{"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":"2026-10-27T00:23:19.442Z"},"content":[{"type":"documentation","id":"d37be63d-bf59-452a-8524-4bc2fadb7f7a","slug":"how-to-conduct-user-interviews","title":"How to Conduct User Interviews: The Complete Step-by-Step Guide","url":"https://www.koji.so/docs/how-to-conduct-user-interviews","summary":"Complete step-by-step guide covering planning, discussion guide writing, participant recruitment, interview facilitation, and qualitative analysis including data saturation and the 5-user rule.","content":"\nUser interviews are one-on-one conversations with real users to understand their needs, motivations, and behaviors. Done right, they're the most powerful tool in a researcher's toolkit. This guide walks you through every step—from writing your research goals through synthesizing your findings—and shows how modern AI tools are compressing weeks of research into hours.\n\n## What Are User Interviews?\n\nA user interview is a qualitative research method where a researcher has a structured or semi-structured conversation with one participant at a time. The goal is to understand the participant's world: their behaviors, goals, frustrations, mental models, and how they currently solve problems.\n\nUnlike surveys that measure *what* people do at scale, user interviews reveal *why* they do it—the motivations, contexts, and stories that explain human behavior.\n\n**Common interview types:**\n- **Generative interviews**: Discover unknown problems and unmet needs (best in early discovery)\n- **Evaluative interviews**: Assess whether your proposed solution addresses a real problem\n- **Jobs-to-be-done interviews**: Understand what \"job\" users are hiring your product to perform\n- **Contextual inquiry**: Interview users in their natural environment while they work\n\n## Why User Interviews Matter\n\nThe evidence for user research is clear:\n\n- Testing with just **5 users reveals approximately 85% of usability problems** in a design. The first user alone surfaces 31% of issues (Jakob Nielsen, Nielsen Norman Group).\n- Teams using iterative research cycles—three 5-user studies vs. one 15-user study—get dramatically better product outcomes because they fix problems between rounds. As Nielsen wrote: *\"The ultimate user experience is improved much more by 3 studies with 5 users each than by a single monster study with 15 users.\"*\n- **42% of startups fail because they build something nobody wants** (CB Insights)—the single largest cause of startup failure. User interviews are the primary defense against this.\n- Companies that prioritize customer feedback programs grow revenues **4–8% faster** than competitors (Bain & Company).\n- **70% of software features are rarely or never used** (Microsoft product data). Discovery interviews help you build the right 30%.\n\nAs Nielsen Norman Group researchers Maria Rosala and Kara Pernice wrote: *\"When performed well, user interviews provide in-depth insight into who your users are, their lives, experiences, and challenges. Learning these things helps teams identify solutions to make users' lives easier.\"*\n\n## When to Use User Interviews\n\n✅ **Use interviews when:**\n- You're in early discovery and don't yet understand the problem space\n- You need to understand users' mental models and decision-making\n- You want to uncover unmet needs before designing a solution\n- You need to explain a surprising quantitative pattern (\"Why is cart abandonment 40% at Step 3?\")\n- You're building personas or journey maps\n- You need to validate that a problem is real and worth solving\n\n❌ **Don't rely on interviews alone when:**\n- You need statistically significant data (complement with surveys or analytics)\n- You want to measure performance against benchmarks (add quantitative methods)\n- You need to know *how many* users have a problem, not just *why* they have it\n\n## Phase 1: Plan Your Study\n\n### Define Clear Research Goals\n\nVague goals produce vague insights. Before scheduling a single interview, define specific questions your research must answer.\n\n**Weak goals:**\n- \"Learn about our users\"\n- \"Understand why people use the app\"\n\n**Strong goals:**\n- \"Understand what triggers someone to start evaluating project management software\"\n- \"Identify the top 3 friction points in the onboarding flow\"\n- \"Discover what workarounds users have invented to compensate for missing features\"\n\nEach goal should directly inform a decision your team is about to make.\n\n### Identify Your Target Participants\n\nUser interviews derive their power from talking to the *right* people. Focus on behavioral criteria, not demographics:\n\n- **Required experience**: \"Must have evaluated or switched project management tools in the last 6 months\"\n- **Behavior of interest**: \"People who have tried and abandoned multiple checkout flows\"\n- **Screening question**: \"Have you signed up for a new SaaS tool in the last 3 months?\"\n\nWrite a screener survey to filter candidates before scheduling. This protects your research budget and ensures valid data. See the [Research Screener Questions guide](/docs/research-screener-questions) for templates.