Customer Interviews That Change Startup Decisions

How do you run a useful customer interview?

Ask a potential customer to reconstruct a recent experience: what triggered the task, what they did, where it became difficult, and what happened next. Keep your proposed solution out of the opening discussion. A useful interview produces evidence that changes a decision about the customer, problem, offer, or next test.

For a founder, the hardest part is often resisting the urge to explain. Someone says a process is frustrating, and a product pitch starts forming. Stay with their story. The workaround they already use may reveal more than their reaction to an imaginary app.

Use this guide when: you have a possible audience and problem, but need to understand whether that problem deserves attention. Open the free interview decision worksheet, then save a copy to fill in. No email is required for this worksheet.

Pick a decision before you pick the questions

Write one sentence: “After these conversations, we need to decide whether to ____.” For example, a founder considering a scheduling product might need to choose between independent cleaners and cleaning companies with several crews. Those audiences can have different buyers, constraints, and consequences when scheduling fails.

Define who qualifies for the conversation using behavior. “People who scheduled multiple crews in the past month” is more useful than “busy business owners.” Include people who use competing tools, people who use spreadsheets, and people who decided the problem was not worth solving. Record how you recruited each person; your own network may share assumptions that a wider market does not.

Start with a small batch you can actually review. Five interviews can be a manageable working batch, but five is not a universal validation threshold or a representative market sample. Keep recruiting when important differences remain unexplained.

Use a 25-minute conversation structure

The following is an original working agenda, not a research standard. Adapt the timing to the person and task.

Explain what you are researching, how notes will be used, and whether you would like to record. Ask permission before recording. Let people decline questions or use a fictionalized example when their work contains confidential information.

Ask about behavior, consequences, and buying

Build follow-up questions from what the person actually says. Useful prompts include:

• What happened immediately before you started this task?
• Which step took the most attention, and why?
• What did you use to get through it?
• How often has that happened recently?
• What changed for the customer, team, or business afterward?
• Have you tried to improve it? What happened?
• Who would need to agree to a different way of doing it?

Ask for a walkthrough or a sanitized example when appropriate. A description of a spreadsheet can conceal five separate handoffs. Seeing the structure may reveal the real obstacle without requiring access to private customer records.

Money questions should follow context. Ask what alternatives cost, who controls the budget, and how a purchase is approved. An answer to “Would you pay $20?” is a stated intention. An actual purchase under clear terms is a different kind of evidence. Record the distinction.

Replace questions that invite agreement

Do not treat every complaint as an opportunity. Some annoyances are inexpensive to tolerate. Some problems are serious but occur too rarely to support your proposed business. Some users care deeply but have no authority to change the process. All three findings can save you work.

Separate the observation from your interpretation

Keep three columns in your notes: what the participant said or demonstrated, what you think it means, and what remains unknown. Mark direct quotations accurately; do not turn a paraphrase into a quote.

Fictional example: a crew manager shows that she copies schedule changes into a group chat because staff do not check the scheduling app. The observation is the copying behavior. “They need a new scheduling app” is an interpretation. Other explanations include notification settings, unreliable connectivity, or unclear responsibility for acknowledging changes.

The next test could be a simple acknowledgement process with the existing tools. If that resolves the difficulty, a replacement platform may be unnecessary. If it does not, investigate the remaining obstacle before adding features.

Turn a batch of interviews into one next test

Review evidence by customer type and recent episode, rather than counting every positive comment together. Look for recurring triggers, meaningful consequences, active workarounds, and access to a buyer. Look just as carefully for contradictory evidence.

Write a short decision memo: the strongest observation, the alternative explanation, the decision, and the next thing that could prove the decision wrong. If interviewees describe different problems, narrow the audience or separate the use cases. If the same problem appears repeatedly, test a bounded offer or prototype. Interviews alone do not establish demand at a particular price.

AI can help organize de-identified notes into candidate themes. Check every theme against the original notes and keep contradictions visible. A generated customer persona is a hypothesis, not a participant. Never add invented quotes or treat simulated interviews as customer evidence.

Questions founders ask

Should I show my product during the interview?

You can, after the open discovery section. Clearly separate the sections in your notes. Reactions to a concept answer a different question from observations about current behavior.

What if people say they love the idea?

Thank them, then ask what specifically fits their work and what would prevent adoption. Seek an appropriate next action, such as reviewing a prototype or joining a clearly scoped pilot. Enthusiasm is useful context, not proof of future revenue.

What should I do after the first batch?

Use your findings to choose a narrower test. The concierge MVP guide helps you deliver a result manually, while the first ten customers guide helps structure outreach and learning.

Further reading and your next step

Y Combinator’s Startup School recap with Eric Migicovsky emphasizes recent experiences, attempted solutions, and the difficulties people encounter. The agenda, tables, worksheet, and fictional example here are StartupCoach editorial tools for putting that principle into practice.

Open the interview decision worksheet and write the decision you need to make before booking your next conversation.

Need a little help applying this? Explore coaching and implementation.