The Mom Test
Market & numbers — every figure sourced
The Mom Test
Rob Fitzpatrick's The Mom Test (2013) answers one brutal question: how do you learn whether your business is a good idea when everyone around you is lying to you? Not maliciously. Your mom lies because she loves you. Strangers lie because being nice is easier than being honest. The book is named for the insight that a well-constructed question is one that even your own mother could not skew with false encouragement.
The whole discipline is roughly a 2-hour read, and it has earned its way into curricula at Harvard, UCL, and accelerators like Seedcamp. Here are the principles in my own words, and why each one earns its keep in a real business.
The core failure it fixes
Founders go looking for validation and call it research. They pitch their idea, watch people nod, hear "I'd totally buy that," and walk out convinced. Then they build, launch, and nobody pays. The interviews were not wrong because the founder was dumb. They were wrong because the questions invited politeness instead of truth.
The three rules
Fitzpatrick boils good conversations down to three habits, and almost everything else follows from them.
- Talk about their life, not your idea. The moment you describe your product, the other person switches from witness to critic-trying-to-be-nice. Keep your idea in your pocket. Ask how they live and work today.
- Ask about specifics in the past, not generics or opinions about the future. "Would you use this?" is worthless. "Walk me through the last time you hit this problem" is gold. People are wildly optimistic predictors of their own future behavior and reliable reporters of their actual past.
- Talk less, listen more. You learn nothing while your mouth is moving. The job is to get them talking about their problems, then shut up.
Why it matters: these rules convert a sales pitch (which gathers compliments) into an investigation (which gathers facts). Compliments feel great and predict nothing. Facts are the only input that should move your roadmap.
The three kinds of bad data to distrust
- Compliments. "That's such a cool idea." Fitzpatrick calls these shiny, distracting, and worthless because they cost the speaker nothing. When you catch yourself feeling good in an interview, get suspicious.
- Fluff. Generic claims ("I always..."), future promises ("I would definitely..."), and hypotheticals ("I might..."). Anchor every fluffy answer back to a concrete past instance: "When did that last happen? What did you do?"
- Ideas (the feature wishlist). Customers will hand you a pile of features. Do not stenograph them. Dig for the motivation and the problem underneath each request, because the request is rarely the right solution.
Why it matters: bad data is worse than no data. It feels like progress and points you confidently in the wrong direction, which is how founders burn a year and a payroll building something nobody needed.
Good questions vs. bad questions
| Bad (invites lies) | Good (forces truth) |
| --- | --- |
| "Do you think this is a good idea?" | "How do you solve this today?" |
| "Would you pay $X for it?" | "What does this problem cost you now (time/money)?" |
| "Would you use a feature that...?" | "Talk me through the last time this came up." |
| "Don't you think this is cool?" | "What have you already tried to fix it?" |
A simple gut-check: if the answer could flatter you, it is probably a bad question.
Commitment and advancement: the only real signals
Talk is cheap, so make people spend something. A meeting is only a success if the prospect gives up one of three currencies, or agrees to a next step that costs them:
- Time - a scheduled follow-up, sitting through your wireframes, agreeing to a trial.
- Reputation - an intro to a peer or the actual decision-maker, a public testimonial.
- Money - a letter of intent, a deposit, a pre-order.
A conversation that ends with "this is great, keep me posted" advanced nothing. A conversation that ends with a calendar invite or a $50 deposit told you something true. This is the test inside the test: did the relationship move forward in a way that cost them?
Run the conversation right
- Keep it informal. The best customer learning happens before it ever looks like a formal "interview."
- Take control early so it stays about their problem, not your demo.
- Write notes by hand and capture exact quotes; typing on a laptop reads as rude and flattens nuance.
- Bring a second person where you can - one leads, one records - and avoid a single-person feedback bottleneck.
- Aim for roughly 5 conversations per segment, and keep going until you stop hearing anything new.
- Push the raw notes and quotes to a shared doc so the whole team learns, not just the person in the room.
How to apply it in a startup tomorrow
Pick the one customer segment you are betting on. Do not mention your product. Ask three people to walk you through the last time they faced the problem you think you are solving: what they did, what it cost them, what they tried, and what they spent to fix it. If the problem is real, they will get animated and you will hear about money already being spent or hacks already in place. If you hear shrugs and polite "sounds useful," you just saved yourself months. Then ask for a commitment of time, reputation, or money. The response to that ask is your truest data point of the whole conversation.
---
Attribution: principles distilled from The Mom Test by Rob Fitzpatrick (2013), summarized here in original wording. Read the source for the full treatment.