By Mehdi HassanWhy do customer interviews lie to you?
Customer interviews lie to you when the questions invite opinions, predictions or compliments instead of facts. People want to be helpful and kind, so they say your idea sounds great and that they'd definitely use it. Then they don't. The interview wasn't dishonest, it was badly designed.
Three things cause most false signals. Compliments, because people don't like disappointing a founder to their face. Hypotheticals, because nobody can accurately predict what they'll do with a product they haven't used. Leading wording, because a question that contains the answer gets the answer back. Rob Fitzpatrick's book The Mom Test is built on this idea: ask questions that even your mother couldn't lie to you about.
This guide is one part of our user research for startups playbook. It focuses on the conversation itself: the script, the wording and the follow-ups. As a copywriter would put it, the questions are the product here, so every word matters.
How do you run customer interviews, step by step?
To run customer interviews, define the decision you need to make, recruit people who have the problem, prepare a short script, run 30 to 45 minute calls focused on past behavior, write up notes right away and synthesize across calls. The steps are simple, and the discipline is in not skipping any of them.
Write down the decision
One sentence: what will this round of interviews help you decide? For example, which of two customer segments to build for first.
List what you need to learn
Three to five learning goals, such as how people handle the problem today and what it costs them. These are for you, not questions to read aloud.
Recruit the right people
Screen for recent behavior: people who dealt with the problem in the last few months. Our guide to recruiting research participants covers sourcing and screening.
Prepare a short script
An opening, context questions, a 'last time' story, current solutions and a close. Keep it to one page so you can glance at it, not read it.
Run the call
Record with permission. Let them talk for most of the call. Follow the story, not the script order.
Write up within an hour
Capture key facts, quotes and surprises while you still remember the tone and context.
Synthesize after five to eight calls
Group observations into themes and connect each to a decision. Our guide to research synthesis shows how.
Which customer interview questions should you rewrite?
Rewrite any customer interview question that asks for an opinion, a prediction or a yes. Most bad questions are only a few words away from good ones. Swap 'would you' for 'when did you last', swap 'do you like' for 'what did you do', and swap 'is this a problem' for 'tell me about'.
Bad question | Why it misleads | Better question |
|---|---|---|
Would you use a tool that does X? | A prediction, and politeness says yes | How do you handle X today? |
Is X a big problem for you? | Leads them to agree it's a problem | Tell me about the last time X came up. |
How much would you pay for this? | Hypothetical money is free | What do you spend on X now, in money or hours? |
Do you like this idea? | Invites a compliment | What have you already tried to fix this? |
Wouldn't it be easier if…? | Contains the answer | What's the hardest part of doing X? |
What features do you want? | Users design poor solutions | What were you trying to get done when it went wrong? |
Do you usually do X weekly? | Generalizing hides reality | When did you last do X? Walk me through it. |
Why don't you use a tool for that? | Sounds like judgment, triggers defense | What made you settle on how you do it now? |
Notice the pattern in the right column: each question asks for a story about something specific that already happened. Stories contain details you can't fake, such as who was involved, which tools were open and what it cost. Those details are your evidence.
What should a customer interview script include?
A customer interview script should include a short opening that sets expectations, a few context questions, one or two 'last time' stories, a look at current solutions and a close that asks for referrals. It's a guide, not a questionnaire. You'll rarely ask every question in order, and that's fine.
A customer interview script you can copy (about 35 minutes)
Opening (3 minutes): 'Thanks for making time. I'm trying to understand how people handle [area], not selling anything today. There are no wrong answers, and the more specific the better. Is it OK if I record so I can listen properly instead of typing?'
Context (5 minutes): 'What's your role, and what does a typical week look like? Where does [area] fit into it?'
The last time (15 minutes): 'Tell me about the last time you dealt with [problem]. What triggered it? What did you do first? Who else was involved? Which tools did you have open? What was the most annoying part? How long did it take? What happened in the end?'
Current solutions (7 minutes): 'What have you tried to make this easier? What did you like about it? What made you stop, or keeps you using it? Have you paid for anything to solve this?'
Close (5 minutes): 'Is there anything I should have asked but didn't? Who else do you know who deals with this? Could I follow up if I have another question?'
Keep the opening honest. Saying you're not selling only works if it's true, so leave your pitch out of the call entirely. If they ask about your product, offer to show it at the end or in a separate call, after you've learned what you came for.
How do you ask follow-up questions without leading?
You ask follow-ups without leading by using short, neutral prompts that invite more detail without suggesting what the detail should be. The best follow-ups are boring. They keep the person talking about what actually happened, and they avoid planting words you'll later mistake for their own.
- 'What happened next?' Moves the story forward without steering it.
- 'Can you give me an example?' Turns a generalization into a specific event.
- 'What do you mean by [their word]?' Uses their language and checks what it means to them.
- 'How did that affect you?' Reveals the cost, in time, money or stress.
- 'Why was that?' Use it sparingly and gently, so it doesn't sound like a challenge.
- Silence. Count to five after they finish. People often add the most useful detail into the pause.
Watch your own wording as closely as theirs. 'How frustrating was that?' assumes it was frustrating. 'How did you feel about that?' lets them decide.
How should you take notes during customer interviews?
Take notes in three separate buckets: what they did, what they said they think and what surprised you. Mixing them together is how opinions sneak into your evidence. A fact like 'exports the report to a spreadsheet every Friday' carries far more weight than 'thinks reporting is clunky'.
Bucket | Example note | How much weight |
|---|---|---|
Behavior (did) | Rebuilds the client list by hand every month from three sources | High: real, specific, verifiable |
Opinion (thinks) | Feels current tools are too complicated | Medium: useful context, not proof |
Prediction (would) | Says they'd switch for a simpler tool | Low: treat as a guess |
Their words | 'I spend Friday afternoons fighting spreadsheets' | High for copy and messaging |
Surprise | Their manager approves every export | High: often the real insight |
If possible, have a second person take notes so you can focus on listening. Afterwards, those exact quotes become valuable beyond product decisions: they're the best source for marketing language. Our guide to conversion research methods shows how teams reuse interview language on their website.
What do you do with customer interview findings?
You turn findings into decisions by grouping observations across all your calls, ranking the themes and writing down what each one changes. A finding that doesn't change a decision is trivia. Aim for a one-page summary per round that the whole team can read in five minutes.
Structure helps here. Jobs to Be Done frames what you heard as the progress people are trying to make, which leads naturally to product and copy decisions. If you're deciding between building personas and writing job statements, our comparison of personas vs Jobs to Be Done helps.
When findings point to a specific design direction, a short, focused exercise is often the fastest way to test it. Our guide to running a design sprint shows how to go from interview insights to a tested prototype. If you want help turning findings into visual directions, our brand & design sprint is $3,000, fixed, and delivers 10 website and product explorations in Figma.
