Product-market fit

Product-market fit is the gate between search and scale. Marc Andreessen's wording is still the useful one: a good market, and a product that satisfies it. You will hear fit described as a feeling — customers pull product out of you. The feeling is real. Feelings also lie.

Pair the intuition with numbers. Retention cohorts say what people do. The Sean Ellis survey says how upset they would be to lose you. If those disagree, segment before you declare victory.

Until fit is real, the company should stay small and learn. After it is real, you earn the right to hire and systematize. Scaling spend first is a top way to die — see why startups fail.

A useful leading proxy is Sean Ellis's question: how disappointed would you be if you could no longer use this product? Companies where ≥40% of recent core users say "very disappointed" tended to grow. Superhuman started around 22%, narrowed to the customers who loved them, and later printed about 58%. Retention remains the tie-breaker: surveys are what people say, cohorts are what they do.

Concepts in this section

Step	Concept	What it answers
1	What fit means	What did Andreessen actually require?
2	Measuring fit	Retention, the 40% Ellis test, Superhuman's loop
3	False positives	Why traffic, friends, and founder sales fool you
4	Before and after	What to do in search, and what changes after the gate

A short picture of the loop you will run in search:

As an early employee, ask how the company defines fit in numbers. "We feel it" without retention or survey data is a yellow flag. As an engineer, reliability and cost become the product after fit; speed of learning dominates before.

When roadmap debates stall, ask: does this deepen fit for our best segment, convert fence-sitters, or only please people who would never be very disappointed? Prioritize the first two.

Common mistakes

Scaling spend during search. Ads and a sales army before repeatable demand.
Trusting launch spikes and press. Judge the cohort four weeks later.
A blended Ellis score with no segments. A 35% miss can hide a 48% beachhead.
Treating fit as permanent. New ICPs, prices, and products restart the test.

Key takeaways

Fit is a good market plus a product that satisfies it.
Watch retention cohorts and organic pull — not launch-day signups.
The Sean Ellis bar (≥40% "very disappointed") is a useful leading indicator.
Distrust founder heroics, friends, press, and one-off logos.
Fit is segment-specific and perishable. Keep measuring after you "have it."