-moderated interview per person\")\n\n## Why this comparison comes up\n\nAgencies and insight teams are rebuilding their qual toolkits around AI. In Qualtrics' 2025 Market Research Trends report, a survey of more than 3,000 researchers in 14 countries, **89% said they use AI tools regularly or in experimental phases** and **83% of organisations planned to significantly increase AI investment** ([Qualtrics](https://www.qualtrics.com/articles/news/ai-to-drive-massive-changes-to-market-research-in-2025-qualtrics-report-says/)).\n\nThat leaves two broad routes for online qual. One keeps the classic methods, like communities and diaries, and uses AI to speed up analysis. Qualzy takes that route. The other hands the moderation itself to AI, so one interviewer can talk to hundreds of people at once. Koji takes that one. The evidence on the second route has grown: in a randomised experiment with 3,160 people on the Verasight panel, AI-moderated interviews produced **4.8 times as many words** as written open-ends, though completion **fell from 99.4% to 40.5%** ([Verasight, August 2026](https://verasight.io/reports/ai-moderated-interviews-vs-survey-open-ends)). Depth goes up, and you have to plan for the people who won't talk to an AI.\n\nLanguage is the other pressure. CSA Research found that **76% of 8,709 consumers in 29 countries prefer product information in their own language** ([CSA Research](https://csa-research.com/blogs-events/csa-in-the-media/press-releases/consumers-prefer-their-own-language)). Both tools take this seriously, in different ways.\n\n## What Qualzy does\n\nQualzy is a qualitative research platform with teams in the UK, US and Australia, built for agencies and client-side insight teams. What it says about itself ([Qualzy](https://qualzy.com/platform.html)):\n\n- **Async methods.** Online communities, diary studies, bulletin boards, homework tasks and in-depth interviews, run over hours, days or weeks.\n- **15+ activity types.** Including video tasks, heat maps, card sorts and mood boards. One feature page counts 18.\n- **AI on every response.** Each response is transcribed and broken into AI key points and clips the moment it's submitted. Maizy Chat lets you ask questions of the whole dataset.\n- **AI follow-ups and translation.** An AI moderator can suggest follow-up questions, and AI translation covers 30+ languages ([Qualzy for agencies](https://qualzy.com/for/agencies.html)).\n- **Video import.** Bring in Zoom or Teams recordings for transcription and clips, so live sessions can sit next to async work.\n- **Security.** Qualzy states ISO 27001:2022 certification and says it is GDPR, HIPAA and CCPA compliant.\n\n**Pricing.** Qualzy offers two models: pay per project, with no monthly fee, or a subscription based on how many active participant slots you need ([Qualzy pricing](https://qualzy.com/pricing.html)). Its app pricing page describes a one-time platform fee of AUD 300 plus GST and a daily charge while a project is live, with an optional Qualzy Assist add-on listed at $700 for expert build, launch and management ([Qualzy app](https://app.qualzy.com/pricing/)). Check the current daily rate with Qualzy, as it isn't shown in public listings.\n\n**Limits worth knowing.** Async methods ask more of participants: several entries over days, not one sitting, and diaries tend to lose people as they run. Someone on your team still designs activities and moderates the community. And public reviews are few; most performance claims come from Qualzy's own case studies.\n\n## What Koji does\n\nKoji is an AI-native research platform for running interviews at scale. You type what you want to learn, the research assistant drafts a brief, and the AI interviewer holds every conversation.\n\n- **Voice, text and phone interviews** in [30+ languages](/docs/multilingual-research-guide). Each participant picks their language on the welcome page. Voice interviews are on the Interviews and Enterprise plans; text works on every plan.\n- **Real-time probing.** The interviewer follows up on each answer, so you get the story behind it without a moderator online.\n- **Structure inside the conversation.** Mix open questions with scales, single choice, multiple choice, ranking and yes/no questions.\n- **Recruitment built in.** Share a link, [import a CSV](/docs/importing-participants-csv) for personal invite links (Interviews and Enterprise), or [recruit from Koji's panel](/docs/panel-recruitment) on paid plans. [Screening questions](/docs/adding-screening-questions-to-your-study) run before the interview.\n- **A quality gate.** Every interview gets a [quality score from 1 to 5](/docs/understanding-quality-scores); only those scoring 3 or more use credits or enter the report.\n- **Reports with evidence.** Findings, themes and personas, each finding linked to the quotes behind it, updating as interviews come in.\n- **For agencies.