The Mom Test: A Complete Guide to Customer Interviews

The Mom Test is Rob Fitzpatrick's set of rules for customer interviews that stop even biased people from misleading you. Instead of pitching your idea, you follow three rules: talk about their life, not your idea; ask about specifics in the past, not opinions about the future; and talk less while you listen more. Good questions become facts.

Quick Answer: The Mom Test works because the best questions are ones people can't lie about. Ask them about their real life and their past behavior — not whether they like your idea — and their answers turn into evidence you can trust instead of polite encouragement.

Here is the trap almost every first-time founder falls into. You have an idea you love, you ask a few people what they think, and they tell you it sounds great. You take that as proof and start building. Months later, nobody buys — and the people who once said "I'd totally use this" have quietly moved on.

Rob Fitzpatrick wrote The Mom Test to explain why that happens and how to prevent it. The title comes from a simple insight: if you ask your own mother whether your business is a good idea, she will lie to you — not out of malice, but because she loves you. The book's argument is that everyone does a softer version of this, so the fix is to ask questions so well-framed that even your mom couldn't give you a false positive.

This guide walks through the three rules, the three kinds of "bad data" that feel like progress but aren't, and the signals that separate a real prospect from a polite one. It sits inside our complete guide to customer research for founders, and it is the right place to start before you run a single interview.

Why Customer Conversations Lie by Default

Customer conversations mislead you by default because the people in them are trying to be nice, and because you are unconsciously fishing for approval. Both sides collude to make you feel good. The result is a conversation that feels productive and teaches you almost nothing true.

Two forces bend the truth. The other person wants to protect your feelings, so they round every lukewarm reaction up to a compliment. And you want to hear that your idea is good, so you ask questions shaped to get that answer — then treat the answer as data.

The moment you mention your idea, the conversation stops being about their world and starts being about your ego. They now know what you want to hear. Fitzpatrick's point is that the problem is not lying customers; it is bad questions. You cannot control whether someone flatters you, but you can control whether your questions leave room for flattery at all.

Consider the single worst question you can ask: "Do you think this is a good idea?" It is worthless for two reasons. First, opinions about the future are guesses. Second, the only people qualified to judge whether your business works are customers spending real money — and they can only prove that by their behavior, not their predictions.

There is also a limit worth naming early. These conversations reduce market risk — whether anyone actually wants what you plan to build — not product risk, whether you can build it. A hundred perfect interviews will not tell you if your code compiles. They will tell you whether it is worth writing. Keeping that distinction straight is a recurring theme across our complete guide to startup idea validation.

The Three Rules of the Mom Test

The three rules of the Mom Test are: talk about their life instead of your idea, ask about specifics in the past instead of opinions about the future, and talk less so you listen more. Together they strip the flattery out of a conversation and leave you with facts. Each rule fixes a specific way interviews go wrong.

The table below turns each rule into a before-and-after, using one running example: a founder testing an app that plans a week of family dinners. Notice how the passing questions never mention the app at all.

The takeaway: the failing questions all invite a comforting guess, while the passing ones pull out a concrete story you can act on. The next three sections unpack why each rule matters.

Rule 1 — Talk About Their Life, Not Your Idea

Keep the spotlight on the customer's world, because the second you introduce your idea, honest data becomes almost impossible to get. Your idea is a magnet for politeness. Their life is a source of facts.

In practice this means you can often run a great learning conversation without ever describing what you plan to build. You are there to understand how they currently handle a problem — the workarounds, the frustrations, the money and time they already spend. That behavior is real. Your pitch is not yet part of their life, so their reaction to it is pure speculation.

This also lowers the stakes enormously for a nervous, first-time interviewer. You are not selling anything or defending a concept. You are asking someone to tell you about their own experience, which most people are happy to do.

It works best when you keep it casual. Fitzpatrick is clear that early learning conversations should not feel like formal, scheduled interviews with an agenda and a recorder on the table. A relaxed chat over coffee gets far more honest answers than a meeting that signals "I am evaluating you." The more it feels like an interrogation, the more the other person performs instead of remembering.

