If you ask two seasoned operators how to measure product-market fit, one will say "look at your retention curve" and the other will say "run the Sean Ellis survey." They're both right. These are the two signals that actually hold up — and they're not rivals so much as a leading indicator and a lagging one. The trick is knowing which is which.
The retention curve
Take a cohort of users who signed up in the same period and plot the percentage still active over the weeks and months that follow. The shape of that line is the signal.
Without product-market fit, the curve decays toward zero — everyone eventually churns. With fit, it flattens: it drops at first, then levels off at a stable plateau of people who keep using the product indefinitely. Brian Balfour names a flat, "smiling" retention curve as one leg of the product-market fit trifecta, alongside meaningful usage and real growth. It's the truest signal there is, because people vote with behavior, not words — they can't fake showing up.
The cost of that truth is time. A retention curve only becomes readable once you have months of history and enough users per cohort for the line to mean something. Run it on a product that launched six weeks ago and you're reading noise.
The Sean Ellis score
The Sean Ellis survey asks your engaged users one question — "How would you feel if you could no longer use this product?" — and your score is the percentage who answer "very disappointed." The benchmark for fit is 40%.
This is an attitudinal signal: it measures how people feel about losing you, not what they've already done. That sounds weaker than behavior, but it buys you something behavior can't — speed. You can get a meaningful read from a few dozen engaged users this afternoon, long before any retention curve has the history to flatten. And the follow-up questions hand you the "why" — who your fans are and what they value — which a retention chart never will. Here's the 40% rule in full.
Behavioral vs attitudinal — the core difference
That's the whole trade-off. One is a rear-view mirror that's almost impossible to fool but only shows you the road you've already driven. The other is a forward read you can take today, at the cost of being a stated feeling rather than a logged action. Neither is "better." They're answering the same question — do people need this? — from opposite ends of time.
Why they usually agree
Here's the reassuring part: these two signals tend to point the same way. The users who say they'd be "very disappointed" to lose you are, overwhelmingly, the same users who keep showing up. So a strong Sean Ellis score is a reliable predictor of a flattening retention curve — the survey sees the plateau coming before the chart can draw it.
When they disagree, it's usually a timing artifact: the survey has picked up a shift in sentiment that the lagging retention data hasn't caught up to yet, or a small/young cohort is making the curve look noisier than the truth. In that case, trust retention for behavior already banked, and the survey for the early read.
When to lean on each
- Early stage, small numbers: lean on the Sean Ellis score. You don't have the months or the cohort size for a credible retention curve, and you need a signal you can act on now.
- Established, lots of data: the retention curve becomes your bedrock — it's the hardest signal to fake at scale. Keep the survey running for the "why" behind the shape.
- You need to move the number: use the survey. You can't really "iterate against" a retention curve week to week, but you can run the survey each cycle, read the open-ends, ship, and watch the score respond — the way Superhuman walked its score from 22% to 58%.
Get the leading signal today
You don't need months of retention data to know where you stand. Run the Sean Ellis survey and get your PMF score now — free, no signup.
Calculate your PMF score → Built on the Sean Ellis 40% method.Use them together
The strongest setup isn't picking one — it's running both, with the right job for each. Let the Sean Ellis score be your leading indicator and steering wheel: the number you check often, mine for direction, and try to push past 40%. Let the retention curve be your lagging confirmation: the slower, sturdier proof that the fit your survey predicted is showing up in real behavior. When your score climbs and, months later, your curve flattens to match, that's product-market fit you can stand behind — to your team and to your investors.
Track the signal you can act on
PMFtracker runs the Sean Ellis survey, scores how many users would be very disappointed to lose you, surfaces your ICP from the open-ends, and tracks the trend — the leading indicator that moves before your retention curve does.
Start measuring free → Set up in 5 minutes · No credit card required
