Truth-seeking series · № 11
Customer interviews
Your mom will lie to you. So will your customers.
Ask your mom whether your startup is a good idea and she'll say yes — she loves you. Strangers do the same thing for a cheaper reason: agreeing is the fastest way to end the conversation. Rob Fitzpatrick's The Mom Test is a set of rules for asking questions even polite liars can't ruin:
- Talk about their life, not your idea.
- Ask about specifics in the past, not opinions about the future.
- Talk less. You're there to collect facts, not applause.
Below are five questions founders actually ask. Each one feels reasonable — and each one manufactures false validation. See if you'd have caught them.
"Would you buy a product that did this?"
It's a hypothetical about the future — the exact kind of question people answer with hopeful fiction. Saying yes costs nothing today, so yes is what you'll hear. Their imagined future self buys everything.
✓ better"How do you deal with this today? What did the last time cost you?"
Past behavior and real costs can't be faked politely. If they've never spent time or money on the problem, that is your answer.
"Do you think it's a good idea?"
Opinions are worthless — this is the literal Mom Test failure. You've pitched, so now they're managing your feelings, not describing their reality. Compliments are data about politeness, not demand.
✓ better"Talk me through the last time you ran into this problem."
A story has verifiable texture: what triggered it, what they tried, where it stalled. No story means no problem — or not one they care about.
"How much would you pay for this?"
Imaginary prices for imaginary products. People are terrible at predicting their own spending, and generous with money they'll never actually part with.
✓ better"What are you paying to solve this now — in money, time, or workarounds? What else have you tried?"
The budget that already exists is the only pricing signal that's real. Solutions they've already tried tell you what the problem is genuinely worth.
"Would you use a feature that automated this?"
Feature fishing. You've named the solution, so the only polite answer is "sure, sounds useful." Everything sounds useful when it's free and hypothetical.
✓ better"What happened the last time this broke? Walk me through what you did next."
If the workflow they describe never hits the pain your feature solves, you just saved yourself a quarter of engineering.
"Don't you hate how long this takes?"
Leading the witness. You've handed them the complaint and asked them to nod. They will. You've learned nothing except that nodding is easy.
✓ better"What's the hardest part of your week around this?" — then stop talking.
Let them pick the pain. If they name a different problem than the one you wanted to hear, you've just been handed better information than any nod.
Now you try
Rewrite each question so a polite liar can't ruin it. There's no single right answer — the test is whether yours asks about past specifics instead of future opinions. Write yours before peeking.
"Would this dashboard be useful for your team?"
"How does your team answer this question today? Can you show me the last report someone actually made?"
"If we built an app for this, would you switch?"
"When did you last switch tools for anything? What finally made you do it?"
The uncomfortable summary
You're not in the meeting to collect validation; you're there to give the truth every chance to hurt you. Words are cheap, so weigh commitments instead: time, money, reputation, an intro, a deposit. Anything less is your mom saying she loves the idea.