How do you run a good customer interview?

The one rule that carries most of the value

Ask about the past, not the future.

Would you use a tool that automatically categorised your expenses? Almost everyone says yes. It costs nothing to agree, the described product sounds pleasant, and the respondent would like to be helpful.

Tell me about the last time you categorised expenses. Now you get a real story: when it happened, what triggered it, what they used, where they gave up, whether they finished. Buried in that story is whether the problem is frequent enough and painful enough to be worth solving — which is the thing you came to find out and the thing no hypothetical can tell you.

Questions that work

• Tell me about the last time you… — the opener for almost every interview.
• Walk me through what you did. Step by step, in order.
• What did you do before that? And after? The surrounding steps are where unmet needs hide.
• What were you using instead? Everyone has a current solution, even if it is a spreadsheet, a colleague, or doing nothing.
• How often does that happen? Frequency separates an annoyance from a business.
• What did you do when that went wrong? Workarounds are the most reliable signal of real pain — people only build them for problems that matter.

Questions to cut

• Would you use…? and Do you think…? — both are hypotheticals.
• How much would you pay for this? People cannot answer this accurately about a product that does not exist.
• Do you like this idea? Nobody tells a stranger their idea is bad.
• Any question naming your feature before they have described their problem.

Recruit for the situation, not the title

The most common recruiting mistake is screening on job title. What predicts a useful interview is whether the person is currently in the situation your product addresses. Somebody who wrestled with this problem last month is worth ten people with the right title who have never encountered it.

Headspace's early design work targeted a specific struggle — people who wanted to meditate and found existing tools intimidating and faintly embarrassing. That is a situation, and it made a far better screen than any demographic.

Handling the urge to pitch

You will want to explain what you are building. The moment you do, the interview is over — everything after it is feedback on your idea from someone being nice to you.

If you must, save it for the last five minutes and mark it in your notes as a separate mode. Everything said after the demo goes in a different bucket from everything said before.

Telling a useful interview from a pleasant one

A useful interview contains at least one thing you did not expect and did not want to hear. If every conversation confirms your plan, you are either asking leading questions or talking to people who like you. Intercom's jobs-to-be-done practice earned its reputation by producing exactly those uncomfortable findings — customers repeatedly described hiring the product for a job the team had not designed for.