Talk to Ten People

Why it matters

Everyone will be nice to you when you describe your idea. That’s the problem. People — especially people who like you — say “that’s a great idea” and “I’d definitely use that” because it’s the polite thing to say, not because it’s true. If you build on top of that politeness, you find out it was fake after you’ve already spent weeks building.

The Mom Test is a simple fix: stop asking people what they think of your idea, and start asking about their life. Ask what they already do about the problem, what they’ve already paid for, what happened the last time it came up. People can lie kindly about the future (“I’d probably use that”) but it’s much harder for them to lie about the past (“here’s the app I actually use for this right now”).

Ten conversations is enough to see a pattern without taking forever. You’re not trying to convince anyone or sell anything yet — you’re trying to find out if the problem you picked is real for people who aren’t you. If seven of ten shrug and say it’s not really a big deal, that’s real information, and it’s worth having now instead of after you’ve built something.

This is half the Spark gate. The other half is turning what you hear into one clear sentence, next lesson.

What to watch for

Eric Migicovsky - How to Talk to Users

• Watch for the core rule: ask about specific past behavior, not hypothetical future behavior.
• Notice how he handles compliments, redirecting “I love this” toward “tell me about the last time this happened to you.”
• Skip trying to memorize a rigid script; the goal is a mindset you can apply to any conversation.

Startup Failure to Bestselling Author: The Story Behind “The Mom Test” | Rob Fitzpatrick

• Watch for why Fitzpatrick says people lie to you — usually to be kind, not to deceive you.
• Notice the difference he draws between a compliment and a real commitment (money, time, an introduction).
• Skip the parts about his own startup’s specific failure story; the lesson is the method he built afterward, not the history.

Start with Claude

Before you talk to anyone, use Claude to build your interview script. Then use it again after each conversation to help you log what you heard honestly, including the parts that don’t support your idea.

I'm running 10 Mom-Test-style conversations for the Spark stage of Make First, about [the problem I picked]. I need a 5-question interview script that asks about people's past behavior, not their opinion of my idea, and never pitches anything.

Ask me about the problem I'm investigating first, then help me draft the 5 questions one at a time, checking each one for whether it's actually asking about the past (what they've done, paid for, or currently use) rather than the future (what they'd want or might use). Don't write generic interview questions — make them specific to my problem. After I run each conversation, I'll come back and tell you what happened; help me log it honestly, including anything that doesn't support my idea.