Answers.Founder questions, by Humble

What should I ask in a customer interview?

Ask about the last time the problem happened, what it cost and what they tried. Here are the questions, the ones to drop, and a free sheet for your notes.

3 min read · Updated 11 Nov 2026

The short answer

Ask about things that have already happened: the last time the problem came up, why it was hard, what it cost, what they've tried and what they dislike about those fixes. Leave out questions about whether they'd use or pay for your idea, because people answer those politely. Write down their exact words, and finish by asking who else you should speak to.

In a Startup School session that Y Combinator wrote up for founders, Eric Migicovsky gave five questions to ask the people you hope will become customers. Every one of them is about something that has already happened (Y Combinator).

That is the habit worth copying, because a story about last time shows you what the problem cost and what the person did about it.

The questions to ask

Migicovsky’s five, in our words:

  1. What’s the hardest part of the job you’re trying to help with?
  2. Tell me about the last time this problem came up.
  3. Why was that hard?
  4. What, if anything, have you done to fix it?
  5. What don’t you like about the fixes you’ve tried?

Rob Fitzpatrick’s book The Mom Test boils the same idea down to three rules, as Michael Lynch summarises them: talk about their life, ask about specific things in the past, and listen far more than you speak (Michael Lynch). The book also suggests asking what the problem costs them now, and ending every conversation by asking who else you should talk to and whether there’s anything you should have asked.

The questions to drop

Some questions feel useful and return nothing you can act on. Swap them like this:

Swap thisFor this
Would you use a tool that does this?How do you handle this today?
How much would you pay for it?What did this cost you the last time it happened?
Do you like the idea?What have you already tried?
Would this be useful to your team?Who dealt with it last time, and what did they do?

The left-hand questions ask people to predict their own behaviour, and most of us answer those kindly. The right-hand questions ask for a memory, which is much harder to dress up.

How to run one

Ask for the conversation without the pitch. Lynch notes the book’s warning that asking to “interview” someone sounds like a chore. A message asking for fifteen minutes about how they deal with the problem gets more yeses.

Keep your idea out of it. Migicovsky is clear that the point of the conversation is to learn. The moment you describe your product, the person starts reviewing it, and you lose the story.

Follow the cost. When they mention a lost afternoon, a bill or an annoyed client, stay there. Ask how often it happens and what they did next.

Write down their words. Keep the exact phrases in one column and your reading of them in another. A line like “I only notice what’s missing when someone leaves” will be far more use to you next week than a note that she seemed keen.

End with two questions. Who else should I talk to? Is there anything I should have asked?

What to write down

The Conversation Notes sheet below is Day 6 of Humble’s 30 days. It has a column for each part of the story and separate columns for what surprised you and what contradicted your assumptions, because those are the lines you’ll need when you decide what to do next.

It is not to sell them on using your product.
Eric Migicovsky on the purpose of a user interview, in Y Combinator's Startup School recap

How many to do

Three conversations with people who match your first customer is a good first target. That’s what Day 6 asks for, and it’s usually enough to see whether the problem turns up in the same shape each time. If you can only get two, write down what stopped the third.

If the stories line up, keep going until you have about ten. If they don’t, that’s useful too: it usually means the customer is still too broad, and the First-Customer Card is the place to narrow it.

When the answers surprise you

The surprising answer is the most valuable thing you’ll hear all week. Write it down in full, even when it undoes something you wrote on your Problem Receipt. Then bring the notes to Day 7, when you decide whether to continue, narrow, change or pause.

The Humble Edge · Sunday email

Get one of these every Sunday.

One founder problem and one thing to do before Monday.

Questions

How many customer interviews do I need?

Three is enough to see whether the problem turns up in the same shape for different people. If it does, carry on to about ten before you decide anything big.

Should I tell them about my idea?

Leave it until the end, if at all. Once you describe your idea, most people start being encouraging, and the conversation stops telling you anything about their week.

How long should a customer interview take?

Ask for fifteen minutes and keep to it. If the problem is real and recent, fifteen minutes is plenty to hear what happened last time and what it cost.

Can I run interviews over email or a form?

A call or a coffee works far better, because the useful parts come from follow-up questions. If email is all you can get, ask for one specific story about last time and reply with a follow-up.

What if nobody agrees to talk to me?

Ask about their problem and leave your idea out of the request. If that still fails, read reviews and forum posts where your customer complains in public, and label those notes second-hand.

Try something

While you're here

Next step

Is your idea worth 30 days? Find out free.

Five questions about your idea and a straight answer. It's week zero of Humble's 30 days, and it costs nothing.

More in validate