User Research for Product Managers: How to Run Customer Interviews Without a Research Team

Most product managers understand the value of customer research in theory. In practice, research often falls to the bottom of the sprint backlog — crowded out by design reviews, stakeholder meetings, and engineering discussions.

The result is products built on assumptions. Features that miss the mark. Roadmaps that prioritize the loudest internal voices over the quietest customer truths.

This guide is for product managers who want to run research consistently — without a dedicated research team, without scheduling 20 individual interviews across 20 different calendars, and without spending two days analyzing transcripts.

Why Product Managers Need Research Skills

The traditional model was that product managers set direction based on intuition, stakeholder requests, and quantitative data — with UX researchers conducting occasional deep-dive studies to supplement. That model is breaking down.

Research cycles that take 6–8 weeks cannot keep pace with 2-week sprints. Dedicated research teams are luxuries that most organizations cannot fund at scale. And quantitative data — clickthrough rates, conversion funnels, retention curves — tells you what is happening, not why.

PMs who can run fast, reliable research have a structural advantage. They make decisions based on what customers actually need, not what the team thinks they need. They kill the wrong features before they are built. They find the hidden activation moment that unlocks retention. They know why users churn and fix it before the data shows up in the dashboard.

With AI-moderated research platforms like Koji, running 20 customer interviews no longer requires 20 hours of PM time. It requires 2–3 hours: roughly 1 hour to design the study, and 1–2 hours to read and act on the report.

The PM Research Playbook

Here is how to embed research into your product process without a dedicated research team.

Step 1: Define the Research Question

Before opening any tool, be clear about what decision this research will inform. Vague questions ("What do users think of our product?") produce vague insights. Sharp questions produce actionable decisions:

• "Why do new users complete the setup flow but not activate within the first week?"
• "What is stopping our power users from sharing the product with colleagues?"
• "Why did users who churned in Q4 leave, and what would have changed their mind?"

A useful test: your research question should complete this sentence — "After this research, I will be able to decide whether to ______."

Step 2: Set Up Your Study Brief in Koji

In Koji, every study starts with a research brief that structures the AI interviewer's approach. The brief has four sections:

Problem Context: Your problem statement, the decision this research will inform, your current hypothesis, and what success looks like. Being explicit about your hypothesis is critical — it is what the AI will attempt to validate or invalidate throughout the interview.

Target Participant: Who should you talk to? Koji asks for required experience ("Has used the product for 30+ days and reached the activation point"), behavior of interest ("Tried at least one collaboration feature"), and a screening question ("Have you invited a teammate to your workspace?"). The more precise your participant criteria, the more relevant your data.

Methodology: Koji supports built-in research methodologies including the Mom Test (talk about their life, not your idea), Jobs to Be Done (understand what progress they are trying to make), and Customer Discovery (validate problem-solution fit). Selecting a methodology loads the AI with the corresponding interview principles, question patterns, and anti-patterns — so even a first-time researcher gets an interview that follows best practices.

Interview Plan: Your key questions, topics to explore, behaviors to probe, and guardrails. This is where you add structured questions — see Step 3.

Step 3: Write Questions That Give You Both Data and Stories

One of the most common PM research mistakes is writing only open-ended questions. Open-ended questions generate great qualitative data, but make it hard to compare across participants. Mixing question types solves this problem.

Scale questions (1–10 ratings) let you measure satisfaction, confidence, or effort quantitatively — then probe for the reasoning. "On a scale of 1–10, how confident did you feel during the onboarding process?" followed by "Tell me what would have made that a 10" gives you a trackable number and a roadmap for improvement.

Yes/no questions check specific hypotheses efficiently. "Before you started using this product, were you doing this manually in a spreadsheet?" verifies your switching story in under 10 seconds, then AI probing captures the specific details.

Single choice questions surface segmentation. "Which team do you work in — Engineering, Product, Design, or Marketing?" lets you cut your report data by role and look for patterns.

Open-ended questions drive discovery. "Walk me through how you handled this task before you started using our product" surfaces the qualitative context that gives numbers meaning.

A well-designed PM research study has 6–10 questions: 2–3 quantitative (scale, yes/no, single/multiple choice) and 4–6 qualitative (open-ended). Koji's AI handles sequencing, probing, and transitions naturally across all question types.

For a deeper look at all six Koji question types and when to use each, see the Structured Questions Guide.

Step 4: Recruit Your Participants

You have three recruitment options in Koji:

CRM import. Upload a CSV or connect your CRM to pull in segmented customer lists. Koji generates personalized interview links for each participant, so you can target churned users, new activations, power users, or any other segment your CRM defines.

Shareable link. Paste the interview link into an email, in-app notification, Slack message, or customer community. Anyone who clicks completes the interview on their own device and timeline — no scheduling required.

Personalized links. For B2B or key account research, send individual links with the participant's name and context pre-filled, making each interview feel tailored rather than generic.

Step 5: Let Koji's AI Run the Interviews

Once participants start clicking your interview link, Koji handles everything. The AI introduces itself, explains what the interview covers, asks your key questions, probes follow-ups, and closes gracefully. Participants can complete it in 15–20 minutes on any device, at any time.

