The Mom Test is a set of rules for customer interviews, published in 2013 by the entrepreneur Rob Fitzpatrick. It holds that whether an interview produces useful information depends on the questions asked rather than on the interviewee, and that good questions ask about the person's past behaviour rather than their opinion of your idea. It is widely used in startup customer discovery.
Before you write a line of code, and every week after launch until you have 50+ paying customers. Anytime you need to find out if a problem is real, and what people actually do about it.
What the Mom Test is
The Mom Test is a set of rules for customer interviews. It was published in 2013 by the entrepreneur Rob Fitzpatrick, in a short book of the same name.
Its central claim is that a bad interview is the interviewer’s fault. If you ask your mother whether your business idea is good, she will say yes, because she loves you. The name is the test: a good question is one your mother could not lie about even if she wanted to, because it asks about her life rather than your idea.
Three rules follow from that. Talk about their life, not your idea. Ask about specific things that already happened, not what they would do in future. Talk less and listen more.
The method is widely used in startup customer discovery, and its vocabulary, particularly the phrase “the Mom Test” itself, has entered general use in product and founder circles.
- Follow three rules The whole method
Their life not your idea. Past specifics not future hypotheticals. Listen more than you talk.
- Recognise three kinds of bad data Mid-conversation
Compliments, fluff and ideas. Each feels like progress and each predicts nothing.
- Ask a better question In the moment
Every bad question has a version that asks about behaviour that already happened.
- Get a commitment Or it was just a nice chat
Time, reputation or money. A meeting that ends with none of the three produced nothing.
Why it matters
The failure mode this method addresses is not that founders skip customer interviews. Most do them. The failure is that the interviews return encouragement, the founder reads encouragement as evidence, and eighteen months later nobody buys.
That failure is invisible while it is happening. A meeting full of compliments feels better than a meeting full of awkward specifics, so the useless conversation is also the more enjoyable one. Founders come away energised and no better informed.
There is a specific sound to it, once you have heard it a few times. Lots of “I would definitely use that”, no dates, no numbers, no names of people who have the problem now. Warmth is not evidence. It is just warmth.
What does the output of doing this properly look like? Across 280 ideas run through ShipFit’s problem-discovery stage, founders surfaced 2,888 distinct problems, which is about ten each. Note the intensity distribution, and in particular note that the average problem scores 70 out of 100. Either founders are very good at finding painful problems, or everybody involved is grading generously. We would not bet against the second.
Across 2,888 problems surfaced from 280 ideas, founders named about ten problems each, and rated the average one 70 out of 100 for intensity.
- Rated "must solve"
- 1,096 problems 37.9%
- Occurring weekly
- 41.4%
- Occurring daily
- 22.1%
- Mean intensity
- median 75 70.4 / 100
Sample: n = 280 ideas, 2,888 problems, 10.3 per idea
What it does not say: Intensity is scored by the engine from the founder’s description, so it inherits both their optimism and its own.
It costs nothing to say and predicts nothing. It is the most common outcome of a meeting that felt like a success.
Deflect it and get back to their life. Do not thank them and move on; treat a compliment as a sign the conversation has drifted into a pitch.
Generic claims, hypotheticals and future-tense promises. All three describe an idealized person rather than the one in front of you.
Anchor it to the last concrete instance. "When did that last happen? Walk me through it. What did you do?"
Feature requests are a symptom, not a specification. The request is real; the proposed solution usually is not.
Dig for the root. "Why do you want that? What would it let you do? How are you coping without it today?"
When to run it
- You have an idea and no evidence anyone but you wants it.
- You are about to commit engineering time to a problem you have not confirmed.
- Your existing interviews keep coming back positive and nothing converts.
- You are entering a new segment and your assumptions about the old one may not transfer.
- You need to know what the problem currently costs someone, in money or hours.
- You know the problem is real and need to decide what to charge. Use Van Westendorp →
- You need to understand what the buyer is hiring the product to do. Use Jobs to be Done →
- You are deciding which features make the first release. Use MoSCoW →
- You need to know what will stop a competitor copying you. Use 7 Powers →
The three rules
- Talk about their life, not your idea
The moment your idea is on the table the conversation stops being research and becomes a pitch. People are polite, so a pitch returns approval, and approval is not information.
