ValidationAugust 6, 2026By Maarten Follaets

What to ask in customer interviews (questions that get honest answers)

Cover reading Ask about their life, not your idea, with a callout quoting The Mom Test by Rob Fitzpatrick and the Foxy mascot in the corner

Quick answer: In a customer interview, ask about the person's life and their past behaviour, never about your idea. The best questions are specific and about things that already happened: what they did last time the problem hit, what it cost them, what they tried, what they pay for today. Those return facts. Bad questions are hypothetical and flattering ("would you use this", "do you like it", "how much would you pay"), and they only return politeness. Rob Fitzpatrick calls this The Mom Test: ask questions so grounded in the past that even your mum could not fib to protect your feelings.

The problem with customer interviews is not that founders skip them.

It is that we run them to feel good, not to learn something.

Why most interviews lie to you

You go in wanting one thing, even if you will not admit it: for the person to say the idea is great. So you ask questions that make that easy. "Would you use a tool that did X?" "Do you think this is a good idea?" And they, being kind, say yes.

Now you have data that feels like validation and is actually noise.

The fix has a name. In The Mom Test, Rob Fitzpatrick argues that you should talk about the customer's life instead of your idea, ask about specifics in the past instead of opinions about the future, and talk less so you listen more. The title is the whole trick: ask questions so anchored in facts that even your mum, who wants you to be happy, could not warp her answer to spare your feelings.

Applause is free. Behaviour is data.

The one rule: ask about their life, not your idea

Every good question you can ask comes from the same move. Stop talking about the thing you want to build. Start asking about how they live now.

The moment you describe your solution, you have poisoned the well. People stop reporting their world and start reacting to yours. So keep your idea in your pocket. You are not there to pitch. You are there to find out whether the problem you imagine is real, sharp, and expensive enough that someone already tries to solve it.

That is the same discipline behind customer discovery interview questions that do not lead the witness: the questions do the work, so you do not have to steer.

Questions that get honest answers

These all point at the past or the present, never a hypothetical future.

"Walk me through the last time you ran into this. What happened?"
"What did you do about it? What did you try before that?"
"How often does this come up?"
"What does it cost you when it happens, in time or money or stress?"
"What are you using for this today, and what do you hate about it?"
"Who else is involved when this goes wrong?"
"The last time you looked for a fix, what did you search for?"

Notice what they have in common. None of them mention your idea. Every one can only be answered with a story about something that already happened. You cannot flatter your way through "walk me through the last time". There either was a last time, or there was not, and both answers are useful.

Questions to never ask

These feel like research and are really just fishing for a compliment.

"Would you use this?"
"Do you like the idea?"
"How much would you pay for it?"
"Would you pay X a month for it?"
"Don't you think it's annoying when...?"

The first three ask people to predict their own future behaviour, which humans are terrible at and generous about. The last two hand them the answer you want. A hypothetical yes costs nothing to give and is worth exactly that. If you want to know whether someone will pay, do not ask; that is a separate test of willingness to pay, and it needs a real commitment, not a spoken promise.

How to run the conversation

Keep it short and let silences sit. Your job is to talk less than a third of the time. When they say something vague ("it's just a pain"), dig for the specific ("when did that last cost you money?"). When they compliment the idea, gently steer back to their life: "that's kind, but tell me what you actually do today."

And listen for the two things that matter more than any yes: a problem that shows up again and again, and a workaround they already pay for in time or cash. A person who has cobbled together a spreadsheet, three tools, and a Sunday-night ritual to cope is telling you something a hundred polite yeses cannot.

You are not collecting opinions. You are collecting evidence of behaviour.

Where Foxy fits

Here is the hard part, and I will be honest that I have got this wrong myself: you are the worst-placed person to hear your own interviews cleanly, because part of you needs them to go well. You will remember the compliments and forget the shrug.

That gap is what we are building Foxy to close. It is an AI co-founder whose job is not to cheer, but to read your interview notes and point out where the answer was flattery rather than fact, and which questions you quietly led. If you want an objective second read on your conversations before you bet months on them, start here.

So before your next call, one question for you. Are you going in to learn what is true, or hoping to hear that you are right?

Frequently asked questions

What should I ask in a customer interview?

Ask about the person's life and their past behaviour, not your idea. Good questions are specific and about things that already happened: what they did last time the problem hit, what it cost them, what they tried, what they pay for now. Those return facts. Hypothetical, flattering questions return politeness.

What questions should I never ask?

Anything that invites a polite yes. 'Would you use this?', 'Do you like the idea?', and 'How much would you pay?' all push the other person to guess about a hypothetical future and to be nice to you. Rob Fitzpatrick's rule in The Mom Test is to talk about their life instead of your idea.

How many customer interviews do I need?

Fewer than you think if the questions are good. You are listening for the same specific pain and the same expensive workaround to repeat across people. When new interviews stop surprising you, you have enough signal on that question. Ten focused conversations often beat fifty vague ones.

Should I tell them my idea in the interview?

Not at the start, and ideally not at all in a discovery interview. The moment you describe your solution you contaminate the data, because people react to you rather than tell you the truth. Save the idea for a separate test where you ask for a commitment, not an opinion.

How do I stop leading the witness?

Ask about the past, not the future, and about them, not your idea. 'Tell me about the last time this happened' cannot be answered with flattery. 'Would you love a tool that fixes this?' can only be answered with flattery. Keep the questions grounded in things that already occurred.

What counts as a good answer versus a bad one?

A good answer is a concrete story with details you did not prompt: a date, a cost, a workaround, a moment of annoyance. A bad answer is a compliment or a hypothetical ('I'd definitely use that'). Compliments feel great and tell you nothing. Facts about past behaviour tell you almost everything.