Most cohort-based businesses (courses, coaching programs, communities) find out a member is about to churn the day they cancel. By then it's too late.

After building dashboards for dozens of cohort operators, here's the pattern: churn is predictable 2-3 weeks out if you're watching the right signals. Most people watch the wrong ones (logins, vanity engagement) and miss it.

The 6 numbers that actually matter:

1. Weekly Active Rate — % of active members who engaged in the last 7 days. A dip here is your earliest warning sign, before anything else moves.

2. Cohort Retention Curve — retention plotted by week-since-join, compared cohort-over-cohort. This tells you if your newest cohort is retaining worse than your last one — a leading indicator of an onboarding problem.

3. Revenue Per Member — tracks whether you're growing loyal payers or just adding low-intent members.

4. Time-to-First-Value — how long it takes a new member to hit their first real win. The longer this stretches, the higher your churn risk.

5. Support/Complaint Rate — spikes here precede cancellations by roughly 2-3 weeks in most cohorts we've measured.

6. Net Promoter Signal — a lightweight, ongoing pulse (not just a survey once a quarter).

The move: don't look at these in isolation. Combine them into a single Weekly Health Score per cohort, and review it for 15 minutes every Monday. That's the whole system — no fancy BI tools required.

We built a plug-and-play version of this (template + course + weekly review framework) if you want to skip building it yourself — check the Cohort KPI Dashboard product on this page.

Happy to answer questions on any of these 6 metrics below 👇