Early traction charts flatter you. Signups, app downloads, and pageviews all climb when you spend on ads or catch some press, yet they say nothing about whether people actually value what you built. Product-market fit is not a feeling or a fundraise. It is measured by whether users come back on their own, without you paying to drag them back. Two honest tools cut through the noise: retention cohorts and the Sean Ellis 40 percent test. Together they tell you whether you have a product people would genuinely miss.
Why retention is the honest signal
A cohort is simply a group of users bucketed by when they first used your product, for example everyone who signed up in January, then everyone from February, and so on. Cohort analysis tracks what percentage of each group is still active after one week, one month, three months, and beyond. Unlike a total user count, a cohort cannot be inflated by fresh ad spend. It shows you the truth about the users you already won: how many of them stayed.
Reading the retention curve
Plot the percentage of a cohort still active against time and you get a retention curve. In almost every product it drops steeply in the first four to eight weeks. What matters is what happens next. Without fit, the curve keeps sliding toward zero, meaning everyone eventually leaves. With fit, the curve flattens into a plateau, because a stable core of users has built your product into a habit and keeps returning. The height of that plateau matters too: a curve that flattens at 30 percent is far healthier than one flattening at 5 percent. Aim to see your cohorts flatten by month three to six. For an even stronger signal, check whether retained users do more or spend more over time rather than less.
The Sean Ellis 40 percent test
Retention shows behaviour. The Sean Ellis test shows intensity of need. Sean Ellis, who helped scale Dropbox and later founded GrowthHackers, popularised one survey question: "How would you feel if you could no longer use this product?" Users choose from Very disappointed, Somewhat disappointed, or Not disappointed. His benchmark, drawn from surveying hundreds of startups, is that if 40 percent or more answer "Very disappointed," you likely have product-market fit. Below roughly 25 percent, you do not yet.
One rule makes or breaks this test: only survey people who have genuinely experienced your product. Superhuman, using this method, surveyed only users who had used the product at least twice in the last two weeks, and reports raising its score from 22 percent to 58 percent within about a year by doubling down on what its "very disappointed" fans loved and fixing what held the fence-sitters back.
Running it as a first-time founder
- Use free tools. Mixpanel, Amplitude, and Google Analytics all have free tiers that build cohort retention charts with little engineering effort.
- Reach real users. Send the survey by email or WhatsApp, which most Indian users check daily, and filter to people who have actually used the product recently.
- Get a real sample. A percentage from five replies is noise. Wait for a meaningful sample, ideally 40 or more responses, before you trust the number.
- Always add follow-ups. Ask the "very disappointed" group what they would miss most, and ask the "somewhat" group what is missing for them. Those answers become your roadmap.
What to do with the answer
If your cohorts flatten and 40 percent or more would be very disappointed, stop tinkering with the core and pour your energy into growth and distribution. If not, resist the urge to scale. Study the users who already love you, learn exactly who they are and why they stay, and rebuild for them. Honest measurement this early saves you from spending years, and your runway, scaling a product the market does not actually miss.