Customer Discovery: The Right Way to Talk to Early Users

Customer discovery is the most important thing a pre-product founder can do — and the most commonly done wrong. Most founder interviews confirm what founders already believe, because they ask leading questions to people who want to be supportive. The result: confident founders building things no one actually wants.

The Mom Test: The One Rule That Changes Everything

Rob Fitzpatrick's 'The Mom Test' frames the problem perfectly: you can't trust what people say about your idea because they'll be nice. But you can trust what they say about their lives, their problems, and what they've already tried.

The rule: don't ask what people think of your idea. Ask about their actual behaviour. 'Would you use this?' is a bad question. 'How do you currently handle this problem?' is a good question. The difference is whether you're asking them to predict behaviour (unreliable) or describe behaviour (reliable).

10 Questions That Produce Honest Insights

• Tell me about the last time you [experienced the problem you're solving].
• What do you do currently when that happens?
• How much time does that take? How often does it happen?
• What have you tried before? Why did it stop working?
• How much does this problem cost you — in time, money, or stress?
• Who else is affected when this happens?
• How did you find the solution you're using now?
• What would a perfect solution look like to you?
• What's stopped you from solving this already?
• If this problem disappeared tomorrow, what would change for you?

What to Listen For

In a good discovery interview, the user talks 80% of the time. You're listening for:

• Workarounds: things people do manually because no good solution exists yet — these signal real pain
• Frequency: problems that happen daily are worth solving; problems that happen twice a year probably aren't
• Budget signals: 'I'd pay anything to fix this' or 'we currently spend £X on a bad solution' — these tell you willingness to pay
• Emotional language: frustration, embarrassment, anxiety — emotional problems produce motivated buyers
• Surprising context: details you didn't expect that reveal the problem is bigger or different than you thought

How Many Interviews Do You Need?

The minimum is usually 15-20. At 5-7 interviews, patterns start to emerge but you can't trust them — your sample might be an outlier. By 15-20, you'll hear the same core problems, workarounds, and contexts repeatedly. That repetition is your signal.

After the Interview: What to Do with the Data

Immediately after each interview, write down: the 3 most surprising things you heard, the exact phrases the user used to describe the problem, and any workarounds they mentioned. Don't summarise — quote. The specific language users use to describe problems becomes your copywriting.