** [Team workspaces](/docs/creating-a-team-workspace) from the Insights plan, and brand themes on Enterprise.\n\n**Pricing.** A free account comes with 10 credits. Insights is €29 a month with 29 credits, Interviews is €79 a month with 79 credits, and Enterprise is priced on request. A text interview uses 1 credit and a voice interview 3. Extra credits are €1 each.\n\n![Koji voice interview in progress with an AI orb asking a follow-up question and a microphone button](/email/shots/shot-voice-interview.jpg \"A Koji voice interview: the AI asks a follow-up based on what the participant just said\")\n\n## Where they differ\n\n### Participant effort\n\nA Qualzy diary might ask someone to log in every day for two weeks, upload a video and sort some cards. That gives you behaviour over time, which a single interview can't. It also costs participants more time, and some will drop out before the end.\n\n![A person writing a daily entry on paper, the kind of repeated task a diary study asks of participants](https://images.unsplash.com/photo-1596797639370-96abbfe337ef?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHw3fHxwZXJzb24lMjB3cml0aW5nJTIwam91cm5hbCUyMHBob25lfGVufDF8MHx8fDE3OTI1NjE4MTR8MA&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Lawrence Aritao](https://unsplash.com/@lawaritao?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\nA Koji interview is one sitting. The participant talks or types, the AI asks follow-ups, and they are done. That makes it easier to reach busy people, B2B buyers or customers who would never join a community.\n\n### Who does the moderating\n\nOn Qualzy, your researchers run the community, with AI helping on follow-ups and analysis. That is a strength when the method needs a skilled moderator's judgement over many days. On Koji, the AI does the moderating, so 200 interviews take the same team time as 20. Guardrails and a free preview let you check how it behaves first.\n\n### Breadth of methods\n\nQualzy has the wider method toolkit: card sorts, heat maps, mood boards and video tasks. Koji focuses on the interview and does it at scale, with structured questions in the same conversation.\n\n### How you pay\n\nQualzy charges for the project and the time it is live. Koji charges per completed interview that passes the quality gate. You can price a Koji study before you start: 50 text interviews are 50 credits.\n\n## Using both\n\nMany agencies don't have to choose. A common pattern is to run 100 AI interviews in Koji to find the main themes quickly, then recruit a smaller group into a Qualzy community or diary to watch those themes play out over two weeks. For more on that split, read our guide to [diary studies](/docs/diary-study-guide) and to [async user interviews](/docs/async-user-interviews).\n\n## Who should choose which\n\n**Choose Qualzy if:**\n\n- Your method is a long-running community or a diary study.\n- You need tasks such as card sorts, mood boards or video uploads.\n- Your researchers want to moderate, with AI speeding up analysis.\n\n**Choose Koji if:**\n\n- You need many one-to-one interviews, quickly, without moderators online.\n- You want probing on every answer and a quality score on every conversation.\n- You run studies in several languages and want one interviewer logic across all of them.\n- You want public pricing and to pay only for interviews that pass a quality check.\n\n## The bottom line\n\nQualzy is a solid choice for agencies whose work centres on async communities and diaries. Koji is the faster route to depth at scale: AI-moderated voice, text and phone interviews in 30+ languages, a report with traceable quotes and prices you can see before you start.\n\n## Try Koji free\n\n[Create a free Koji account](https://www.koji.so), type your research question and preview the interview yourself. Your 10 signup credits cover ten text interviews. If you run an agency, read our [guide to user research for agencies](/docs/user-research-for-agencies) or the [multi-market qualitative playbook](/blog/multi-market-qualitative-research-agency-playbook).\n","category":"Comparisons","lastModified":"2026-10-29T13:18:28.165558+00:00","metaTitle":"Koji vs Qualzy 2026: Async Qual vs AI Interviews","metaDescription":"Qualzy runs async communities and diaries; Koji runs AI-moderated interviews. Compare method, participant effort, languages and pricing, and when to use both.","keywords":["koji vs qualzy","qualzy alternative","qualzy pricing","async qualitative research","online community research platform","diary study software","ai moderated interviews"],"aiSummary":"A comparison of Qualzy, an asynchronous qualitative platform for online communities, diary studies and homework tasks, and Koji, which runs AI-moderated voice, text and phone interviews in 30+ languages. It covers method, who moderates, participant effort, languages and pricing models, and suggests running AI interviews for breadth and a community or diary for behaviour over time.","aiKeywords":["qualzy","async qualitative research","online research communities","diary studies","ai moderated interviews","qual research agencies","research pricing","koji"],"aiContentType":"comparison","faqItems":[{"answer":"Qualzy is an asynchronous qualitative research platform for agencies and insight teams. It runs online communities, diary studies, bulletin boards and homework tasks from more than 15 activity types, and uses AI to transcribe and summarise each response as it arrives.","question":"What is Qualzy?"