How to run a customer discovery interview (without leading the witness)

The idea comes from Steve Blank’s customer development method, summed up in three words: get out of the building. You cannot learn whether a problem is real by staring at a spreadsheet or asking your co-founder. You learn it by talking to the people who live with the problem. But there’s a catch that Rob Fitzpatrick built an entire short book (The Mom Test) around: people are wired to be kind, so if you ask them about your idea, they’ll tell you what you want to hear. The whole craft of a discovery interview is getting truthful data out of polite humans. Here’s how.

Step 1 — Recruit the right five to ten people

You’re not looking for a representative national sample; you’re looking for people who plausibly have the problem. Start with the narrowest, most obviously-affected group you can reach — a specific role, situation, or community — because a clear signal from a narrow segment beats mush from a broad one. Recruit through communities where these people already gather, warm introductions, or a simple ask (“I’m researching how people handle X, can I buy you a coffee and ask about your experience?”). Aim for 5–10 to start; you can always do more once you know what you’re listening for.

Step 2 — Ask about the past, never the future

This is the entire game. Future and hypothetical questions (“Would you use…?”, “Do you think this is a good idea?”, “How much would you pay?”) invite people to imagine a flattering version of themselves. Past-behavior questions force them to report what actually happened. Your core script:

• “Walk me through the last time you dealt with [problem].”
• “What did you do about it?”
• “What did that cost you — time, money, stress?”
• “What have you already tried, and why did you stop using it?”
• “When’s the last time it really annoyed you? What happened?”

Notice none of these mention your idea. You’re mapping their real behavior, not auditioning your solution.

Step 3 — Shut up and let the silence work

The most common interviewing mistake is talking too much. Ask an open question, then stop. The uncomfortable three-second pause is where the real answer lives — people fill silence with the thing they didn’t plan to say. Aim to talk maybe 20% of the time. Follow up with “tell me more about that,” “why?”, and “what do you mean?” rather than steering. And write down their exact words — the language people use to describe their problem is gold for your eventual marketing.

Step 4 — Avoid the four data-poisoning traps

• Leading the witness. Any question containing the answer you want (“Wouldn’t it be great if…?”) is worthless. Ask neutral, open questions.
• Pitching too early. The second you describe your idea, the person switches from “honest reporter” to “polite reviewer.” Save any pitch for the very end — or skip it.
• Collecting compliments. “I love this” and “that’s so cool” are data about their manners, not their wallet. Fitzpatrick’s rule: compliments are the fool’s gold of customer conversations.
• Accepting fluffy generalizations. “I usually…” and “I always…” are stories, not facts. Anchor them: “Tell me about the last specific time.”

Step 5 — Read the results honestly

After a handful of interviews, look for repetition. When the same problem, the same clumsy workaround, and the same words keep surfacing, you’ve found a pattern worth acting on. The signals that matter are behavioral: people who’ve already spent money, cobbled together tools, asked around for a fix, or described real frustration. The anti-signal is a room full of pleasant nods with no history of anyone ever trying to solve it — that’s a polite way of learning the problem isn’t painful enough to build on. Either way, you learned it in a week of coffees instead of a year of building.

Where the free scorer fits (honestly)

Interviews are legwork, and nothing replaces the conversations themselves. What a structured scorer can do is hold your findings honest afterward — so a warm feeling from a few nice chats doesn’t quietly become “validated.” IdeaSweetSpot scores an idea live, free, on your device across four circles — Care, Capability, Need, and Paid path — and surfaces the weakest one, which is often exactly the thing your interviews should be probing next. Scoring is free and needs no account; the full exportable one-page build brief, the guided journey, and the multi-idea compare board are the Builder plan ($49/yr, cancel anytime). It’s a reflection and prioritization tool, not financial or legal advice, and it makes no promise of income or success.