Do this instead Ask how they handle the problem today, and never name what you are building.
- Ask about specifics in the past, not generics or the future
Predictions about future behaviour are close to worthless, including your own. People are not lying; they genuinely do not know what they will do.
Do this instead Ask what happened the last time, what it cost them, and what they did about it.
- Talk less and listen more
A founder who talks half the time has spent half the meeting learning nothing, and has usually steered the answers by signalling what they hoped to hear.
Do this instead Aim to speak for well under a third of the conversation.
Good questions and bad ones
Most bad questions are bad in the same way: they ask for a prediction, an opinion, or a yes. All three are free to give.
| Question that fails | Why | Ask this instead |
|---|---|---|
| “Do you think this is a good idea?” | Asks for an opinion, and opinions are free. | “How do you handle this today?” |
| “Would you buy a product that did X?” | A hypothetical about the future. Everyone says yes. | “What are you using now, and what does it cost you?” |
| “How much would you pay for X?” | Invites a made-up number with nothing behind it. | “What is in the budget for this already, and who signs it off?” |
| “Do you ever have a problem with Y?” | Leading, and answerable with a polite yes. | “Talk me through the last time Y happened.” |
| “What features would you want?” | Outsources product design to someone who has not thought about it. | “What have you already tried, and why did you stop?” |
The pattern in the right-hand column is that every good question asks about something that has already happened. Past behaviour is a fact the person cannot be polite about. Future behaviour is a guess they are making on the spot, to be nice.
Commitment: what a good meeting actually produces
This is the part most summaries of the method leave out, and it is the answer to the question every founder has after a promising conversation.
- TimeCounts
A follow-up meeting with a fixed date, a working session, a trial with their real data.
Does not countA vague "send me something" with no date attached.
- ReputationCounts
An introduction to their boss or their peers, a public reference, bringing a colleague to the next call.
Does not countA LinkedIn connection, or "I will keep an eye on it".
- MoneyCounts
A deposit, a letter of intent, a pre-order, a signed pilot.
Does not countEnthusiasm about the price you floated.
A conversation that ends with none of the three went well socially and told you nothing. That is the test, and it is a harder one to pass than it sounds: most founders leave their best meetings with a compliment and a promise to stay in touch.
The Mom Test in practice: Quibi
The most expensive demonstration available of what happens when the research says yes and the behaviour was never asked.
Quibi · August 2018 to December 2020
$1.75bn, two of the most experienced executives in media, and six months of operation.
Quibi launched in April 2020 with $1.75bn in funding, Jeffrey Katzenberg as founder and Meg Whitman as chief executive. The thesis was short-form premium video for the commute: ten-minute episodes, phone-first, filmed in both orientations at real production budgets.
It announced it was shutting down in October 2020 and wound up in December. Katzenberg publicly attributed the failure partly to the pandemic, which removed the commute the product was designed around.
That explanation is not wrong, and it is also not sufficient. The product had no sharing, so a moment could not travel; the audience assumption had been tested in survey rather than in behaviour; and the commute was a context, not a job. Nobody was walking around with an unmet need for premium content in ten-minute chunks that they could not send to anybody.
- Raised before launch
- $1.75bn
- Launch to shutdown announcement
- ~6 months
- Reported paying subscribers at wind-down
- under 500,000
What it shows: Seniority and capital do not substitute for evidence. Quibi could afford to build anything, which meant it never had to find out whether anybody wanted this before it existed.
The Mom Test vs other ways of asking
Is this problem real, and what does it already cost them?
Gives you: Facts about past behaviour, plus a commitment or a clear no
How common is a pattern I already understand?
Gives you: Counts. Useless before you know what to ask about
Can someone operate the thing I built?
Gives you: Friction points. Assumes the thing should exist
What does a room full of people say when watching each other?
Gives you: Consensus, mostly manufactured by the loudest participant
When it won’t help you
- It tells you the problem is real, not that your solution sells
The rules are built to keep your idea out of the conversation, which is exactly why the conversation cannot validate it. A confirmed painful problem is compatible with your product being the wrong answer to it.
