ajayyadav.dev
All books
Book · Nov 2026

The Mom Test

by Rob Fitzpatrick

Customer discoveryStartups

By Rob Fitzpatrick · about 130 pages · roughly 3 hours to read

If your mom couldn't lie to you in the conversation, you asked the right questions.

The problem it solves

Ask anyone if your idea is good and they'll say yes. Your mom, your friends, strangers. Not because they're lying, but because they're being nice.

That's the trap: you leave ten conversations feeling great, build for six months, launch, and nobody shows up. The compliments were sincere. They just weren't evidence.

The fix isn't "ask better questions about your idea". It's stop mentioning your idea at all, and ask about their life instead.

The 3 rules

  1. Talk about their life, not your idea. "Do you think a grocery-tracking app is a good idea?" gets you encouragement. "Walk me through the last time you felt you overspent on groceries" gets you facts.
  2. Ask about specific past events, not the future. "Would you buy this?" is worthless. "Do you usually…?" is nearly as bad, because people describe who they wish they were. "What did you do last Tuesday?" is where the truth is.
  3. Talk less, listen more. Aim for them talking about 90% of the time. A pitch triggers politeness; silence triggers detail.

The book's example: a mother says she'd love a healthy-eating app. Ask about the past and it turns out she downloaded three, used each for four days, and still orders takeaway twice a week. Nothing she said was dishonest. It just predicted nothing.

Swap these questions

❌ Don't ask✅ Ask insteadWhy
Does this happen often?Talk me through the last time it happened.A story has dates, tools and costs. A frequency guess has none.
Would you use a tool for this?What else have you tried?Past attempts prove the pain. A hypothetical proves nothing.
Is this a big problem?What happens because of it?Separates a real cost from a minor annoyance.
Would you pay for this?What are you doing about it now?Money or time already spent is the only budget signal that holds.
Do you like my idea?Who else should I talk to?An introduction is a commitment. Praise is free.
Do you want feature X?Why do you bother?Finds the motivation behind the request.

The worst question in the book: "Would you buy a product that did X?" Everyone says yes, so the answer tells you nothing.

Three kinds of fake progress

These feel like progress. They aren't.

  • Compliments ("Cool idea!", "I love it"): they show up when you pitched instead of asked. Deflect: "Thanks, but how do you handle this today?"
  • Fluff: generics ("I always…"), hypotheticals ("I would…") and promises ("I'll definitely…"). Pull them back to a real example: "When did that last happen? Walk me through it."
  • Feature requests ("Can it export to CSV?"): don't take the feature, find the reason. A CSV request usually means someone rebuilds a report by hand every Monday. That report is the real product; the CSV is one guess at it.

Get a commitment, not a compliment

A conversation that ends well but goes nowhere is a failed conversation. The test: did they give up something they value?

CurrencyWeakStrong
Time"Send me something."A booked session to walk through their workflow
Reputation"I'll keep you in mind."An intro to two people with the same problem, by name, this week
Money"Let me know when it launches."A pre-order, a deposit, a letter of intent

"Send me a link when it's ready" means no. Ask for the commitment and let them say no. A clean no is useful. An endless maybe (a "zombie lead") costs months.

Who to talk to

"Everyone" isn't a customer. Keep narrowing by who, where and what until you have a group you can actually find, that has the problem badly, and that can pay.

"Young professionals" is useless. "Freelance video editors who bill clients hourly and lose an hour a week to timesheet admin" is a group you can go and find on Tuesday.

Go where the problem already lives (communities, forums, comment threads), ask for advice rather than feedback, and close every conversation with "who else should I talk to?"

How to run it

  1. Write your 3 big questions before the call: the three things that, if you learned them, would change what you build.
  2. Aim at your riskiest assumption: the one that kills the plan if it's wrong. Don't spend the call on parts you're already confident about.
  3. Write down their exact words, especially numbers and emotional words. "It's annoying" and "I lose a day a month to it" are different findings. Keep commitments separate from opinions.
  4. Stop when you stop being surprised. The signal is repetition, not a magic number of interviews.

A worked example: a creator tool

The idea: a tool that turns a list of text into on-screen visuals for reels.

The wrong conversation: "I'm building a tool that turns your bullet points into reel visuals. Would you use it?" → "Yeah, sounds useful!" You learned nothing.

The right conversation (no mention of the tool):

  • "Talk me through making your last list-style reel, from the moment you had the idea."
  • "How long did the on-screen text part take? What did you make it in?"
  • "Have you ever dropped an idea because the editing took too long?"
  • "What are you paying for in that workflow right now?"

What the answers mean:

You hearIt means
"About 20 minutes in Canva, it's fine."No pain worth paying for. Wrong customer, or wrong problem.
"Two hours, and I dropped three ideas last month because of it."A real cost, already measured in lost output.
"I pay for a template pack and still redo half of it."Money already spent plus an unmet need: the strongest signal there is.
"I'd love something that automates it."Fluff. Ask what they tried last time.

Then ask for a commitment: "Can I watch you make the next one?" If they say yes, the pain is real. If it's "sure, send me a link when it's ready", it isn't, however excited they sounded.

How I'm using it

This is the exact method behind my product plan: 15+ conversations before a single line of code, and the same pain has to show up in at least 8 of them, with 3+ people already spending time or money on it. If it doesn't, I change the problem or the customer, not the bar.

I'll share the real numbers (DMs sent, replies, calls, what people said) on the timeline.

Where it falls short

  • Thin on synthesis. It teaches the conversation, not how to turn 20 sets of notes into a decision.
  • Leans B2B. For consumer products, where people can't explain why they like something, watching them often beats asking.
  • Assumes you can reach people. Getting in front of the right customers is often the hardest part, and the book doesn't spend long on it.

Read next: Continuous Discovery Habits by Teresa Torres (making interviews a weekly habit) and Interviewing Users by Steve Portigal (the craft of the conversation itself).

The one-line version

Don't ask people what they think of your idea. Ask what they did last time they had the problem, and what it cost them.