What it means
Product-market fit is the state where a defined group of people want your product enough that keeping up with them becomes the problem. Before it, you're pushing: every user is individually persuaded and individually retained. After it, you're pulled: people arrive, stay, and tell others without you engineering each step.
It's specific to a market, not absolute. A product can have clear fit with regional travel agencies and none at all with large tour operators. "Do we have product-market fit" is an incomplete question until you say with whom.
Signals, real and false
| Real signal | False signal |
|---|---|
| Users return without prompting | Signups after a launch post |
| People complain when it's down | People say they like it |
| Users tell colleagues unprompted | Followers and mailing list size |
| They'd be genuinely annoyed if it disappeared | They'd "probably keep using it" |
| Support asks about doing more, not about basics | Press coverage |
| Retention curve flattens | A growing total-users chart |
The last row is the one that matters most and is most often misread. Cumulative user counts only ever go up, including for products everyone abandons. They are a chart of your marketing effort, not of your product.
Retention is the evidence
The clearest measurable signal is a retention curve that flattens. Group users by the week they joined and track what share are still active weeks later. Every product loses people early. The question is whether the line eventually goes horizontal.
Week: 0 1 2 4 8 12No fit: 100% 38% 19% 8% 3% 1% -> heads to zeroFit: 100% 45% 34% 30% 29% 28% -> flattens ~28%A flat line means you've found a group for whom this is genuinely useful. The percentage it flattens at matters less than that it flattens at all — and the flat portion tells you who your actual market is, which is often narrower than the one you were aiming at.
Before it, and after it
| Before fit | After fit | |
|---|---|---|
| Main activity | Talking to users, changing the product | Serving demand, removing bottlenecks |
| What to avoid | Scaling anything | Ignoring the retention curve |
| Hiring | Almost none | Support and delivery first |
| Marketing spend | Small tests to learn language | Scaling what already converts |
| Roadmap driven by | What's blocking usefulness | What's blocking growth |
The expensive mistake is running the right-hand column's playbook while in the left-hand column's situation. Hiring, ad spend, and process before fit all make the same wager — that the product is finished — and they make being wrong much more costly.
Getting there
- Narrow the market until something worksFit with a small, specific group beats indifference from a large one. Narrowing is nearly always the right move when nothing is sticking.
- Find the users who did stay and study themEven a failing product usually has a handful of committed users. What they have in common is often the market you should have been aiming at.
- Talk to the ones who left, specificallyChurned users give you the most direct information available and are the least contacted.
- Change one substantial thing at a timeChanging the audience, the pricing, and the core flow simultaneously means learning nothing from the result.
- Watch the cohort curve, not the totalIt's the only chart that can tell you the answer.
Common mistakes
- Reading a rising cumulative user chart as evidence of fit
- Measuring retention on logins rather than on the valuable action
- Hiring and spending as though fit exists because growth looks good
- Asking about fit without specifying which market
- Widening the audience when nothing sticks, instead of narrowing it
Key takeaways
- Fit is when the market pulls rather than you pushing — and it's always fit *with someone*
- A flattening cohort retention curve is the clearest evidence available
- Total-user charts go up regardless and prove nothing
- Before fit, change the product; after fit, remove bottlenecks
Try it yourself
Build one cohort retention table: group users by join week, and count how many performed the valuable action in each week after. If the line still descends at week eight, that's the most useful thing you'll learn this month.