A Customer Interview Is Not a Sales Pitch: Test the Problem Before You Build

Entrepreneurs often ask potential customers whether they like an idea. The answer may feel encouraging, but it is weak evidence. People are generous with compliments, optimistic about future behaviour and reluctant to criticise a founder who is visibly excited.

A useful customer interview does something different. It investigates a real problem before the founder becomes committed to one solution. The aim is not to win approval. It is to understand what people already do, what the difficulty costs them and whether the problem is important enough to change their behaviour.

Start with a decision, not a conversation

Before arranging interviews, write down the decision the evidence will inform. Are you deciding which customer group to serve, whether a problem deserves a prototype, which workflow needs support or what someone might reasonably pay? Without a decision, interviews become interesting conversations that produce notes but no direction.

The UK government’s business-plan guidance treats market and competitor information as part of planning, not decoration added after a product exists. A business plan will evolve, but it should connect the offer to credible evidence about customers, operations and finance.

Create a short assumption register before the first interview:

• Customer: who experiences the problem?
• Situation: when and where does it occur?
• Consequence: what time, money, risk or frustration does it create?
• Current response: what workaround or alternative is used?
• Priority: what evidence would show that this is worth solving now?

Recruit people with recent experience

Do not begin with friends simply because they are available. Recruit people who have faced the relevant situation recently. Someone who filed expenses last week can describe the sequence, tools and delays more reliably than someone imagining how they might file expenses next year.

Use a simple screening question: “When did you last do this?” A specific answer suggests lived experience. A vague answer may still be useful for exploring awareness, but it should not carry the same evidential weight.

A balanced sample should include more than enthusiastic early supporters. Speak to people who tried an alternative, abandoned the task, chose a competitor or decided the problem was not worth solving. Rejection data can reveal weak urgency, budget constraints or a mismatch between the founder’s language and the customer’s reality.

Ask for a story, not an opinion

“Would you use an app that organises your receipts?” invites a hypothetical promise. “Talk me through the last time you prepared receipts for your accountant” invites evidence. The second question can uncover where documents came from, how they were stored, who was involved, what went missing and what happened next.

Useful follow-ups include:

• What triggered you to begin?
• What did you do first?
• Which part took the longest?
• What did you try before this method?
• Who else had to help or approve something?
• What happened when the process failed?
• Did you pay for a tool, service or extra labour?
• What have you decided not to fix—and why?

These questions focus on observed behaviour. They do not prove future demand, but they provide stronger evidence than praise because they expose existing effort and trade-offs.

Keep the prototype hidden at first

If you show the solution too early, the interview changes. The participant begins reacting to colours, features and your presentation rather than describing the problem in their own terms. Spend the first part of the conversation mapping their current process. Only then, if appropriate, introduce a rough prototype to test comprehension and task flow.

When you do show it, give the smallest possible instruction: “What do you think this screen allows you to do?” or “How would you complete the task from here?” Avoid teaching the interface. If the participant succeeds only after an explanation, you have tested the quality of the explanation, not the product.

Separate evidence from interpretation

After each interview, divide the notes into three columns:

• Evidence: what the participant said they did, used, paid or abandoned.
• Interpretation: what you think the behaviour means.
• Open question: what still needs testing.

For example, “three people emailed themselves photographs of receipts” is evidence. “They need a dedicated receipt app” is an interpretation. The same behaviour might instead indicate that email is convenient, that existing tools are poorly integrated or that the problem is too small to justify another app.

This distinction protects the founder from confirmation bias. A pattern should be considered alongside contradictory cases, not built by discarding them.

Do not convert enthusiasm into fake demand

“That sounds brilliant” is not a purchase. Neither is a social-media like, survey tick or request to be kept informed. Stronger signals involve cost or commitment: a person gives time for a follow-up test, introduces a decision-maker, shares a redacted example, joins a genuine waiting list or agrees to a clearly described paid pilot.

Even these signals require caution. A deposit is stronger than an email address, but early buyers may not represent the wider market. Customer discovery reduces uncertainty; it does not eliminate it.

Record responsibly

Ask permission before recording audio or video, explain how the material will be used and collect only what the research needs. Do not request confidential employer documents or personal data merely because they would make a vivid case study. The Information Commissioner’s Office guidance on data-protection principles emphasises lawful, fair and transparent processing, purpose limitation and data minimisation.

Offer an interview that does not require recording. Store notes securely, remove identifying details where possible and set a deletion point. If a quote might appear publicly, obtain clear permission for that separate use.

Use a decision rule before the results arrive

Founders can always reinterpret mixed feedback to protect a favourite idea. Reduce that temptation by deciding in advance what will happen after the interviews. For example:

• Proceed to a prototype if repeated recent examples show meaningful cost and active workarounds.
• Narrow the audience if the problem is urgent for one group but occasional for others.
• Change the problem statement if participants consistently describe a different bottleneck.
• Pause if interest is polite but behaviour shows little urgency or willingness to change.

A pause is not entrepreneurial failure. It is a comparatively inexpensive result. Money not spent building the wrong product remains available for a better opportunity.

The founder’s job is to become less certain

The strongest customer interview does not end with a triumphant pitch. It ends with a more accurate map: who has the problem, when it appears, what people already do and which uncertainties still matter.

Build only after listening has changed something—your audience, your wording, your priority or your proposed solution. If every interview confirms exactly what you believed before it began, the research may have been designed to persuade rather than to learn.

---

Sources and practical guidance

• GOV.UK: Write a business plan
• British Business Bank: How to conduct market research
• Information Commissioner’s Office: Data-protection principles