2.1 · The Mom Test: Ask About the Past, Not the Future
Module 2 · Lesson 2.1 · [CORE] · From Idea to First Paying Customer
Input: a Founding Hypothesis sentence (from Lesson 1.1) + your
[CUSTOMER]blank from that sentence (the role, company size, and situation it names - that description is your ICP, your Ideal Customer Profile)Output: the 5-question Mom Test template + a draft question list (5-8 questions) ready to sharpen in Lesson 2.2 and then run in real interviews after Lesson 2.3-2.4 recruitment. The scoring rubric becomes your reference card once you have transcripts in hand.
Progress: M2 · 1 of 6 · Results so far: all Module 1 artifacts - Module 2 starts here
Run ten interviews where the only question is “would you pay for this?” and you’ll close the week with a stack of yeses and an empty launch. The technique below switches every question to the past tense - what did you do last time, what did it cost, show me the spreadsheet - because whoever paid in the past keeps paying, and whoever did nothing will keep doing nothing no matter what they promise you over coffee.
After this lesson you will be able to: write interview questions that ask about past behavior - so the answers tell you what people actually did, not what they would politely promise.
Rob Fitzpatrick’s book The Mom Test (2013) named the technique that prevents the polite-yes problem. The core idea: ask interviewees to recount what they actually did the last time the problem happened, not what they think they’d do about a product you describe to them.
The 5 questions
Every question a first-time founder writes first feels like it gets to the truth - and each one is built to surface a polite lie. The fix is the same each time: switch to past behavior. Read the rewrites as written; “would you” instead of “did you” flips the answer back into hypothetical polite.
| The question you’d write | Why it produces polite-yes | Fitzpatrick’s past-behavior rewrite |
|---|---|---|
“Do you find [PROBLEM] frustrating?” | Leading. The interviewee hears that you want a yes; their politeness reflex supplies one. | “Tell me about the last time [PROBLEM] happened.” |
“Would you use a tool that solved [PROBLEM]?” | Hypothetical-future. Their answer is a guess about a person who doesn’t exist yet. | “Walk me through what you did the last time you tried to handle [PROBLEM].” |
“How important is solving [PROBLEM] for you?” | Asks for a self-rating, which everyone inflates. People rate everything 7/10 to avoid sounding rude. | “What did the workaround cost you - in time, money, or sanity - the last time?” |
| “Would you pay $X for a solution?” | Pricing hypothetical. They have no skin in the game; saying yes costs them nothing. | “What have you already tried, paid, or built to deal with it?” |
| “Does this idea sound good to you?” | Compliment-fishing. The interviewee can’t refuse without being mean to you. | “On a scale of 1-10, how would you rank fixing [PROBLEM] this year against the other 3 things on your list?” |
What a passing answer sounds like, versus a polite one, on the very first question:
- Fail (polite): “Yeah I usually struggle with reporting.” No date, no mechanic - autopilot polite mode.
- Pass (real story): “Last Tuesday at 9pm I spent 40 minutes copying numbers from three spreadsheets into a slide for the board.” Date, time, tool, duration, feeling.
The first answer is always thin - excavate. Ask “tell me about the last time” and most people answer at the surface: “Oh, last night after dinner.” Help them re-enter the moment and walk it forward - “Where were you? What happened right before? What did you do next?” - until the actual sequence comes out. Then stop talking. Leave a few seconds of silence; the pause is uncomfortable enough that people fill it with the detail they were about to skip. (Teresa Torres calls this story excavation.)
The full Mom Test reference has the pass/fail signal and follow-up for all five questions, plus the interview flow and the score-and-recover routine.
Score each call with the 3 emotional flags
While the script runs, listen for three patterns. They do more work than the words “yes” and “no.”
| Flag | Example Phrases | What it signals |
|---|---|---|
| Frustration language | “I hate this.” “It drives me crazy.” “Every single week.” | The problem is felt. Polite interviewees suppress feeling - the opposite of what validation needs. |
| Workaround language | “I’ve been meaning to…” “We hacked together…” “I pay [TOOL] $X for this.” | The interviewee already spent time or money on a solution that doesn’t fully work. That budget is the line item your product would replace. |
| Urgency language | “Last week.” “This morning.” “I missed my kid’s birthday because of this.” | A problem that happened today is felt more sharply than one that happens “sometimes.” Time-anchored urgency is the strongest signal in the set. |
Scoring the flags: A passing call has 3 or more flags spread across the five answers. A failing call has 0 or 1 - the interviewee is being polite to you. Two flags is ambiguous; treat as a 5/10 default. This 1-10 score per call is what Lessons 2.4-2.6 count to make the build / pivot / kill decision, so score every call the same way. The printable Interview Scorecard runs this pass/fail check and flag tally after each call.
Write:
- Open a new Google Doc titled
Mom Test draft - [DATE]in the sameFounder OSfolder as your Founding Hypothesis.- Copy the 5 verbatim rewrite questions from the table above.
- Add 2-3 ICP-specific probes of your own (e.g., for a chiropractor ICP: “Walk me through your last insurance-claim resubmission. When did it happen?”).
- Success check: you have a 5-8 question list, every question anchored in a past event, none asking “would you.”
That list is the input Lesson 2.2 sharpens against AI personas.
If this fails: your questions still sound hypothetical.
- Why: a question asks the interviewee to predict, rate, or evaluate instead of recount a specific past event.
- Fix: rewrite it to start with “Tell me about the last time…” or “Walk me through what you did when…”. If you can’t anchor it to a past moment, cut it.
Read your draft list aloud. Which question would let a polite stranger agree without ever having lived the problem? Rewrite that one first - it’s the one most likely to hand you a false yes.
Done: you understand the 5 Mom Test questions, can spot hypothetical phrasing, and have a draft question list (5-8 questions) anchored in past behavior.
You have now: all Module 1 artifacts + a draft Mom Test question list (2.1). Sharpening and recruiting come next - the problem itself isn’t validated until you run the 10 real interviews.
Next: 2.2 · Sharpen Your Question List with AI Personas - takes the draft list you just wrote and stress-tests each question against a simulated ICP before you spend a real interview slot on it.
If blocked: open the Mom Test Interview Script - it has the 5 questions verbatim with follow-ups.
Deeper reference: Per-question pass/fail signals, the interview flow, and the score-and-recover routine
See it in action: Module 2 walkthrough: Mia interviews ten parents
Built by JetThoughts as part of the From Idea to First Paying Customer curriculum.