Answered by 4 people who ran growth · Updated
When the share of users doing your product's key action holds steady across consecutive cohorts, read over at least twelve weeks — and flat alone is still not permission to spend.
Most founders default to logins or page views, which is the wrong signal. The y-axis should be the percent of users performing the key action — the thing that cannot happen unless value was delivered. At Grubhub that was placing an order; at Pinterest, saving a pin.
Flattening is necessary but not sufficient. A curve that settles at 2% is flat and still won't support any acquisition economics. The real test is whether you can recover acquisition spend inside a three-month payback period.
The mistake is putting a vanity action on the y-axis, like “opened the app.” That tells you about your funnel, not your product. If you can't name the thing that makes your product worth coming back to, the retention curve is the wrong problem to be solving right now.
Stability matters more than the absolute level. A curve holding at 15% D30 across four cohorts is a better signal than one that bounced between 12% and 22% and averaged 18%. And there is no universal “good enough” number — anyone who gives you one without knowing your category is guessing.
Seed-stage teams consistently declare victory too early because they're looking at 4-week cohorts when they need 12-week cohorts. A curve still declining 1-2 points per period is not flat — it's falling slowly.
Stress-test the curve by acquisition source before you conclude it's ready. If you've been growing on word of mouth, expect it to drop when you open up paid — the curve you see is partly a function of who you've been acquiring.
Founders looking at a 2-week curve and declaring victory are almost always seeing the novelty bump flatten, not true retention. You want at least two full usage cycles past the initial drop before you trust the floor.
Before opening the tap I want three things together: the curve flat for at least two consecutive cohorts, retained users showing organic referral behaviour, and unit economics on even a small paid channel trending the right way. Any one alone is not enough.
Ask a follow-up
Group chat with Casey, Brian, Adriel, Luc
With the retention curve. Until it flattens at a non-zero level, growth work is premature.
After product-market fit, and after the founders have drafted the growth model themselves.
Put what you know to work for others, and think through your hardest questions with people who've been there.
© 2026 Super Me Network, Inc.
