Why blended churn lies
You ran a Twitter launch in March (45% M3 retention) and a paid YouTube ad in April (12% M3 retention). Blended churn says 25% — useless. Cohort view tells you to kill YouTube ads and double Twitter. Same data, opposite decision.
What the table shows
Rows: cohort by join week. Columns: % retained at M0, M1, M2, M3, M6, M12. Cells colored by retention band — green ≥70%, yellow 40–70%, red <40%. Hover a cell to see the exact members who churned that month.
Decisions it unlocks
Channel ROI (which acquisition channel returns positive LTV). Onboarding changes (M1 retention jumps after you ship a welcome DM sequence). Price elasticity (M3 drop after a price increase). Seasonality (cohorts joining in December retain worse — pause ads in Q4).
Wire it live against your community.
tools4skool plugs into Skool and runs these analytics on autopilot — engagement, cohorts, MRR.
Book a demo →Frequently asked
How many members do I need for cohort analysis to be useful?
Roughly 30 members per cohort, so 30/week = ~1500/year minimum. Below that, cohorts are too noisy — use simple monthly churn and member-level surveys instead.
Where does the data come from?
tools4skool exports member join date, plan, last active, and cancellation date from your Skool community. Cohorts are computed in your browser — nothing leaves your machine.
Can I see cohort by source/UTM?
Yes, if you tag members at signup via the tools4skool extension. Each cohort can then be sliced by source — twitter, youtube, referral, organic.
Is M3 retention or M6 the better KPI?
M3 — long enough that trial-curiosity members have churned, short enough to iterate quickly. M6 is great for board decks but too slow to drive weekly decisions.