},{"answer":"Qualzy offers pay-per-project pricing with no monthly fee, or a subscription based on active participant slots. Its app pricing page lists a one-time platform fee of AUD 300 plus GST and a daily charge while a project is live. Koji publishes its prices: a free account with 10 credits, Insights at €29 a month and Interviews at €79 a month.","question":"How is Qualzy priced?"},{"answer":"An async community or diary collects several entries from each participant over days or weeks, with researchers moderating. An AI interview is a single conversation in which the AI asks follow-up questions in real time, so you can run many interviews at once without moderators online.","question":"What is the difference between an async community and an AI interview?"},{"answer":"Yes. A common pattern is to run a large set of AI interviews in Koji to find the main themes quickly, then recruit a smaller group into a community or diary study to watch those themes over time.","question":"Can I use Koji and Qualzy together?"},{"answer":"Koji runs interviews in 30+ languages. Each participant chooses their language on the welcome page, and the whole interview, including follow-up questions, runs in that language by voice or text.","question":"Which languages does Koji support?"}],"relatedTopics":["qualzy","async qualitative research","online communities","diary studies","ai moderated interviews"]},{"type":"blog","id":"8bab0014-60ec-4a25-9425-7ea9cd40b0ff","slug":"koji-vs-playbookux-2026","title":"Koji vs PlaybookUX: AI-Moderated Customer Research vs Unmoderated UX Testing (2026)","url":"https://www.koji.so/blog/koji-vs-playbookux-2026","summary":"Koji and PlaybookUX target different research problems. PlaybookUX is an unmoderated usability testing platform with a 6M-participant panel, best for observing task completion, card sorts, tree tests, and prototype usability. Koji is an AI-moderated customer interview platform — voice or text — that probes for the \"why\" behind behavior and automatically synthesizes themes across hundreds of conversations. Koji runs as low as €1 per qualified interview and €3 per qualified voice interview, starting free with 10 credits and no card; PlaybookUX starts at $5,400/year. Use PlaybookUX for behavioral observation; use Koji for customer discovery, JTBD, churn, pricing, and value-prop research.","content":"# Koji vs PlaybookUX: AI-Moderated Customer Research vs Unmoderated UX Testing (2026)\n\n**Quick answer:** Koji and PlaybookUX solve different research problems. PlaybookUX is built for unmoderated usability testing — you write tasks, recruit from a 6M-person panel, and watch screen recordings of people clicking through your product. Koji is built for AI-moderated customer research interviews — an AI voice or text interviewer asks open-ended questions, probes follow-ups in real time, and synthesizes hundreds of conversations into themes automatically. If you want to know **what users do** on a flow, choose PlaybookUX. If you want to know **why they do it, what they want, and what would make them pay**, choose Koji.\n\nThis guide breaks down both platforms across methodology, pricing, AI capabilities, analysis depth, and the use cases each one wins.\n\n## TL;DR Comparison\n\n| Dimension | Koji | PlaybookUX |\n| --- | --- | --- |\n| Core method | AI-moderated voice & text interviews | Unmoderated usability testing |\n| AI interviewer | Yes — two-way conversational AI | No — task prompts only (AI personas via add-on) |\n| Probing follow-ups | Automatic, real-time | None (preset task instructions) |\n| Question types | 6 structured types + open conversation | Task prompts, surveys, card sort, tree test |\n| Participant panel | Built in: describe your audience, approve a credit quote | 6M+ vetted participants |\n| Thematic analysis | Automatic across all interviews | Manual tagging + AI summaries |\n| Multilingual | 30+ languages with native voice | English + select languages |\n| Free plan | Yes: 10 credits at signup, 1 active study | 14-day trial (no forever-free) |\n| Starting paid price | From €1 per qualified interview; pay as you go, no subscription | $5,400/year (Scale plan) |\n| Best for | Customer discovery, JTBD, churn, pricing, message testing | Click-through usability, task completion, prototype testing |\n\n## What PlaybookUX Is (and Where It Shines)\n\nPlaybookUX is a usability testing platform launched as an affordable alternative to UserTesting. According to its [own pricing page](https://www.playbookux.com/pricing/), it supports unmoderated and moderated usability tests, card sorting, tree testing, surveys, first-click testing, five-second testing, and preference testing across desktop, tablet, and mobile.