Instead: Follow it with something that puts a real offer in front of someone: a pre-sale, a deposit, a pilot.
- It is slow, and it does not scale
Ten to twenty good conversations take weeks and cannot be delegated to a form. There is no version of this that runs overnight across a thousand people.
Instead: Use it to work out what to ask, then use a survey to find out how common the answer is.
- It assumes you can reach the right people
Every rule is about question quality, and none of them helps if the people answering are not buyers. Screening is left to you, and screening is often the harder problem.
Instead: Define the buyer first. A perfect interview with the wrong person returns clean, precise, useless data.
- Past behaviour is a weak guide in genuinely new categories
The method leans hard on what someone already does. When no adequate substitute exists, there may be very little past behaviour to ask about.
Instead: Look for the workaround rather than the product. People rarely have no solution; they have a bad one made of spreadsheets and time.
ShipFit and the Mom Test
ShipFit applies Mom Test discipline at Stage 3 (What Hurts?). It ranks real pain points by frequency and intensity, then generates the questions you should be asking your buyers, phrased in past-tense behaviour rather than future-tense intention. The output is a ranked problem list, the buyer profile to recruit against, and a synthesis sheet to log what you actually hear, so the answers feed forward into pricing, MVP scope and launch.
Where this sits in the sequence
The Mom Test comes first. Everything downstream, the buyer, the price, the scope, the positioning, assumes somebody has confirmed the problem is real and worth money.
- The Mom Test
Confirm the problem is real and find out what it already costs.
You are here
- Jobs to be Done
Work out what the buyer is hiring a solution to do.
- Buyer Persona
Turn the interviews into something the whole team can act on.
- Van Westendorp
Find out what that buyer will actually pay.
Further reading
- Rob Fitzpatrick, The Mom Test (2013). The source. Around 130 pages and readable in an evening.
- Rob Fitzpatrick, The Workshop Survival Guide (2019). Useful if your research is turning into group sessions, where the rules change.
- Jobs to be Done. Where the conversations go next, once you know the problem is real.
- Van Westendorp. What to run once you need a number rather than a narrative.
- Buyer Persona Canvas. How to turn a stack of interview notes into something the whole team can use.
- Lean Startup validation. The loop these conversations feed.
- Churn rate. The number that eventually tells you whether the problem you confirmed was worth solving.
How to apply The Mom Test
- 1
Pick who you are talking to before you pick what to ask
Screen for people who would plausibly buy: the right role, the right company size, and evidence they have the problem now. A flawless interview with the wrong person returns clean, precise, useless data, and no rule in this method protects you from that.
- 2
Never mention your idea
The moment the idea is on the table the conversation becomes a pitch, and a pitch returns approval rather than information. If they ask what you are building, deflect and return to their life. This is rule one and it is the one founders break without noticing, because being upfront feels like honesty.
- 3
Ask about specific things that already happened
Replace every question about the future with one about the past. Not 'would you use this' but 'what did you do the last time this happened'. Predictions about future behaviour are close to worthless, including your own, and people are not lying when they get them wrong.
- 4
Spot bad data in the moment and deflect it
Compliments, fluff and ideas. A compliment means the conversation drifted into a pitch, so deflect and return to their life. Fluff means generic or hypothetical, so anchor it to the last real instance. An idea means a feature request, so dig for the problem underneath it. Recognising these live is most of the skill.
- 5
Talk less than a third of the time
A founder who talks half the meeting has learned nothing for half the meeting, and has usually steered the answers by signalling what they hoped to hear. Silence is a tool; leave it there and let them fill it.
- 6
Push for a commitment, or record that there wasn't one
A good conversation ends with the other person giving up time, reputation or money: a scheduled working session, an introduction to their boss, a deposit. If none of the three was offered, the meeting was pleasant and produced nothing. Write that down honestly rather than logging the enthusiasm.
Common mistakes
- **Pitching in the first 2 minutes.** The moment you describe your idea, interviews become useless for validation. Only pitch at the end, if you need to.
- **Asking 'would you pay for this?'.** Leading question, useless answer. Ask instead: 'how much did you spend trying to solve this last time?'
