Customer Discovery Interviews: How to Get Honest Answers
A customer interview is the cheapest experiment in the validation toolkit and the easiest one to get wrong. Ask the wrong questions and you collect polite enthusiasm that tells you nothing. Ask the right ones and, within a dozen conversations, you can see whether a problem is real.
The core rule: talk about their life, not your idea
Rob Fitzpatrick's short book The Mom Test (2013) gave founders a memorable standard: ask questions that even your mother could not lie to you about. People are kind. If you describe your idea and ask whether they like it, most will say yes. If you ask how they handled the problem last time, what it cost and what they tried, they cannot flatter you, because they are reporting facts.
Three habits follow from that rule:
- Ask about specifics in the past, not generalities about the future. "When did this last happen?" beats "Would you ever...?"
- Talk less and listen more. If you are pitching, you are not learning.
- Chase commitments. At the end, ask for something small: an introduction, a follow-up call, access to a messy spreadsheet.
Who to interview
Start with the narrowest segment that you believe has the problem worst. Recruiting through your own network is fine for practice, but friends are the least reliable sources. Better sources are online communities where the segment already gathers, industry events, LinkedIn searches and people who have publicly complained about the problem. For B2B ideas, try to speak with both the user and the buyer; they often want different things.
How many conversations? There is no magic number. Many practitioners keep going until new interviews stop telling them anything new, which for a tightly defined segment often happens somewhere between 15 and 30 conversations.
Questions that work
| Instead of | Ask |
|---|---|
| Would you use an app that...? | How do you deal with this today? |
| How much would you pay for this? | What have you spent trying to fix it? |
| Is this a big problem? | What happened the last time it went wrong? |
| Do you like this feature? | What did you try before, and why did you stop? |
| Would your team buy it? | Who would need to approve a purchase like this? |
Reading the room: what poker players know about tells
Poker has a romantic idea of the "tell": the twitch or glance that gives away a bluff. Experienced players will tell you that physical tells are unreliable and that the information that really matters is how people bet. Over many hands, a player's actions, how much they put in, when and in what situations, reveal far more than their face does.
Customer interviews work the same way. Excitement, nodding and "that's a great idea" are the twitches at the table: easy to misread and easy to fake. What people have actually done with their money and their time is the betting pattern. Someone who has built a spreadsheet, hired a freelancer or paid for two competing tools to solve the problem has shown you their hand. Someone who says they would "definitely" pay but has never spent a minute on a workaround is, in poker terms, talking a big game with no chips in the pot. We look at more of these parallels in poker skills for founders.
Running the interview
- Open with context: you are researching a problem, not selling anything. This lowers people's urge to be polite.
- Ask them to walk you through the last time the problem came up, step by step.
- Dig into the cost: hours, money, stress, lost customers.
- Ask what they have tried and why it was not good enough.
- Close with a commitment ask and permission to follow up.
Take notes in the interviewee's words, not your interpretation. Afterwards, tag each note: problem confirmed, problem absent, workaround, budget, objection. Patterns across interviews matter more than any single vivid quote.
Warning signs
- Compliments without specifics.
- Hypotheticals ("I might", "I would probably").
- The problem only appears after you describe it.
- Nobody has tried any workaround.
- Strong interest from people outside your target segment, and indifference inside it.
Interviews will not prove demand on their own; they tell you which bet to test next. The natural follow-up is a landing-page smoke test or, if you need numbers from a wider group, a carefully designed validation survey.