You do not need to be present. You do not need to schedule anything. And unlike a static survey, every participant has a genuine conversation — the AI adapts its follow-up probing to what each participant says, surfacing context you would never get from a checkbox form.

Step 6: Read the Report and Act

After collecting responses, Koji generates an AI research report that includes:

• A synthesized executive summary of the key findings across all participants
• Per-question analysis: thematic summaries with representative quotes for open-ended questions, distribution charts for scale, yes/no, and choice questions
• Individual interview summaries if you want to go deeper on specific participant responses
• The ability to ask follow-up questions about the data using Koji's Insights Chat

As a PM, your job is to take the report and answer one question: "What did we learn, and what are we going to do about it?" Share the report directly with engineering, design, and stakeholders via a published link — no login required for viewers.

Research Templates for Common PM Jobs

Activation research: Why do newly signed-up users not return after day 1?

• Opening: "Walk me through what happened when you first signed up for the product."
• Include: scale question for first impression, yes/no for whether they completed setup, open-ended for what blocked them from returning

Churn research: Why did users cancel or stop engaging?

• Opening: "When did you realize this product wasn't going to work for you?"
• Include: open-ended for the moment of decision, yes/no for whether they tried an alternative, single choice for what they switched to

Feature prioritization: Which unbuilt features would users actually use?

• Use Jobs to Be Done methodology in Koji's brief settings
• Focus on past behavior: "Walk me through how you currently handle this task"
• Scale questions to measure how painful the current approach is

Competitor research: Why do users choose you vs. alternatives?

• Opening: "When you were evaluating options, what alternatives did you consider?"
• Include: ranking question for the most important decision factors, open-ended for the deciding moment

Integrating Research into Your Sprint Cycle

Research does not need to be a big, planned event. With Koji, you can run micro-research between sprints:

Pre-sprint (30 min setup, 24-48 hours collection, 30 min to read): Run 5–8 interviews to validate the direction of the next sprint before the team commits to building.

Post-launch (2 weeks after shipping): Run 10–15 interviews with new users. Do they understand the feature? Are they using it the way you designed it? What are they doing instead?

Quarterly: Run a larger study (20–30 interviews) to reset your understanding of the customer and update your core assumptions about their pain points, behaviors, and alternatives.

The key insight is that research becomes a habit, not an event. The compound advantage of consistent customer contact — even just 5–8 interviews per sprint — is enormous over 6–12 months.

Common PM Research Mistakes (and How to Avoid Them)

Asking hypothetical questions. "Would you use a feature that did X?" People are unreliable predictors of their own future behavior. Ask about what they have actually done. "Have you ever tried to do X?" is a stronger question than "Would you pay for something that does X?"

Pitching instead of listening. The purpose of customer interviews is to understand the customer's world, not to get them excited about your solution. Koji's AI-moderated format enforces this discipline — the interviewer stays curious and question-driven by design, without the conversational drift that affects human-moderated sessions.

Only talking to enthusiasts. Your most engaged users are not your typical users. Talk to people who tried your product and left. Talk to people who fit your target persona but are still doing the thing manually. These are often your most valuable conversations.

Treating research as a one-time event. The teams with the deepest customer understanding run research continuously — not as a quarterly initiative, but as a background process that never stops.

Frequently Asked Questions

Q: I am not a trained researcher — will my study design be effective? Koji's brief builder and methodology frameworks do substantial heavy lifting. The Mom Test principles embedded in Koji's Customer Discovery methodology will steer you away from leading questions and toward past-behavior questions automatically. A well-completed Koji brief produces good interviews even without formal research training.

Q: How many interviews do I need for a product decision? For directional decisions, 6–8 interviews with the right participants often surface enough themes to act on. For significant product bets, 15–20 is more reliable. For quantitative benchmarks like NPS or satisfaction scores, aim for 30+ to get statistically meaningful distributions.

Q: What if participants give very different answers? Variance is a signal, not a problem. If half your churned users cite price and the other half cite missing features, that is telling you something important about your customer segments. Look for patterns within the variance — who says what, and why?

Q: How do I get busy customers to complete an interview? Koji's async model removes the biggest barrier: scheduling. Busy users will complete a 15-minute interview on their own time when they would never block 30 minutes for a calendar invite. Personalized links that address participants by name and explain why their feedback matters increase completion rates significantly.

Q: Can I share Koji interview reports with stakeholders who do not have an account? Yes. Published Koji reports are accessible via a shareable link with no login required. You can send the report URL to your entire engineering team, present it in a stakeholder meeting, or embed it in a Notion document.

Q: How much does it cost to run PM research in Koji? New accounts receive 10 free credits on signup. Text interviews cost 1 credit each; voice interviews cost 3 credits each. The Insights plan (€29/month) includes 29 credits — enough for 15–29 text interviews per month, which is more than most PMs currently run.

Related Resources

• Structured Questions Guide: Mixing All 6 Question Types for Better Research
• How to Write User Interview Questions That Surface Real Insights
• Customer Discovery Interviews: The Complete Guide
• Feature Adoption Research: Interviewing Users Who Are Not Using Your Product
• Research-Driven Roadmap Prioritization
• Reading Your Koji Research Report