Milestone map
Milestone map
3 milestones
Define the PMF signal you are targeting — and why
1–2 weeks (definition) + ongoing measurement
Define, before you measure it, what product-market fit will look like for your specific business. PMF is often described as a feeling ('you know it when you have it') but that description is not actionable. The most operationalised PMF signals are: Sean Ellis's '40% of users would be very disappointed if the product disappeared', NPS above 50, retention curves that flatten rather than trend to zero, or a specific cohort-based retention benchmark for your category. The choice of PMF signal matters because different signals are appropriate for different business types — a consumer app and a B2B enterprise product have fundamentally different PMF indicators. Defining your signal before measuring it prevents post-hoc rationalisation.
Proof required
Write a PMF hypothesis document covering: (1) the PMF signal you are targeting and why it is the right signal for your specific business model; (2) the current measurement result for that signal (even if below target); (3) the leading indicators you are watching that might predict PMF before the lagging signal confirms it.
What gets checked
- PMF signal is specific and pre-defined — not 'when it feels right' but a named quantitative signal with a target threshold.
- Current measurement is shared honestly — if you are below the target, that is the most valuable data in this document.
- Leading indicators are named — what do you watch day-to-day that predicts PMF before the signal confirms it?
Common mistakes
- PMF signal defined after the fact: 'we chose this signal because we've already achieved it' — defining PMF on outcomes you've already seen is not a hypothesis, it is a rationalisation.
- No current measurement — 'we haven't measured yet' means the PMF hypothesis has no evidence basis.
- PMF signal is qualitative only: 'customers say they love it' without a quantitative threshold.
Resources
Foundationstart here
Depthgo deeper
Arena
What a verifier looks for
- The PMF signal must be pre-defined — ask 'when did you define this signal?' and whether it preceded the current measurement.
- Current measurement should be honest — a business that claims to have PMF at M1 without evidence is either very early or rationalising.
- Leading indicators should be specific — ask 'what metric do you look at every morning that tells you whether you're moving toward PMF?'
You'll sign in first, then come straight back here.
Cross the PMF threshold — evidence from the target signal
4–12 weeks (depending on measurement cycle)
Cross the PMF threshold you defined in M1. The evidence requirement is simple: show the PMF signal measurement crossing the threshold. The more important requirement is the context: PMF is not a permanent state, it is a signal at a specific time with a specific cohort. The proof must state what cohort was measured, when, and whether the signal has been consistent across at least 2 measurement periods.
Proof required
Share the measurement result for the PMF signal you defined in M1 — the data showing you have crossed the threshold (survey results, retention curve screenshot, NPS data, or equivalent). State the cohort: who was measured (paying customers, users who signed up in the last 90 days, a specific segment) and when. State whether you have measured this signal twice and whether both results were above the threshold.
What gets checked
- PMF signal measurement shown with actual data — not a description of the results but the data itself.
- Cohort is defined specifically — PMF measured on all users ever is weaker signal than PMF measured on users who joined in the last cohort.
- Two measurement periods covered if available — single-measurement PMF is preliminary; consistent PMF signal across 2 periods is stronger.
Common mistakes
- PMF signal changed between M1 and M2 — if you changed the signal because you couldn't reach the original threshold, that is not PMF.
- Cohort definition is 'all users' without narrowing — PMF measured on all-time signups conflates highly engaged early users with disengaged recent users.
- Single measurement only — one survey result is weak; two results above threshold are the minimum for a credible PMF claim.
Resources
Foundationstart here
What a verifier looks for
- The measurement data must match the M1 signal definition — if the signal changed, ask why.
- Cohort definition matters — ask 'did you measure all users or a specific recent cohort?' A recent cohort result is more credible for PMF.
- Two measurement periods: ask for the first measurement date and the second. If only one measurement, ask when the next one is planned.
You'll sign in first, then come straight back here.
Describe the segment with PMF — and the segment without it
2–4 weeks (analysis and write-up)
Identify both the segment that has PMF and the segment that does not. PMF is almost never universal — it is usually concentrated in a specific customer type, use case, or geography. Articulating the segment with PMF (and the signal that reveals it) and the segment without PMF (and why) is the most actionable insight from the PMF measurement process. It tells you where to focus acquisition and where not to. This is the insight that separates founders who 'found PMF' from founders who understand what they found.
Proof required
Write a PMF segmentation analysis covering: (1) the specific segment in which you have PMF — named by customer type, use case, or observable characteristic; (2) the specific segment in which you do not have PMF — and your hypothesis for why; (3) what the existence of a PMF-negative segment revealed about your original product assumptions.
What gets checked
- PMF-positive segment is named specifically — not 'engaged users' but 'solo technical founders building their first B2B product who discovered us through organic search'.
- PMF-negative segment is named honestly — not 'people who don't understand the product' but a specific customer type or use case where your product genuinely does not work well.
- What the PMF-negative segment revealed is specific — at least one product assumption that turned out to be wrong for that segment.
Common mistakes
- PMF-negative segment absent: 'we have PMF with all our customers' is almost never true and suggests the segmentation analysis was not done.
- PMF-positive segment described by behaviour only: 'people who use the product regularly' — behaviour is a proxy for segment; the segment is a type of customer with observable characteristics.
- What the PMF-negative segment revealed is absent: 'they just weren't the right fit' without naming what assumption was wrong.
Resources
Foundationstart here
What a verifier looks for
- The PMF-positive segment description should be specific enough to be a targeting criterion — ask 'if you wanted 10 more customers exactly like your PMF-positive segment, how would you find them?'
- The PMF-negative segment should be a real customer type you have, not a hypothetical — ask 'do you have current customers in this segment?'
- The assumption the PMF-negative segment revealed should be traceable to an original product decision — ask 'was this assumption in your original pitch deck or founding hypothesis?'
You'll sign in first, then come straight back here.
Part of