\n\nIts standout asset is the **6 million-person participant panel**, which means you can launch a test in the morning and have a dozen screen-recorded sessions back the same day. PlaybookUX also added **Synthetic Participants** in 2025 — AI personas grounded in your past video research that you can deploy in new studies — and it integrates with data warehouses, CRMs, and internal tools.\n\nThe pricing on PlaybookUX is structured into three tiers: a **Pay-as-you-go** option for ad-hoc tests, **Scale** at $5,400/year (billed annually) for card sorting, tree testing, and recruitment, and **Pro** at $8,800/year that adds approval flow, custom consent forms, invisible observers, and SSO + SAML. Enterprise pricing is custom.\n\n**PlaybookUX is the right choice when you need to:**\n- Watch people use a clickable prototype or live site\n- Run a card sort or tree test on your information architecture\n- Validate that a redesigned flow actually reduces friction\n- Test five-second first impressions of landing pages\n- Recruit a large, vetted panel quickly without building your own\n\n## What Koji Is (and Where It Shines)\n\nKoji is the AI-native customer research platform. Instead of giving participants a task and watching them click, Koji runs **two-way AI-moderated interviews** — voice or text — where an AI interviewer asks your discussion guide questions, listens to answers, and **probes for the \"why\" in real time** with follow-up questions. The AI never gets tired, never leads the witness, and runs hundreds of interviews in parallel.\n\nWhere PlaybookUX captures *behavior*, Koji captures *meaning*. You learn why a customer almost churned, which job they \"hired\" your product to do, what they would pay, which message resonates, and what would make them recommend you. After interviews complete, Koji automatically synthesizes:\n\n- Thematic clusters traceable back to every quote\n- Quality scores per interview\n- Sentiment and emotion analysis\n- Pain points, feature requests, and key insights\n- A one-click research report ready to share with stakeholders\n\nKoji supports **6 structured question types** — open-ended, scale, single-choice, multiple-choice, ranking, and yes/no — that mix seamlessly with conversational follow-ups. See the [structured questions guide](/docs/structured-questions-guide) for how teams combine them.\n\nPricing is built for accessibility. Interviews start as low as **€1 per qualified interview** and **€3 per qualified voice interview**. Start free with 10 credits at signup (one active study at a time), no card. Then pay as you go, with no subscription needed. There is no overage. You pay as you go, and you only ever pay for the conversations that bring insights. If your balance hits zero, your study pauses until you top up, and nothing is ever charged automatically. Volume pricing and plans are there when you want them.\n\n**Koji is the right choice when you need to:**\n- Run customer discovery interviews at the speed of insight\n- Validate pricing without hiring a [pricing consultant](/blog/pricing-research-without-consultant)\n- Understand the real reasons customers churn (not just \"price\")\n- Test value propositions and positioning messages\n- Run Jobs-to-Be-Done switch interviews at scale\n- Conduct in-language research in 30+ languages\n\n## Head-to-Head: 7 Decision Factors\n\n### 1. Research Method\n\n**PlaybookUX**: Task-based. You write instructions (\"Find a blue jacket and add it to your cart\"). Participants complete the task while their screen and voice are recorded. You watch playback to find usability friction.\n\n**Koji**: Conversation-based. You write a discussion guide. Koji's AI moderator asks the questions, probes deeper based on each answer, and adapts the conversation to what each participant says. The output is a transcribed interview with synthesized themes.\n\n**Verdict**: PlaybookUX wins for *behavioral observation*. Koji wins for *qualitative depth and motivation*.\n\n### 2. AI Capabilities\n\nThis is where the gap is widest. PlaybookUX added AI personas in 2025 (Synthetic Participants), and offers AI-assisted summaries of session recordings. The AI helps *after* the test — it doesn't moderate the test itself.\n\nKoji's AI is the moderator. It conducts the interview, decides when to probe deeper, adapts to non-answers, and handles tangents gracefully. After interviews finish, the same AI synthesizes patterns across every conversation, generates [AI-powered insights](/docs/ai-generated-insights), and writes the report. See [AI-moderated vs human-moderated interviews](/blog/ai-moderated-vs-human-moderated-interviews) for why this matters.