Half Your Members Will Be Gone in a Year. The Ones Who Quit Mostly Decided in Week One.
Benchmark data says the average paid community loses 51% of its members annually — and the churners never got far enough in to see the content.

The content library is almost never the problem. The people who cancel didn't get to it.
Start with the benchmark. RetentionCheck puts membership community churn at 5.8% a month — 51% a year — with median ARPU of $42 for operators in the 500 to 50,000 member range. A community with 1,000 paying members loses roughly 510 of them every year at that rate.
Now the reason. The top cancellation driver is low engagement and infrequent participation, at 32% of cancellations. Second is perceived value not justifying cost, at 27%.
Read those two together and you get an uncomfortable picture: the single biggest reason people leave a paid community is that they never really joined it.
The week-one number
At the cheaper end of the market the figures get rougher, and they come from vendors rather than audited data, so treat them as directional. Tools4Skool, a Skool automation vendor, puts average community churn at 12 to 18% a month and says 60% of churners decide to leave in week one.
That price tier is real and large. RevenueGeeks' July 2026 sample of 1,000 trending Skool communities found a median of 205 members and a median price of $27 a month.
The arithmetic at that level is brutal. At $27 a month, a 200-member community at 15% churn loses about $810 a month and needs 30 new members just to stand still. About 14 of every 100 members remain after twelve months, and a $27 member is worth roughly $180 in total.
Most operators respond to that by buying more members. The decision that killed the old ones was made on day three.
What day one through seven has to do
Kourses frames it the same way: the single biggest predictor of long-term retention is what happens in a member's first week. The job of those seven days isn't to impress anyone with the archive. It's to produce one visible outcome and one human interaction before the welcome-email dopamine wears off.
The course-platform data points at the same mechanism. RetentionCheck calls early activation — completing lesson one within 48 hours — the single highest-leverage retention metric.
And the cancellation itself is a lagging indicator. As Kourses puts it, a member who has not logged in for 30 days is gone — they just have not told their credit card yet.
Two levers that aren't content
Billing: annual pre-pay memberships churn at roughly a third the rate of monthly plans, because the commitment horizon extends past the typical quit window of months 2-4.
Price and curation: most paid communities sit at 4-8% monthly, while tightly curated high-ticket rooms above $200 a month with active programming often reach 2-3%.
Neither of those is a content fix. Both are structural decisions made before anyone joins — which is roughly when the retention outcome gets set anyway.
Sources
- Membership Communities Churn Rate: Benchmarks & Analysis — RetentionCheck
- Paid community churn: what 12 to 18% a month really costs — Scale With Fulfillment
- Member Retention 2026: Strategies, Benchmarks, Formulas — Kourses
- Online Course Platforms Churn Rate: Benchmarks & Analysis — RetentionCheck
Editor’s note: Every community operator has watched members vanish without explanation; the benchmark data says the exit was decided in week one, long before the library mattered.
Keep reading
One Extra Second of Loading. That's All It Took to Lose 4.6% of Learners.
A new Nature Human Behaviour study of nearly 82,000 learners puts a hard number on the thing no course creator ever audits: speed.

89% of Fans Trust Your Expertise. Mention Something Once and Half of Them Assume You Don't Actually Use It.
A new Wondery study of 3,000 podcast and video fans found that a single mention reads as an ad. Repetition reads as belief.

The Same Course Finishes at 65% or 42%. The Difference Isn't the Content.
Ruzuku's 2026 platform data finds a 23-point completion gap between courses with active community and courses without. Completion is a revenue number.

The Students Who Used ChatGPT Scored 57.5%. The Ones Who Didn't Scored 68.5% — On a Test Nobody Warned Them About.
A randomized trial in Brazil found AI study help made learning faster and remembering worse. That gap is your refund rate.
