20 Customer Discovery Interview Questions and Useful Follow-Ups

Good customer discovery interview questions help you reconstruct how someone handles a problem today. Start with a recent experience, follow the sequence of events, and ask what the person actually tried. That gives your team something more useful to investigate than a prediction about whether someone might use a future product.

Below is an original question bank and worksheet from Stetos.co. Adapt it to the decision you need to make. Select the questions that matter to your study and leave room for follow-ups; this is not a checklist to rush through in every conversation.

Download the customer discovery interview worksheet.

Define the learning goal

Write a sentence containing the audience, the experience, and the decision. For example: “We want to understand how small agency owners collect client approvals, so we can choose which part of that workflow to prototype.”

Next, define who has relevant experience. Someone who recently collected approvals can describe the work. Someone who likes the concept of an approval tool may have little evidence to offer about that workflow.

Keep a note of how you found each participant. A friend, an existing customer, and a person who rejected your product bring different context. Preserve those differences when you compare the interviews.

A suggested opening script

Explain who you are, the subject, how you will use the conversation, and any recording or data handling details relevant to your study. Confirm the person is comfortable taking part.

Here is a starting script to adapt:

We are learning how people handle client approvals. I would like to hear about your experience, including anything that did not work well. You can skip any question or stop. Before we begin, I will explain how we record and use the conversation and answer your questions.

For an AI interview, also explain that the interviewer is AI. Use your actual research and privacy arrangements in the final invitation. The GOV.UK guidance on informed consent explains the information participants need to understand.

Questions 1–4: establish context

1. What is your role in this workflow? Follow up on which steps the person owns and which depend on someone else.
2. When did you last need to do this? Ask for a specific recent occasion rather than a typical day.
3. What was happening that made it necessary? Explore the trigger: a deadline, a request, a problem, or a routine task.
4. What did a good outcome look like on that occasion? Ask how the person knew the work was finished or acceptable.

These questions establish the boundaries of the story. If the participant describes an approval process but someone else does the work, ask about their own involvement and consider interviewing that colleague separately.

Questions 5–8: reconstruct the experience

5. What did you do first? Let the participant establish the sequence in their own terms.
6. What happened after that? Repeat this where the sequence is unclear, rather than jumping to a prepared theory.
7. Which tools, documents, or people were involved? Ask what each contributed to the process.
8. At what point did you have to pause, repeat work, or change your approach? If nothing went wrong, explore what made that occasion straightforward.

A useful follow-up stays close to the answer. If someone says they “chased the client,” ask what that involved. Avoid translating it immediately into your preferred solution, such as automated reminders. The task might involve an unclear decision, a missing reviewer, or a disagreement that reminders would not solve.

Questions 9–12: understand the consequences

9. What happened because of that difficulty? Explore the effect on the actual task.
10. Who else was affected, and how? Distinguish what the participant observed from what they assume others felt.
11. How did you handle the delay or extra work? Ask about the workaround and whether it succeeded.
12. How often has a similar situation happened recently? Ask about the period the person can reasonably recall; record uncertainty rather than forcing a number.

Cost can include time, money, rework, a missed deadline, or an unacceptable outcome. Let the participant describe it before suggesting categories. If they estimate time spent, label it as an estimate. Later, you may decide to check it against operational records with appropriate permission.

Questions 13–16: explore alternatives and decisions

13. What have you already tried to improve this? Ask when and why they tried it.
14. What worked well enough that you kept it? Existing tools have strengths worth understanding.
15. What made you stop using or reject an alternative? Follow the actual evaluation rather than asking for a speculative feature wish list.
16. Who took part in choosing the current approach? Ask about approval, budget ownership, and constraints when relevant.

For B2B research, the person using a tool and the person approving the purchase may be different. An interview with one cannot stand in for the other. Keep user needs and buying conditions separately identifiable in the notes.

Questions 17–20: test your understanding and close

17. Which part of this experience would you most want to change? Ask what a better outcome would allow them to do.
18. What have I misunderstood or missed about how this works? Give the participant a real chance to correct the account.
19. Is there another recent example that played out differently? Look for circumstances that challenge your initial explanation.
20. Would you be comfortable with a follow-up conversation? Ask about contact permission separately and respect the answer.

If you plan to show a concept, separate that activity from the discovery portion and record when the participant saw it. Reactions after a demonstration have a different context from an unprompted description of the problem.

Rewrite questions that steer the answer

Instead of “Would automatic reminders solve your problem?”, ask what happened the last time a review stalled and how the participant responded.

Instead of “How frustrating is your current tool?”, ask which parts of the workflow went smoothly and which took more effort.

Instead of “Would you pay for a better solution?”, explore what they have already spent time or money on, who approved it, and what prompted the decision. A hypothetical price answer is not a purchase commitment.

These are proposed rewrites for this guide. The broader principle of asking open, neutral questions and exploring concrete experiences is covered in GOV.UK’s interview guidance.

Capture a useful record after each conversation

Use the following fields in your worksheet:

• Participant reference and relevant experience.
• Recruitment source and interview-guide version.
• Trigger and desired outcome.
• Steps, tools, and people involved.
• Difficulty and reported consequence.
• Current workaround and previous alternatives.
• Supporting excerpt with a session reference.
• Your interpretation, recorded separately.
• Contradictory evidence and unanswered questions.
• Proposed next investigation or experiment.

Avoid turning one memorable quote into a market-wide conclusion. Review patterns within the relevant participant groups and preserve cases that do not fit. Our customer interview analysis guide explains how to move from these records to a decision.

Use this guide in an AI interview

In Stetos.co, set the research goal, add business context, and use the selected topics to configure an agent. Test whether its follow-ups stay neutral and whether it allows uncertainty. Share the interview link with the people you chose, then review transcripts and summaries before expanding the study.

For a customer who has already left, use a more focused churn interview guide. For selecting the research method, see our guide to AI-moderated interviews.