\n\n**Verdict**: Koji is AI-native end-to-end. PlaybookUX bolts AI onto a traditional usability testing pipeline.\n\n### 3. Pricing & Plan Flexibility\n\nPlaybookUX's lowest paid tier (Scale) costs **$5,400 billed annually** — roughly $450/month commitment. There is a pay-as-you-go option for individual tests, but ongoing research locks you into a yearly contract.\n\nKoji starts **free with 10 credits**, with interviews as low as **€1 per qualified interview** and **€3 per qualified voice interview**, then **pay as you go** with no subscription needed. No annual commitment required. For comparison, see the [user research budget template](/blog/user-research-budget-template-2026).\n\n**Verdict**: Koji is typically 60-90% cheaper for teams running ongoing discovery research, depending on volume. PlaybookUX makes more sense if you specifically need its 6M panel.\n\n### 4. Recruitment\n\nPlaybookUX includes recruitment from its 6M-person panel as part of paid plans. You can filter by demographics, occupation, behaviors, and tech use.\n\nKoji interviews your own lists (customer list, NPS detractors, churned users, free trial signups, sales prospects) and also has **panel recruitment built in**: describe the audience you need, and Koji returns a live per-respondent quote from Koji's panel. The quote is shown and paid in credits, and you approve it before launch (see [participant recruitment platforms](/blog/participant-recruitment-platforms-2026)). Koji is also ideal for customer-base research where panel recruitment misses the point.\n\n**Verdict**: Both recruit fresh participants fast. Koji wins for actual customer research: reach your own people with personalised links when you want to track who answered, or a generic anonymous link when candour matters, or let Koji recruit an audience you do not have, and either way the respondents land in AI-moderated interviews rather than task recordings.\n\n### 5. Analysis Depth\n\nPlaybookUX gives you video recordings, transcripts, heatmaps, click paths, and AI-assisted highlight clips. Analysis is largely manual — you watch sessions and tag insights.\n\nKoji **automatically synthesizes** themes, pain points, sentiment, quality scores, and feature requests across every interview. Researchers describe going from raw interviews to a polished report in hours rather than weeks. The [turning interviews into insights](/docs/turning-interviews-into-insights) guide walks through the workflow.\n\n**Verdict**: Koji is dramatically faster for qualitative synthesis. PlaybookUX is better if you specifically need session replay analysis.\n\n### 6. Multilingual Reach\n\nPlaybookUX supports English and select major languages.\n\nKoji conducts AI-moderated **voice and text interviews in 30+ languages** with native voice models — useful for global product teams running discovery in non-English markets.\n\n**Verdict**: Koji is the clear winner for multilingual research.\n\n### 7. Use Case Fit\n\n| If your question is… | Use… |\n| --- | --- |\n| \"Can users find the checkout button?\" | PlaybookUX |\n| \"Why did 30% of trial users not convert?\" | Koji |\n| \"Does the new IA match users' mental model?\" | PlaybookUX (tree test) |\n| \"What would make customers pay 2x?\" | Koji |\n| \"Which onboarding screen confuses people?\" | PlaybookUX |\n| \"Why are enterprise customers churning?\" | Koji |\n| \"Does the new homepage hero communicate value in 5 seconds?\" | PlaybookUX (five-second test) |\n| \"What jobs are customers hiring our product to do?\" | Koji (JTBD interviews) |\n\n## The Real Difference\n\nPlaybookUX is the right tool for **UX teams testing interfaces**. Koji is the right tool for **product, founder, marketing, and research teams uncovering customer truth at scale**.\n\nMost growth-stage companies need both — but they need them for different jobs. The mistake is paying $5,400/year for PlaybookUX and then trying to use task prompts to figure out *why* customers churn, or paying for an enterprise survey platform and trying to extract motivation from a 5-point Likert scale. Those tools were not built for those questions.\n\nKoji was built specifically for the *why*. Voice or text. Open-ended or structured. 5 interviews or 500. In any of 30+ languages. With automatic synthesis that makes the [time to insight](/docs/time-to-insight) feel almost unreasonable.\n\n## When to Use Both Together\n\nA common workflow we see at growth-stage SaaS companies:\n\n1. **PlaybookUX** runs unmoderated usability tests on the redesigned checkout flow → identifies a friction point at the address step\n2. **Koji** runs follow-up AI interviews with users who abandoned at that step → reveals that the friction isn't UX, it's a trust gap around how billing is described\n3. Marketing rewrites the trust copy, design adds a clarification microcopy → conversion lifts\n\nBehavior tells you *where* the problem is. Conversations tell you *what* the problem actually is.