- **Interviewing friends.** They lie out of love. Interview strangers, ex-colleagues, or people one LinkedIn-degree away.
- **Not writing down the exact words they use.** Your copy lives in their vocabulary. Their words beat your words every time. Record (with permission) or write real-time.
- **Counting interest as validation.** Interest is free. Only commitments (money, time, reputation) count.
- **Doing 3 interviews and declaring victory.** The signal emerges around interview 15. Ten interviews is the floor, not the ceiling.
How ShipFit operationalizes this
ShipFit applies the Mom Test at Stage 3 (What Hurts?) of the 9-question flow. The stage uses Mom Test discipline to rank real pain points by frequency and intensity, then surfaces the questions you should be asking your buyers, phrased in past-tense behavior, not future-tense intentions. The conversations themselves are yours to have, and ShipFit equips them: the question list, the buyer profile to recruit against, and a synthesis sheet to log what you actually hear. Start a Quick Take to see the framework applied to your idea.
ShipFit runs 55 frameworks across 9 decision stages
The Mom Test is one tool in a bigger toolkit. The full library covers market sizing, buyer discovery, MVP scoping, pricing, and launch.
The Mom Test
Q3Rob Fitzpatrick
Validation question methodology, real interviews, not theater
Jobs-to-be-Done
Q2-Q4Clayton Christensen
Functional, social, and emotional jobs your product fulfills
7 Powers
Q4Hamilton Helmer
Strategic moats: Scale, Network, Counter-positioning, Switching, Brand, Cornered Resource, Process
Van Westendorp PSM
Q6Feature-weighted price sensitivity analysis without guessing
Blue Ocean Strategy
Q4Kim & Mauborgne
ERRC framework: Eliminate, Reduce, Raise, Create
Fake Door Testing
Q7Pre-build behavioral validation with landing pages and apology modals
+ 49 more: TAM/SAM/SOM Analysis, Porter's Five Forces, Market Timing Analysis, Unit Economics (LTV/CAC)...
Frequently asked questions
How many Mom Test interviews do I need before I know something?
Can I Mom Test over video calls?
Do I need to record?
What if they ask what I'm building?
What's a real commitment vs. a fake one?
Is the Mom Test still relevant for B2B SaaS in 2026?
How is the Mom Test different from Jobs-to-be-Done?
What are the three rules of the Mom Test?
What are the three types of bad data?
What counts as a commitment in the Mom Test?
How many customer interviews do I need?
Does the Mom Test work for B2B?
Keep exploring
The 9-step playbook from market verdict to ship-ready spec.
MoSCoW sorts features into Must, Should, Could and Won't against a fixed date. The 60% capacity rule is the part most teams drop, and dropping it breaks the method.
Adele Revella's Five Rings of Buying Insight, the questions that produce each one, and why a persona built without buyer interviews informs no decision.
Default-prompted AI is a slop machine: agreeable, plausible-sounding, useless for validating an idea. Here's how to use AI for the parts where it actually adds signal, and where to keep it out of the way.
Most founder market research is a TAM slide that nobody believes. The numbers that actually matter are smaller, harder to defend, and tell you whether the market exists for the ten-customer version of your business.
Van Westendorp in 4 numbers. Skip the survey-platform fees.
The state where a product satisfies a strong market demand from a specific buyer segment such that customer pull on the product exceeds the founder's effort to push it. Coined by Marc Andreessen in 2007 ('the only thing that matters'). Operational measure: 40%+ of active users would be 'very disappointed' if they could no longer use it (Sean Ellis test, popularized by Rahul Vohra at Superhuman in 2018). Below 40%, you don't have PMF yet, regardless of revenue or press.
Idea validation for technical founders who can build anything. ShipFit forces the buyer and pricing decisions your engineering skill lets you skip. Start free.
Lovable builds apps by chatting with AI: describe it, see it built in real-time, deploy. ShipFit makes 9 decisions before you open Lovable so you build the right thing. ShipFit even exports a Lovable-optimised prompt that encodes every decision. Use both. ShipFit first.
Ready to make your next product a success?
9 decisions between your idea and a product worth building.