The short answer
Good customer interview questions ask about specific past behaviour, not opinions or predictions. Ask people to walk you through the last time the problem happened, what they tried, what it cost and what they use now. Avoid pitching, avoid asking whether they would buy, take notes in their words, and synthesise across interviews by looking for repeated workflows and spend.
On this page (8)
The principle behind every good question
People are poor forecasters of their own behaviour and generous critics of other people's ideas. Ask 'Would you use this?' and you will hear yes far more often than the truth justifies, because yes is polite and costs nothing. Ask 'What did you do last Tuesday when this went wrong?' and you get a fact.
Rob Fitzpatrick's book The Mom Test is built around this idea: talk about their life rather than your idea, ask about specifics in the past rather than generics or the future, and listen far more than you talk. The test is that even someone who loves you and wants to protect your feelings could not mislead you with their answers, because you never asked for their opinion of your idea.
A 30-minute interview script
- Opening, 2 minutes — Thank them, say you are researching how people in their role handle a particular area, that there is nothing to sell, and ask permission to take notes.
- Context, 5 minutes — 'Tell me about your role and what a typical week looks like.' 'Where does [area] fit into that?'
- Last occurrence, 10 minutes — 'Can you walk me through the last time you had to deal with [problem]?' Follow with 'What happened next?' until the story is complete.
- Workarounds and spend, 6 minutes — 'What do you use for this today?' 'What have you tried before?' 'Roughly how much time or money does it take each month?'
- Priority, 4 minutes — 'Where does this rank against the other problems you deal with?' 'What happens if it just does not get done?'
- Close, 3 minutes — 'Is there anything I should have asked but did not?' 'Who else should I talk to?' 'Would you be open to seeing something rough later?'
Question bank by stage
You will not use every question in every interview. Pick the ones that fit the stage you are in, and let the conversation follow the most interesting thread.
| Stage | Goal | Example questions |
|---|---|---|
| Context | Understand their world | What does a normal week look like? Which tools do you open first each morning? |
| Problem | Find the real pain | What is the hardest part of [task]? When did that last go wrong? |
| Workflow | Map what they actually do | Walk me through it step by step. Who else is involved? Where does the information come from? |
| Workarounds | Find the real competition | What do you use today? What did you try before that, and why did you stop? |
| Cost | Size the pain | How long does this take each time? Who does it, and what else could they be doing? |
| Buying | Understand the purchase path | How did you end up with your current tool? Who signed off? What budget did it come from? |
| Priority | Separate annoyance from urgency | If you could fix one thing in your job this month, what would it be? |
Questions not to ask, and what to ask instead
- 'Would you use a product that does X?' — Invites a polite yes. Ask instead: 'How do you handle X today?'
- 'How much would you pay for this?' — A hypothetical number with no commitment. Ask instead: 'What do you spend on this now, in money or time?'
- 'Do you think this is a good idea?' — Asks for an opinion about you. Ask instead: 'What would have to be true for you to change how you do this?'
- 'Wouldn't it be great if...?' — Leading; the answer is in the question. Ask instead: 'What would make this easier?' and stay quiet.
- 'Would you buy this if it had feature Y?' — Turns the interview into a feature wish list. Ask instead: 'Why does Y matter? When did you last need it?'
- 'Do you ever have problems with X?' — Yes-or-no, and primes them to invent a problem. Ask instead: 'Tell me about the last time X happened.'
How to take notes you can actually use
If possible, have a second person take notes so you can focus on listening. If you record, ask permission first and still take brief notes, because recordings are rarely re-watched. Write as close to their actual words as you can, especially for descriptions of pain and workarounds; paraphrase quietly turns their problem into your hypothesis.
Mark each note with a simple tag as you go, for example: P for a problem, W for a workaround, $ for a cost or spend, B for something about buying, Q for a quote worth keeping, and F for a fact about their context. Straight after the call, spend ten minutes writing a short summary while it is fresh: the one most important thing you learned, what surprised you, and what you would ask differently next time.
Synthesising ten interviews into a decision
One interview is a story; patterns only appear across several. After each batch of five or so, put the tagged notes side by side and look for the same workflow, the same workaround and the same cost appearing independently. A simple grid works: one row per interviewee, columns for role, main problem, current workaround, spend, urgency and quotes.
Then answer a few blunt questions in writing. Did most people describe the same painful task in similar terms? Is there evidence of money or significant time already going into it? Can you name what they would stop using? If the answers are mostly yes, you have something worth testing further, perhaps with a landing page or a pre-sale. If the answers are scattered, either your segment is too broad or the problem is not as sharp as you thought. Narrow the segment and run another batch rather than forcing a conclusion.
Be especially wary of compliments, enthusiasm and feature requests. They feel like progress and are not evidence. The useful signals are facts about past behaviour and concrete commitments, such as agreeing to a follow-up, introducing a colleague, or offering to pay for an early version.
Your action checklist
0/8 doneSaved in this browser only. Nothing is sent anywhere.
Questions
How many customer interviews should I do?
Enough to see a pattern within one specific segment. For many early ideas that means somewhere around ten conversations with people in the same role. If you have done more than that and every person describes a different problem, the issue is usually that the segment is too broad, not that you need more interviews of the same kind.
Should I show my product or prototype in a customer interview?
Not in the first round. Early interviews should focus on the problem and current workaround. Showing a prototype too soon shifts the conversation to opinions about your solution. Once you understand the problem well, a separate round where you show a rough concept can test whether you have understood the job, ideally ending with a request for a real commitment.
What is The Mom Test?
The Mom Test is a short book by Rob Fitzpatrick about how to talk to customers without collecting misleading answers. Its core rule is to ask about the customer's life and past behaviour rather than your idea, so that even someone trying to be kind could not mislead you. It is widely recommended reading for founders doing customer discovery.
How do I find people to interview?
Start with your own network and ask for introductions to people in the exact role you are studying. Industry communities, professional associations, and people who write publicly about the problem are other good routes. Be clear that you are researching rather than selling, keep the request short, and offer to share what you learn. Always ask each interviewee who else you should speak to.
Should I pay people for customer interviews?
Small thank-you payments or gift cards are common and can help when reaching busy professionals. The risk is attracting people who are there for the incentive rather than because they have the problem, so screen carefully. For business buyers, a genuine interest in the problem and the promise of a summary of findings is often a better motivator than money.
Terms in this guide
Finished? Mark it so your paths and the library remember.