Been tracking churn patterns across different cohorts and noticed some weird shifts lately.
Wondering what models other teams are using to predict dropoffs and what’s actually moving the needle on retention right now.
Switched from standard cohort analysis to a scoring model 8 months ago. Instead of just tracking usage, we look at in-app actions within the first 72 hours.
Users completing 3 or more core actions in their first session stick around 40% longer. We redesigned onboarding to encourage these behaviors.
For retaining users, push notifications at day 14 are more effective than at day 7. It seems counterintuitive, but data shows day 7 can feel too pushy, while day 14 hits when users start to lose interest.
Also testing a “pause subscription” option instead of just canceling. Early results show 30% of those pausing return within 60 days.