The single most important chart for product health. It plots what share of a cohort is still active over time — and its shape reveals whether you have a real product or a leaky one.
▸ Try the interactive toolThe retention curve is the single most important chart for understanding product health. It plots what percentage of a cohort is still active at each point in time after their first use — day one, day seven, day thirty, and beyond. The shape of that curve reveals more about a product's future than almost any other metric.
Three shapes tell three stories. A curve that declines to zero means no one sticks — you have an acquisition machine, not a product. A curve that flattens into a plateau (the 'smile' or stabilising curve) means a core of users found lasting value — a real product with a foundation to grow on. A curve that smiles upward (rises after the dip) is the rarest and best — users not only stay but deepen engagement. The single most diagnostic question is: does the curve flatten, or does it go to zero?
Declines to zero — no retention; an acquisition machine, not a product.
Flattens to a plateau — a core finds lasting value; a real product.
Smiles upward — users deepen over time; the rare ideal.
The key question: does it flatten, or hit zero?
A flattening retention curve is the closest thing to proof of product-market fit there is — and a curve that decays to zero is proof of its absence, no matter how good acquisition looks.
| The shortcut | What it costs | What it gives you instead |
|---|---|---|
| Ignoring the curve's shape | Watching acquisition while the curve decays to zero. | The shape reveals whether you have a product or just sign-ups. |
| No plateau | Users all eventually leave; no durable core. | A flattening curve shows a core found lasting value. |
| Reading only day-1 retention | Early retention can look fine before the curve collapses. | The full curve shape matters more than any single point. |
| Acquisition masking decay | Pouring in users hides that none of them stay. | The curve exposes the truth regardless of acquisition. |
For a cohort, plot the percentage still active at day 1, 7, 30, 90. The curve, not any single number, is the diagnostic.
Ask the key question: does the curve flatten into a plateau, or decline toward zero? A high day-1 number means little if the curve later collapses.
A stabilising curve reveals the core of users who found lasting value — your real product. No plateau means no durable value yet, which is the deeper problem to solve.
Are newer cohorts retaining better than older ones? An improving curve shape over cohorts means the product is genuinely getting stickier (ties to cohort analysis, Tool 09).
No plateau → the priority is finding and delivering lasting value, not acquiring more. A low-but-flat plateau → raise the plateau. Let the curve direct the work.
A product had strong day-1 retention and rising sign-ups, and the team felt good. But plotting the full retention curve told the real story: it declined steadily toward zero — there was no plateau. Users tried the product and almost all eventually left. The strong acquisition was masking the absence of any durable core.
The curve's shape was unambiguous: an acquisition machine, not a product. No amount of additional acquisition would fix it — they were filling a bucket with no bottom. The work shifted from growth to finding why no core of users stuck, and only once the curve began to flatten (a real plateau emerged) did scaling acquisition make sense.
The deliverable is the cohort retention curve, read for its shape — the most diagnostic chart of product health.
| Curve shape | Means | Implication |
|---|---|---|
| Declines to zero | No durable value | Not a product yet — find the value |
| Flattens to a plateau | A core found value | Real product — raise the plateau |
| Smiles upward | Users deepen over time | The rare ideal — protect & scale |
The retention curve answers the most fundamental product question there is: do people, having tried this, keep coming back? Its shape — flatten or zero — is the clearest signal of product-market fit a single chart can give.
What makes the curve uniquely powerful is that it's almost impossible to fool. Acquisition can inflate every top-line number, vanity metrics can flatter a dashboard, but the retention curve simply shows what fraction of users stay — and a curve that decays to zero exposes the absence of lasting value no matter how impressive the growth looks. The flattening, conversely, is genuinely good news: it means some segment found durable value, which is the foundation everything else builds on. This is why the diagnostic question is binary and brutal — flatten or zero — and why the right response to a non-flattening curve is never 'acquire more' but always 'find why no one stays.' Retention is the foundation; the curve is how you read it.
Sign-ups mean nothing if the curve goes to zero. Read the curve's shape.
A good start can precede a total collapse. The full curve matters.
More users into a bottomless bucket is pure waste. Fix retention first.
Whether newer cohorts retain better is the improvement signal. Compare them.
Retention curves are plotted by cohort (Tool 09) — the two are inseparable.
The curve shows retention shape; growth accounting (Tool 07) shows the new/retained/churned mix.
A flattening curve is the strongest quantitative PMF signal (Module 1, Tool 22).
The retained core is who the North Star (Tool 04) ultimately measures value for.
Sketch two retention curves from memory or imagination: one that declines to zero, one that flattens to a plateau. For each, state what it says about the product.
Then ask: for the zero curve, would more acquisition help? For the flat one, what would 'raising the plateau' mean?
If you concluded that acquisition can't fix a zero curve — only finding lasting value can — you've understood why the curve's shape, not its starting height, is the real verdict on a product.