What is a good retention curve and how do I know if mine is flattening?
The short answer
Plot the percent of a cohort still active by week or month since signup, not cumulative churn. A healthy curve drops fast at first then flattens into a horizontal plateau, that flat line is your real retained base and the honest measure of product-market fit. If your curve keeps sliding toward zero with no plateau, you do not have retention yet and pouring money into acquisition just fills a leaky bucket faster.
Go deeper, your way
3 hand-picked resources, 3 link-checked. Pick how you want to dig in.
✍️ Essay
✓ Link checkedFreeIntermediate
Why we picked it
This is the clearest essay on why a flattening cohort retention curve is the thing that actually matters, and it lives on Andrew Chen's blog where the flattening-curve idea got popularized. It walks you from week-to-week churn to why the curve should stop dropping and level off, and makes the honest case that virality is worthless until retention is stable. A good starting point if you are staring at a retention chart and trying to figure out whether it is bottoming out or still bleeding.
A healthy curve drops fast in the first few weeks, then flattens into a plateau. The plateau, not the starting number, is the signal you are looking for.
If your curve never flattens and keeps sliding toward zero, you do not have retention yet, and chasing more users just fills a leaky bucket faster.
Fix retention before you spend on growth or virality: users who churn cannot invite anyone, so retention compounds where top-line growth alone does not.
Why we picked it
Once you know your curve should flatten, the next question is where it should flatten, and this piece answers it with concrete benchmark ranges by business type instead of hand-waving. Lenny pooled numbers from 20-plus growth leaders and investors, so you can put your own 6-month retention next to a real bar for your category. Treat the ranges as a starting point for judgment, not a pass-fail line, since a consumer social product and an enterprise SaaS live in completely different worlds.
The bar is category-specific: roughly 25% good / 45% great for consumer social, 40% / 70% for consumer subscription, and 70% / 90% for enterprise SaaS at 6 months.
Judge yourself against your own business type, not a global average. Comparing a marketplace to a SaaS number will just mislead you.
It also splits user retention from net revenue retention, a useful reminder that a subscription business can look flat on users while still growing revenue per account.
Why we picked it
Benchmarks are useless until you can actually plot your own cohort curve, and Mixpanel's free tier lets you build a real retention chart without writing SQL. This guide covers the two knobs that trip people up first: the retention criteria (which event starts the cohort, which event counts as returning, and the time bucket) and how to read the resulting grid. It is a genuinely free starting point for an early-stage founder, though you will point to the sample board and cohort features to go deeper.
From
Mixpanel Docsby MixpanelDocs guide, ~15 min to set up
Set your starting event (for example Signup) and your return event, then pick a time unit (day, week, or month) that matches how often your product should be used.
The retention grid reads down as cohorts and across as periods, so a flattening curve shows up as columns that stop shrinking to the right.
Mixpanel's free plan is enough to build this, so you can measure your curve before spending anything on analytics tooling.