\n\n## Phase 2: Write Your Discussion Guide\n\nAn interview guide is NOT a script—it is a flexible framework. You do not read questions verbatim; you use them as landmarks while following the participant's natural story.\n\n### Guide Structure\n\n**1. Warm-up (5 minutes)**\nEasy, rapport-building questions with no wrong answers:\n- \"Tell me a little about yourself and your role\"\n- \"Walk me through a typical day at work\"\n\n**2. Context-setting (10 minutes)**\nUnderstand their current situation and behaviors:\n- \"How do you currently handle [the problem area]?\"\n- \"Tell me about the last time you did [Y]\"\n- \"What tools or methods do you use for [Z]?\"\n\n**3. Core topic exploration (25–30 minutes)**\nOpen-ended questions with prepared follow-up probes:\n- \"Tell me about a recent time when [core topic]. What happened?\"\n- \"What is the most frustrating part of [process]?\"\n- \"Walk me through exactly what you did when [event] occurred\"\n\n**4. Probing follow-ups**\nUse freely throughout:\n- \"Tell me more about that\"\n- \"What happened next?\"\n- \"How did that make you feel?\"\n- \"Why did that matter to you?\"\n- \"Can you give me a specific example?\"\n\n**5. Closing (5 minutes)**\n- \"Is there anything else you think I should know?\"\n- \"Is there someone else I should talk to?\"\n\n### Question Principles\n\n- Ask about **specific past events**, not general habits or hypotheticals\n- Use **open-ended questions** that invite stories, not yes/no answers\n- **Avoid leading questions** that embed assumptions\n- Keep questions **short and single**—never combine two questions into one\n\n**Always pilot your guide** with one colleague before launch. This catches confusing phrasing and calibrates timing.\n\n## Phase 3: Recruit Participants\n\nFor most research goals, recruit **5–8 participants** per distinct user segment. The 5-user sweet spot for cost-effectiveness is well established—five users reveals roughly 85% of usability issues at a fraction of the cost of a larger study.\n\n**Recruitment channels:**\n- **Panel recruit in Koji**: 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\n- **Your customer list**: Email existing users with a brief invite and incentive offer, and import the list so each person gets a personalized link and you can see who responded; share the generic link instead when responses should stay anonymous\n- **CRM segmentation**: Filter by behavior, churned users, power users, recent signups, and import the segment so the report is already segmented\n- **Embed or call**: put the interview inside your product, or have Koji call the people on a list\n- **LinkedIn**: Ideal for B2B research targeting specific job titles or industries\n- **Social communities**: Relevant Slack groups, Subreddits, or LinkedIn communities\n- **Research marketplaces**: If you prefer to run recruitment yourself, Respondent.io, User Interviews 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**Incentives**: Standard rates are $50–$150 per 45-minute session for consumers; $150–$300 for B2B professionals. Always compensate fairly—it signals respect and dramatically improves show rates.\n\n## Phase 4: Conduct the Interview\n\n### Setup Checklist\n- Record the session (with explicit consent)\n- Have a second person take notes so the interviewer can focus on listening\n- Use a quiet, distraction-free environment\n- For remote: camera on, good lighting, test audio first\n\n### The Interview Flow\n\n**Opening (5 min)**: Welcome the participant, explain the purpose, clarify you are testing the problem—not them—and ask for consent to record.\n\n**Warm-up (5 min)**: Easy questions to build comfort. Slow your speech and let natural pauses breathe.\n\n**Core exploration (30–35 min)**: Follow your guide but prioritize interesting threads over the script. The best insights often come from unexpected detours.\n\n**Active listening cues**: Lean in, nod, use brief verbal acknowledgments (\"I see,\" \"go on,\" \"interesting\") without endorsing or judging. Let silences hang—participants often fill them with their most revealing thoughts.\n\n**Closing (5 min)**: Ask if there is anything they wanted to cover. Thank them and explain how findings will be used.\n\n### What Great Interviewers Do\n\n- Listen 70%, talk 30%\n- Ask \"what happened?\" not \"what would you do?\"\n- Follow up on emotion: \"You mentioned that was frustrating—tell me more\"\n- Never pitch or describe the product you are building\n- Stay genuinely curious—the best researchers are always a little surprised\n\n## Phase 5: Analysis and Synthesis\n\n### How Many Interviews Before Saturation?\n\n**Data saturation** is reached when new interviews stop surfacing new themes. For most research questions this occurs between 8–15 interviews for a homogeneous population. For heterogeneous groups or complex topics, plan for 15–20.\n\n**Run iteratively**: after 5 interviews, pause to review patterns. If major new themes keep emerging, continue. If findings are converging, you likely have enough.