Founder-market fit gets discussed two ways: as credentials — build in a domain you already know — or as encouragement. Neither is answerable. The version you can actually decide is narrower and less comfortable: can you build this, explain it, sell it and support it, and if the answer is no, what happens to the idea?
In this product the answer to that last part is unusual. This is the only analysis where a fact about you can kill the idea outright, and the rest of the page is what that costs.
The checks, and none of them is about your CV
The framing matters more than the list. This is execution-understanding fit, not resume matching and not personality testing — the questions are about a specific buyer and a specific job, so a decade in an adjacent industry earns you nothing here on its own.
understands_buyerunderstands_workflowcan_judge_output_qualitycan_support_userscan_create_distributioncan_tolerate_support_shapelearnabilityRendered from the check list the analysis runs, in its own order. The last one is not a check on you — it is a check on the other answers.
The third one is the one people have never considered. Can you tell whether the output is good? Plenty of buildable products fail here: you can ship the thing and have no way to know if what it produced was right, which means you cannot improve it and cannot defend it when a customer says it is wrong.
The last one changes the meaning of everything above it. A gap is only a problem if it does not close, so the analysis grades each check in a vocabulary of 4 — STRONG, LEARNABLE, WEAK, FOREIGN — where the difference between the third and the fourth is not severity. It is whether learning is a plausible route at all.
What the score does downstream
The aggregate lands on a four-rung ladder, and each rung has consequences that are computed rather than editorial:
| Fit | Report score | Caps sales fit at | Kill filter | Decision |
|---|---|---|---|---|
| STRONG | 95 | 10 | PASS | BUILD |
| LEARNABLE | 70 | 8 | PASS | VALIDATE |
| WEAK | 40 | 5 | UNKNOWN | HOLD |
| TOO_FOREIGN | 10 | 2 | FAIL | KILL |
Every cell is the return value of the mapping function the product calls, not a description of it. The middle column is a ceiling on one scorecard category, not a score.
Read the bottom row. It is the only rung that maps to a kill-filter failure, and the filter it fails is not one of the overridable ones — a platform-dependence failure can be carried forward with a flag, and this cannot. A verdict about the founder is a hard kill on the idea.
That is a strong thing for a tool to do, so it comes with a matching obligation: the bottom rung is required to cite a specific fact from your own profile. Not a vibe about the domain being hard — a named thing you said about yourself.
The middle rows are gentler and more common. The point of the cap column is that a mediocre fit does not subtract points from the whole scorecard; it puts a ceiling on the one category it has any business affecting, and leaves the rest of the evaluation to the idea.
The same idea under two founders
Use an invoice-exception monitor for agencies as the fixed idea. Change only the person carrying it:
| Founder profile | Fit rung | Mapped decision |
|---|---|---|
| Former agency operator who owns the buyer list and has run the workflow | STRONG | BUILD |
| Engineer with no agency workflow knowledge, no buyer access and no support tolerance | TOO_FOREIGN | KILL |
The profiles are an illustration; the decision in each row is generated by the real mapping for the stated fit rung. The idea stays fixed so the founder difference is visible.
The sentence that beats the score
Alongside the score, the analysis returns a plain recommendation to keep the idea or drop it. Walk every rung with that flag both ways and the ladder stops being the decision:
| Fit | Kept in the shortlist | Recommended dropped |
|---|---|---|
| STRONG | BUILD | KILL |
| LEARNABLE | VALIDATE | KILL |
| WEAK | HOLD | KILL |
| TOO_FOREIGN | KILL | KILL |
Both columns generated by calling the same projection with the flag set each way. The right-hand column is identical all the way down, including the top row.
A top-rung fit with a drop recommendation lands on the same decision as the bottom rung. That is deliberate: if the analysis has said in words that this idea is not the one for you, letting a good numeric fit outvote it would be the numbers laundering a judgement. Worth knowing because the score is the part people quote and the sentence is the part that decided.
Where this kind of analysis usually lies, and the two rules against it
It invents an attribute about you. The first live evaluation of this mode caught exactly that: checks asserting a founder had, say, a low tolerance for something the profile never mentioned. The fix was an explicit grounding rule — only facts that literally appear in the idea or the profile, and an attribute that was not provided is treated as unknown rather than scored against you. That is the most common failure of founder-fit advice generally, and it is convincing because an invented attribute usually sounds like a reasonable inference.
It hands you a study plan made of nouns. Every gap it names has to survive one test — Could I search for this and get a how-to? — and a bare topic fails it:
Specific enough | Too vague |
|---|---|
| Stripe billing webhooks | fintech compliance |
| How SOAP clinical notes are structured and validated | the buyer's workflow |
| Cold-outbound email sequences to seed founders | distribution to founders |
Both columns are quoted from the contract's own examples. The right-hand column is what a learning plan looks like when nobody applied the test.
There is also a short list of things the output is not allowed to say, and it is the list you would expect from most advice on this topic: no motivational language, and no coaching — no mentors, no advisors, no "get a co-founder". Those are all real options and none of them is an answer to whether you can sell this thing next month.
Running the checks on yourself
- Answer for a named buyer, not a market. "I understand small businesses" is not an answer to any of the checks. Name the person.
- Take the output-quality check seriously. If you cannot tell good output from bad, you are proposing to sell something you cannot inspect.
- Write your gaps as searchable things. Apply the test above. If an item is not something you could start on this week, it is a category, not a gap.
- Separate weak from foreign. Weak closes with effort. Foreign does not, and the honest response to foreign is a different idea, not a longer study plan.
- Notice which answer you are arguing with. The check you want to relitigate is usually the one that was right.
The short version
- 7 checks, and not one is about credentials. It is whether you can build, explain, sell and support this specific thing.
- 4 verdicts per check, and the gap between the last two is learnability, not severity.
- The bottom rung forces a kill-filter failure, on a filter that cannot be softened. A fact about the founder is a hard kill on the idea — and it must cite something you actually said.
- A drop recommendation lands on KILL at every rung, including the top. The score is not the decision.
- Gaps have to be searchable, unstated attributes must not be invented, and coaching is banned from the output.