2.1b · Mom Test Synthesis: Build, Pivot, or Kill

Module 2 · After Step 1 · From Idea to First Paying Customer

Input: 10 scored Mom Test transcripts (from Ch 2.1) + completed interviews (from Ch 2.3 (a + b))

Output: a build / pivot / kill decision + a one-page validated problem statement

TL;DR: Score 10 transcripts, count strong signals, make one of three calls. 90 minutes. The decision you avoid here costs you a quarter of build time later.

You should be here AFTER your Ch 2.3 (a + b) interviews are done. If you don’t have 10 scored transcripts in hand, return to Ch 2.1 for the technique, then Ch 2.3 (a + b) for recruitment. This page is the synthesis pass - you cannot complete it without real interview data.

After all 10 interviews are done, you have scored transcripts in a folder and a number. Synthesis is the 90-minute step that turns those transcripts into the one-page validated problem statement you’ll carry into Module 3. Skip this step and go straight to Lovable, and you have not validated anything - you have a folder and a hypothesis.

The 3-step synthesis

Synthesis runs on three moves. You don’t need a framework. You need 90 minutes alone with the 10 transcripts, a printed template, and the willingness to write down a number that might be a 3.

Step 1 - Score each interview 1-10. Open the transcripts in order. For each call, read your handwritten Q4 score and your emotional-flag count from the Ch 2.1 script. Combine the two into one number from 1 to 10. A score of 7+ means the interviewee gave you a 7 or higher on Q4 with a comparison (the polite-default 7 with no comparison rounds to 5) and at least 3 emotional-language flags across the five answers. A 4 to 6 means partial signal - a real story but a weak workaround, or a high Q4 score with zero frustration language. Below 4 means polite-yes mode: vague Q1 answers, “nothing yet” on Q3, a hedged Q4 number under 7.

Write the number on the first page of each transcript within 5 minutes of hanging up. The score you write immediately is more honest than the one you’d write after a week of wanting the number to be higher.

Step 2 - Count the strong signals. On a single sheet of paper, list the 10 scores in a column. Circle every score that is 7 or higher. That circled count is your strong-signal number.

The pattern matters more than the average. Eight 7+ scores and two 3s is a strong signal - you found a problem two ICPs share. Five 7+ scores and five 5s is muddled - the ICP definition is too broad. Three 9s and seven 4s is the dangerous one: you talked to your three best friends in the industry and they validated the idea while seven strangers told you the truth.

Step 3 - Write the one page. Open the Validated Problem Statement Template on a second screen. Fill it in within 30 minutes. Five sections, no exceptions: who has the problem (named persona, named industry, strong-signal count); what it costs them (time, money, and one specific quote from a real transcript - avoid “frustrating” and “time-consuming”); what they’ve tried (named workarounds and why each failed - these are your real competitors); why now (the trigger event or market shift that makes this solvable in 2026); how big is the pain (average score plus strong-signal count - print both, not just the average). A single side of paper. If you spill onto a second page, the persona is too broad or the pain is too vague.

The decision: build / pivot / kill

Your strong-signal count from Step 2 routes you to one of three outcomes.

7+ strong signals: build. You have a problem that 70%+ of a stranger sample confirmed with felt urgency. The validated problem statement is your input to The One-Page Product Brief.

Before you start writing code, run the 3 pre-orders test: ask 3 of your strongest-signal interviewees for a pre-order, a paid letter of intent, or a deposit. Strangers who told you their problem score is a 9 should be willing to put a small commitment behind it. If 3 of your top 5 say yes, you have validation with money attached - the strongest signal there is. If 0 of 5 say yes, the 7+ scores were politer than you thought.

4-6 strong signals: pivot. The signal is partial. Most often this is an ICP problem, not a problem problem. Pick the cleanest segment, sharpen the ICP definition, run 5 more interviews against that narrower group. Don’t build yet. The 5 sharper interviews cost you a week. A built MVP against a fuzzy ICP costs you a quarter.

Below 4 strong signals: kill. Strangers were polite. The market said no in the only way the market knows how to say no before a launch: by not feeling the pain enough to put a number on it. Write down what you learned about the wrong ICP, the wrong framing, or the wrong trigger event. Start Find 10 People With the Problem again with a different hypothesis.

What good looks like vs what bad looks like

Bad problem statement (vague, unfilled):

Founders need a better way to validate their startup ideas. Many of them waste time and money.

Good problem statement (specific, named, signed):

Pre-seed B2B SaaS founders running their own discovery do customer interviews, but 9 of 10 (per our 10-call sample, Apr-May 2026) use hypothetical-future questions and get polite-yes answers. The average interviewee currently spends 6-12 hours running interviews and learns the problem wasn’t real only after their first launch flops - typical sunk cost is 6 weeks of build time plus $15K-$30K of contractor spend. Workarounds tried: YC Library essays (too high-level), $1,500 SurveyMonkey panel (taught one founder I spoke with nothing in the survey style), free templates downloaded but not used. Why now: AI-built MVPs accelerated this failure mode - the prototype lands in 4 days instead of 12 weeks, so the validation gap surfaces faster. Pain average 7.6/10 across 10 calls, 8 strong signals.

The good answer has named industry, dated sample, named workarounds with named failure modes, a quantified cost, a why-now, and a strong-signal count. A peer can argue with it. If your statement has the word “many” or “a lot,” cross it out.

The Validated Problem Statement Template is the artifact for this section. Print it, fill it in 30 minutes, get 2 signatures, and the problem validation checkpoint is closed.

Writing the one-page statement is the validation step. Ten transcripts in a folder don’t count - until you’ve scored them, counted the strong signals, and written down what the pattern says, you have raw material and a hypothesis, not a validated problem.

Done when: You have a build / pivot / kill decision backed by your strong-signal count, and a one-page validated problem statement. Next click: If build - 2.4 · Build a Clickable Prototype to test the shape with 5 of your strongest-signal interviewees, then 3.1 · The One-Page Product Brief. If pivot - return to 2.3a · Find 10 People: Where to Look to rebuild your list around a sharper hypothesis (same hypothesis, different list). If kill - the hypothesis is wrong, not the list; return to 1.1 · Form Your Founding Hypothesis and rewrite the weakest blank using verbatim quotes from your dead transcripts, then re-run 2.3a. If blocked: If the numbers aren’t adding up, re-read Ch 2.1’s scoring rubric - the Q4 score and flag-count combination is what separates a 7 from a 5.

Case Study: Tomas & Mia

Tomas (ReconcileBot, B2B SaaS): scored 7 of 10 controllers at 8/10 or higher. Strong-signal count = 7. Verdict: BUILD. Writes the validated problem statement using a verbatim quote (“CFO billed $800 last week on a 4-hour reconciliation”) and moves to 2.4 prototype.

Mia (TutorMatch, B2C marketplace): scored 8 of 10 parents at 9/10 or higher. Strong-signal count = 8. Verdict: BUILD. Validated problem statement anchors on a verbatim quote (“missed a $2,000 project deadline on hold with a tutoring center”) and moves to 2.4 prototype.

Built by JetThoughts as part of the From Idea to First Paying Customer curriculum.