\n\n### Analysis Process\n\n1. **Transcribe** interviews (AI transcription tools save hours)\n2. **Code** for themes: highlight quotes representing needs, pain points, behaviors, mental models\n3. **Cluster** codes into themes using affinity mapping\n4. **Identify patterns** across participants—what is universal vs. idiosyncratic?\n5. **Synthesize** into insight statements, personas, or journey maps\n\nLook for:\n- **Jobs to be done**: What task is the user ultimately trying to accomplish?\n- **Pains**: What is blocking or frustrating them?\n- **Workarounds**: What hacks have they invented to solve their problem?\n- **Emotional peaks**: What moments carry the most emotional weight?\n\n## Common Mistakes to Avoid\n\n| Mistake | What Goes Wrong | Fix |\n|---------|----------------|-----|\n| Asking hypotheticals | \"Would you use X?\" gets false positives | Ask \"Tell me about the last time you...\" |\n| Skipping the pilot | Guide questions are confusing or too long | Pilot with 1 colleague before launch |\n| No probing | Surface-level insights only | Build the \"tell me more\" reflex |\n| Researcher bias | Nodding at \"good\" answers shapes responses | Stay neutral; reward all answers equally |\n| Solo interviewing | Detail lost while facilitating | Always pair an interviewer with a notetaker |\n| Wrong participants | Biased or unrepresentative data | Use a screener with behavioral criteria |\n\n## The Modern Approach: AI-Moderated Interviews with Koji\n\nTraditional user interviews create a painful bottleneck: recruiting, scheduling, transcribing, and coding can take 2–3 weeks from brief to insight. Koji is an AI-native research platform that automates the time-consuming parts while preserving conversational depth.\n\n**How Koji changes the process:**\n\n- **Automated moderation**: Koji's AI conducts interviews 24/7—no scheduling required, participants answer when it suits them\n- **Intelligent probing**: When a participant mentions a pain point, Koji follows up contextually, drilling deeper the way a skilled interviewer would\n- **Structured + open questions**: Mix open-ended discovery questions with [6 structured question types](/docs/structured-questions-guide)—scale, single choice, multiple choice, ranking, and yes/no—to capture both qualitative depth and quantitative patterns in one study\n- **Automatic thematic analysis**: Koji identifies themes across all interviews automatically, no manual coding required\n- **Real-time reporting**: See insights emerge as interviews complete, not days later\n- **Voice and text modes**: Voice interviews for natural conversational depth, text interviews for async global reach\n- **Recruitment included**: Panel recruit finds the audience for you at a price you approve first, or reach your own people with personalized links, the generic anonymous link, an embed, or a call from Koji\n\nWhile traditional approaches might take 2–3 weeks from brief to insights, Koji users typically go from question to shareable report in under 48 hours. Teams using AI-assisted research report 60% faster time-to-insight compared to traditional manual methods.\n\n## Related Resources\n\n- [AI Interview Questions Generator: Write Better Questions Faster](/docs/ai-interview-questions-generator)\n- [Structured Questions Guide: 6 Question Types for Better Research](/docs/structured-questions-guide)\n- [Research Screener Questions: Finding the Right Participants](/docs/research-screener-questions)\n- [How Koji's AI Follow-Up Probing Works](/docs/ai-probing-guide)\n- [Note-Taking in User Research](/docs/note-taking-user-research)\n- [Research Interview Template Library](/docs/research-interview-templates)\n\n\n## Further reading on the blog\n\n- [How to Conduct Remote User Interviews: The Complete Guide (2026)](/blog/how-to-conduct-remote-user-interviews-2026) — Remote user interviews are now the default for most research teams. This guide covers everything — from recruiting and scheduling to running\n\n<!-- further-reading:blog -->\n","category":"Research Methods","lastModified":"2026-09-20T15:27:24.436965+00:00","metaTitle":"How to Conduct User Interviews: The Complete Guide (2026)","metaDescription":"Step-by-step guide to conducting user interviews: planning, discussion guide writing, recruiting, facilitation, and analysis. Includes Nielsen's 5-user rule, sample questions, and AI-powered approaches.","keywords":["how to conduct user interviews","user interview guide","conducting user interviews","user research interviews","discussion guide","user interview best practices","qualitative research interviews"],"aiSummary":"Complete step-by-step guide covering planning, discussion guide writing, participant recruitment, interview facilitation, and qualitative analysis including data saturation and the 5-user rule.","aiPrerequisites":["Basic familiarity with product or UX concepts","Understanding of what user research is"],"aiLearningOutcomes":["Plan a user interview study from scratch","Write an effective discussion guide","Recruit and screen participants","Conduct interviews with minimal bias","Analyze findings and identify themes","Determine the right sample size for qualitative research"],"aiDifficulty":"beginner","aiEstimatedTime":"20 min read"}],"pagination":{"total":1,"returned":1,"offset":0}}