Customer discovery

A 30-minute customer discovery interview script

A timed customer discovery script with moderator prompts, contingency branches, a worked example, and a practical post-interview debrief.

Instant Expert EditorialPublished 6 min read

A 30-minute discovery interview works best when it has one main job: reconstruct a relevant experience well enough to inform a decision. It is too short to explore every workflow, test a prototype, discuss pricing, and collect a company history with equal care.

The script below is an original, adaptable guide. Its timing is a planning aid, not a research benchmark. The worked scenario is fictional: a founder is learning how facilities teams coordinate non-emergency maintenance requests across office locations. No example response should be treated as a customer finding.

Before the call: choose the event and the uncertainty

Write one sentence describing what you need to learn. For the example: “Understand where responsibility becomes unclear between receiving a maintenance request and confirming completion.”

Choose a participant who has direct experience with that sequence. If you need to understand purchasing, recruit for that separately; a coordinator’s knowledge of daily work does not automatically include software budgets.

Prepare a short guide, check the meeting setup, and leave time afterward for notes. GOV.UK’s guidance on planning a research round includes preparation, practice, and arrangements for note-taking or recording. Read the preparation guidance.

Keep three items visible during the call: the main uncertainty, the event you want to explore, and the closing time. Everything else is optional.

Minutes 0–3: explain the conversation

Use an opening that lets the person understand the request:

“Thanks for joining. I’m learning how facilities teams handle maintenance requests across locations. I’d like to discuss one recent, non-emergency example, from the original request through completion. I’m interested in what happened, including anything that worked well. Please avoid confidential details, and feel free to skip a question or stop.”

Then explain who is present, whether you plan to record, how notes or recordings will be used, and what compensation was agreed, if any. Confirm that they understand the research and agree to participate before beginning the interview. Ask separately before recording.

This short opening does not replace the participant information you should provide in advance. GOV.UK’s consent guidance identifies purpose, recording, observation, data use, and voluntary participation as information participants need. Adapt the process to your organization and applicable requirements. Read the informed consent guidance.

Minutes 3–6: establish just enough context

Ask:

“What part of maintenance coordination do you personally handle?”

Then:

“Can you think of a recent non-emergency request you were involved in from the beginning, or from when it reached your team?”

Clarify the setting only as needed. How many people were involved in this event? Which location or kind of location did it concern? What responsibility did the participant actually hold?

Avoid spending this section collecting a complete organizational chart. You can ask about another person when they appear in the story.

If the participant did not directly handle a suitable event, establish the limit openly. You can explore their actual responsibility or end the interview appropriately; do not coach them into the role the study expected.

Minutes 6–17: reconstruct one event

Start with:

“What first brought the request to your attention?”

Follow the sequence. Use these prompts only where the answer needs detail:

  • “What happened next?”
  • “What did you need to know before you could act?”
  • “Where did that information come from?”
  • “Who took responsibility at that point?”
  • “How did you know the next step had happened?”
  • “What told you the work was complete?”

Draw a simple timeline in your notes. Mark the trigger, handoffs, decisions, and outcome. If the account jumps ahead, bring it back gently: “Before we get to completion, how was the contractor assigned?”

GOV.UK’s in-depth interview guidance recommends designing the discussion around the research questions and exploring relevant experiences. The timeline here is an original way to apply that approach to a bounded workflow. Read the interview guidance.

Minutes 17–24: examine the important exception or workaround

Choose the part of the story most relevant to your uncertainty. Suppose the fictional participant says two people contacted the same contractor. Do not immediately propose an assignment dashboard.

Ask:

“How did you find out both people had contacted them?”

Then trace the consequence: “What happened because of that?” Perhaps there was duplicate work. Perhaps it caused no difficulty. Perhaps the second contact was intentional. The answer should decide your next question.

Explore the response: “What did you do?” and “Is that how you would usually handle it?” If a tool appears in the account, ask what role it played rather than asking for a general rating of the tool.

This section is also where you can discover that a supposed problem is a useful safeguard or an acceptable tradeoff.

Minutes 24–28: test how far the example applies

Ask:

“How similar was this request to other ones you handle?”

Follow with a counterexample:

“Can you think of a request where responsibility was clear throughout? What was different?”

Do not ask the participant to estimate a precise frequency if they do not track it. Record their description and its basis. “I remember two examples” means something different from a count taken from a system.

If your main question remains unanswered, use these minutes to resolve the missing step. A second anecdote is less useful than finishing the sequence you already started.

Minutes 28–30: check your understanding and close

Offer a brief factual summary:

“Let me check the sequence: the office manager submitted the request, you assigned it, and completion was confirmed by a message from the location. Have I missed or misunderstood anything?”

Then ask what was important that you did not cover. Explain any agreed next step. If you want another conversation, ask separately rather than treating the first interview as permission for continuing contact.

End at the agreed time unless the participant explicitly wishes to continue and it fits the session arrangements.

Use these branches when the plan breaks

The answer becomes general: Ask for the last instance they can remember, without suggesting what must have happened.

The person offers feature ideas: Thank them, note the idea, and ask what experience prompted it.

A sensitive detail appears: Pause and redirect to the process at a level they can discuss comfortably.

Time is running out: Skip optional context and finish the main event plus the closing explanation.

The workflow works well: Investigate why. A reliable process is evidence about the conditions your product would need to improve on.

Immediately afterward: preserve the limits

Write a short debrief with the event, direct observations or reported facts, your interpretation, counterevidence, and questions still open. Label paraphrases as paraphrases. Mark uncertain dates and estimates.

The useful result of this half hour is a clearer account of one experience and a better-informed next decision. Compare it with other evidence before turning it into a claim about an entire customer segment.

The working model

A 30-minute interview with one main job

  1. 1

    0–3 · Explain

    Purpose, participation, data use, and agreement.

  2. 2

    3–6 · Context

    Responsibility and one suitable recent event.

  3. 3

    6–17 · Reconstruct

    Trigger, steps, handoffs, and outcome.

  4. 4

    17–24 · Examine

    One important exception or workaround.

  5. 5

    24–28 · Compare

    Typicality, counterexamples, and missing detail.

  6. 6

    28–30 · Close

    Check the sequence and explain next steps.

Original example schedule. Adapt timing to the participant, topic, and research question.