Rule 2 — Ask About the Past, Not Opinions About the Future

Ask what people have actually done, not what they think they will do, because past behavior is a fact and future intention is a wish. "Would you use this?" and "Would you pay for that?" feel like the core validation questions, but they are the two least reliable questions you can ask.

People are wildly optimistic about their future selves. They will tell you they would exercise more, read more, and yes, happily pay for your product — and then not. Anchor every claim to something that already happened. Instead of "would you pay," ask what they paid last time. Instead of "would you use it," ask how they solved this yesterday.

This is where a good question bank earns its keep. Our list of customer interview questions that pass the Mom Test gives you past-tense phrasings you can adapt on the spot when a conversation drifts toward hypotheticals.

Rule 3 — Talk Less and Listen More

Say less so the customer says more, because every minute you spend talking is a minute you are not learning. First-time founders tend to fill silences, explain their vision, and lead the witness. Discipline here is mostly about biting your tongue.

A useful habit: ask a short question, then let the pause sit. People rush to fill silence, and what they add unprompted is often the most honest part of the conversation. Your job is to steer gently and follow the threads that carry emotion, not to perform a monologue about your roadmap.

Listening well is easier when you have prepared. Before each conversation, Fitzpatrick suggests writing down your three biggest, scariest questions — the assumptions that would sink the whole idea if they turned out false. That short list keeps a casual chat pointed at what actually matters, so you are not left afterward realizing you learned a lot of pleasant trivia and nothing that moves your decision.

Bad Data — Compliments, Fluff, and Ideas

Bad data is any interview response that feels like progress but tells you nothing you can build on, and it comes in three flavors: compliments, fluff, and ideas. Each one is dangerous precisely because it is pleasant. A meeting full of bad data leaves you energized and no smarter.

The table below breaks down the three types, what each sounds like in the wild, and how to steer back to something real.

The takeaway: none of these responses are lies, and none of them are useful on their own. The skill is catching them mid-conversation and converting each into a fact.

Compliments are the sweetest poison. They light up the reward center in your brain and teach you nothing. When someone says your idea is great, do not celebrate — deflect. Steer straight back to their life: "I appreciate that. Has this actually been a problem for you recently?" A compliment that survives that follow-up is worth something; most do not.

Fluff is anything generic, hypothetical, or set in the future. It sounds specific but evaporates when you press. The cure is to anchor it: when someone says "I always meal-plan on Sundays," ask them to walk you through last Sunday. Vague habits collapse into real ones — or reveal that the habit does not actually exist.

Ideas are trickier because they feel collaborative. Customers will pitch you features endlessly, and it is tempting to write them all down as a to-do list. Resist. An idea is a clue about a problem, not an instruction. When someone says "you should add X," your job is to dig underneath it: why do they want X, what would it let them do, and what are they doing about that today?

Commitment and Advancement — The Signals That Count

Commitment and advancement are the two signals that prove a conversation was real, because they cost the customer something. Advancement means the prospect agrees to a concrete next step in your real process. Commitment means they give up something they value to get there. A meeting that ends with a compliment but no next step has failed, no matter how warm it felt.

The currency they spend is what makes it credible. Fitzpatrick names three currencies of genuine interest: time, reputation risk, and money. Anyone can hand out praise for free; only a real prospect will part with these.

The table below contrasts the hollow version of each currency with the real thing, so you can score your next conversation honestly.

The takeaway: the left column is the sound of a conversation going nowhere, and the right column is the sound of demand. If you cannot get anyone to spend a single currency, you have your answer — and it is cheaper to hear it now.

Beware the zombie lead: the prospect who never says no but never advances either. They take every meeting, love every demo, and commit to nothing. Zombie leads are the most expensive kind of false hope because they keep your roadmap alive on borrowed optimism. The fix is to always end a conversation by asking for a real commitment; the ask itself flushes out who is serious. For a deeper breakdown of reading these signals, see our guide to commitment and advancement as the interview signals that count.

