The most useful customer interview questions ask someone to reconstruct a recent problem, the actions they took and the consequences they experienced. Avoid asking them to predict whether they would buy your idea before you understand the job they already do.
An interview is a way to learn, not a performance in which the founder earns approval. You can leave with less confidence in your proposed product and still have used the conversation well. The goal is a more accurate picture of the customer's situation.
Set an honest context
Explain who you are, what you are investigating and how the notes will be used. If you want to record the conversation, ask permission first. Make it easy to decline and avoid collecting information unrelated to the research question.
A simple opening might be: I am exploring how small teams prepare project handovers. I do not have a finished answer. I would like to understand how you handle it today, and you can skip anything you cannot share.
Nielsen Norman Group's interview guidance describes limitations of self-report, including imperfect recall and social desirability. An interviewer's confidence should not exceed the quality of the evidence they collect.
Use this sequence of questions
Start with context: what is your role in this process, and who else is involved? This prevents the founder from assuming that the person experiencing the problem is also the buyer.
Move to a recent event: can you walk me through the last time you completed this task? Ask what triggered it, where the information came from and what happened next. Allow pauses rather than supplying the expected answer.
Explore friction: where did the process become difficult? What did you do when that happened? Ask what the person tried, who helped and whether the issue changed the outcome.
Explore alternatives: what tools or services do you use today? How did you choose them? What works well enough that you would not want to lose it? Existing strengths matter as much as complaints.
Finally, explore decisions: when did your team last change a tool for this kind of work? Who was involved, what needed approval and what made the change worthwhile? This is more concrete than asking for an imaginary budget.
Replace leading questions
| Avoid | Ask instead |
|---|---|
| Would an AI assistant save you time? | Which part took the most effort last time? |
| Do you hate your spreadsheet? | What does the spreadsheet make easy, and where do you leave it? |
| Would you pay for this? | What have you already spent or committed to addressing the problem? |
| Is missing information a huge issue? | What happened when information was missing on the last project? |
| Would your team adopt our product? | What happened the last time your team introduced a new tool? |
These questions are an original practical script, not a validated research instrument. Adapt the language to the person and task. Do not turn a conversation into an interrogation simply to complete every row.
A hypothetical interview correction
Suppose a founder believes small podcast studios need automated client reminders. This example is hypothetical. The founder asks a studio manager how often clients forget recording details.
The manager instead describes last-minute guest changes. The hard work is updating everyone without sending contradictory instructions. If the founder keeps asking about reminders, they may miss the actual coordination problem.
A useful follow-up is: can you show how you handled the most recent guest change, using a redacted example? The answer may suggest a change log, a communication template or no product at all. The founder should preserve the original hypothesis and note how the evidence changed it.
Keep three kinds of notes separate
Record the participant's account, any observed material and your interpretation in separate fields. Mark quotations as quotations and avoid turning a paraphrase into a supposedly exact statement.
For example: the manager said they sent several updates; a redacted thread showed different recipients; the founder inferred that version confusion may be important. Those are related but different claims.
Strategyzer's guide to customer experiments distinguishes what people say from what they do. Follow the interview with an appropriate observation or task test when the decision warrants it. A conversation alone cannot establish repeated product use.
Close without manufacturing commitment
Ask what you misunderstood and whether there is someone else whose role would provide a different view. If a prototype test would answer a useful question, describe the request clearly. Do not treat a friendly conversation as consent to marketing or a public testimonial.
The counterargument is that interviewing can become a comfortable way to avoid selling or shipping. Set a decision for the research: which uncertainty must this conversation reduce? Stop collecting more of the same story when the next useful evidence requires a real task or purchase decision. Listening should change what you do next.