\n\n## Try Koji Free\n\nIf your team's research questions look more like *\"why did they churn,\"* *\"what would they pay,\"* or *\"what job are they hiring us to do,\"* — Koji is the right starting point. The free account includes 10 credits and one active study, with no credit card. Most teams launch their first study in under 20 minutes and have synthesized insights in 24-48 hours.\n\n[Start your first AI-moderated study free →](https://www.koji.so)\n\nFor more context, see [best AI user research tools 2026](/blog/best-ai-user-research-tools-2026), our [usability testing guide](/blog/usability-testing-guide-2026), and [customer discovery interviews for founders](/blog/customer-discovery-interviews-startup-guide).\n","category":"Research","lastModified":"2026-10-29T13:18:27.86031+00:00","metaTitle":"Koji vs PlaybookUX 2026: AI Interviews vs Unmoderated UX Testing","metaDescription":"Compare Koji and PlaybookUX in 2026. AI-moderated voice interviews and automatic theme synthesis vs unmoderated usability testing with a 6M-person panel — which research platform wins for your team?","keywords":["koji vs playbookux","playbookux alternative","ai moderated interviews","unmoderated usability testing","customer research platform","ux research tools 2026","ai user research"],"aiSummary":"Koji and PlaybookUX target different research problems. PlaybookUX is an unmoderated usability testing platform with a 6M-participant panel, best for observing task completion, card sorts, tree tests, and prototype usability. Koji is an AI-moderated customer interview platform — voice or text — that probes for the \"why\" behind behavior and automatically synthesizes themes across hundreds of conversations. Koji runs as low as €1 per qualified interview and €3 per qualified voice interview, starting free with 10 credits and no card; PlaybookUX starts at $5,400/year. Use PlaybookUX for behavioral observation; use Koji for customer discovery, JTBD, churn, pricing, and value-prop research.","aiKeywords":["ai customer research","unmoderated usability testing","playbookux comparison","voice interviews","thematic analysis","research methodology","question-to-insight workflow"],"aiContentType":"comparison","faqItems":[{"answer":"Koji runs AI-moderated voice and text interviews that probe for the 'why' behind customer behavior, then automatically synthesize themes. PlaybookUX is an unmoderated usability testing platform where participants complete tasks while their screen is recorded — designed for observing UX friction, not surfacing motivation.","question":"What is the main difference between Koji and PlaybookUX?"},{"answer":"Koji is substantially cheaper for ongoing research. Interviews start as low as €1 per qualified interview, and €3 per qualified voice interview. Koji has a free account (10 credits at signup, one active study, no card) and you pay as you go, with no subscription and no annual commitment. PlaybookUX's lowest annual plan is Scale at $5,400/year, with Pro at $8,800/year.","question":"Which is cheaper, Koji or PlaybookUX?"},{"answer":"No. PlaybookUX added Synthetic Participants (AI personas) and AI-assisted summaries in 2025, but the platform itself does not conduct moderated AI conversations. Participants complete preset tasks unmoderated. Koji's AI actively moderates each interview in real time.","question":"Does PlaybookUX have AI interviews?"},{"answer":"Use PlaybookUX when you specifically need unmoderated usability testing of a prototype or live site, card sorting, tree testing, five-second tests, or quick recruitment from its 6M-person panel. These behavioral observation methods are what PlaybookUX is built for.","question":"When should I use PlaybookUX instead of Koji?"},{"answer":"Yes — Koji supports usability research methodology as one of its 8 supported methods, including post-task AI follow-up interviews that probe why participants made specific choices. But for pure click-through task observation with screen recording, dedicated tools like PlaybookUX go deeper. Koji excels at the qualitative why, not the click path.","question":"Can Koji run usability testing?"},{"answer":"PlaybookUX is positioned as an affordable alternative to UserTesting with overlapping unmoderated/moderated usability testing functionality and recruitment panel access. UserTesting is enterprise-priced; PlaybookUX is mid-market. Both are behavior-observation tools, not customer interview platforms.","question":"Is PlaybookUX the same as UserTesting?"}],"relatedTopics":["ai customer research","unmoderated usability testing","customer interview platform","ux research comparison","playbookux alternative"]}],"pagination":{"total":1554,"returned":100,"offset":0}}