This is also where a validation platform like Edmired earns its keep. Interviews only compound if you capture what was actually said — the specific stories, the commitments offered, the currencies spent — instead of the rosy summary your memory writes on the drive home. A structured place to log those facts keeps the search honest and your evidence, not your optimism, in charge of the decision.

Common Mistakes Founders Make Running Mom Test Interviews

The most common Mom Test mistakes all share one root: the founder slips back into seeking approval instead of hunting for the truth. Knowing the rules is easy; holding to them under the emotional pull of a real conversation is the hard part. These are the traps that catch people most often.

• Pitching instead of learning. The urge to explain your beautiful idea is strong, and it poisons the well. Save the pitch; the early conversation is for their life, not your product.
• Asking for opinions and predictions. "Would you use it?" and "Do you like this?" hand the customer a script for flattery. Ask what they did, not what they think they'd do.
• Treating compliments as validation. Feeling good after a meeting is not the same as learning something. If you can't name a fact you gathered, the meeting was bad data in a nice wrapper.
• Writing down every feature request. Ideas are raw material, not a roadmap. Dig for the problem behind each one before you let it near your build list.
• Making it a formal meeting too soon. Early discovery works best as a casual chat, not a scheduled interview with a slide deck. Formality makes people perform; casualness makes them honest.
• Talking to anyone who will listen. Vague segments give vague answers. Narrow to a specific, findable group before you start so you can compare like with like.
• Never asking for a commitment. A conversation with no next step is a dead end you mistake for progress. Always close by asking for time, reputation, or money.

Avoid these seven and your interviews will produce something rare: information you can actually make a decision with. If your friends and family have been telling you the idea is great, this is the discipline that tells you whether that praise means anything.

Key Takeaways

• The Mom Test is a set of rules for asking questions people can't lie about — you fix your own questions rather than hoping customers become more honest.
• Follow the three rules every time: talk about their life not your idea, ask about specifics in the past not opinions about the future, and talk less so you listen more.
• Compliments, fluff, and ideas are bad data — they feel like progress but carry no evidence, and the skill is catching and redirecting each one mid-conversation.
• Anchor everything to real past behavior — what someone did last week is a fact, while what they say they would do next month is a wish.
• Commitment and advancement are the signals that count — a real prospect spends time, reputation, or money, and a meeting with no next step has failed however warm it felt.
• Watch for zombie leads — prospects who never say no and never commit are the most expensive false hope, so always end by asking for a concrete commitment.
• These conversations reduce market risk, not product risk — they tell you whether the thing is worth building, never whether you can build it.

Frequently Asked Questions

What is the Mom Test in simple terms?

The Mom Test is a way of asking customers about their lives so honestly that even your own mother couldn't give you a false positive. You never ask whether your idea is good. You ask about what they actually do, what it costs them today, and what they've already tried — questions whose answers are facts, not flattery.

Why is it called the Mom Test?

It is named after the idea that if you ask your mother whether your business is a good idea, she'll say yes because she loves you. Rob Fitzpatrick's point is that most people soften the truth the same way. The rules are designed so the questions themselves prevent that bias, making even a biased person's answers reliable.

What are the three types of bad data in customer interviews?

The three types are compliments, fluff, and ideas. Compliments are praise that costs nothing and proves nothing. Fluff is anything generic, hypothetical, or future-tense. Ideas are feature requests that hide the real underlying problem. All three feel like progress, so the skill is spotting each one and steering back to concrete facts.

How do I know if a customer interview actually went well?

A good interview ends with a commitment or an advancement, not a compliment. If the customer spent something they value — time in a booked follow-up, reputation via an introduction, or money as a deposit or pre-order — the interest is real. If they only said nice things and agreed to nothing, you gathered bad data.

Can I use the Mom Test for a consumer app, not just B2B?

Yes. The rules work anywhere people can flatter you, which is everywhere. For a consumer idea, "commitment" simply looks different — a signup that converts, a small pre-order, or someone bringing friends into a test. You still ask about past behavior and still insist on a real next step instead of a warm "I'd totally use it."