Aha-moment finder
Find which early action, repeated how often and within how many days, predicts lasting value.
End of analysisNext steps and reference material below. Not part of the analysis result.
Next step
What resonates
Full journey
Which action should I drive to reduce early churn?
Find the action count and window most associated with users who stay.
Method and FAQOpen only when you need the methodology
See the search method and interpretation limits
When to use it
Use it to prioritize onboarding work, to choose an optimization event for campaigns, or to locate where retention diverges. You need an event CSV with user id, event name, and timestamp.
What it calculates
For each window-and-count pair, threshold membership is crossed with long-term outcomes to compute lift and F1, and only candidates with enough sample rise to the top. Mapping a segment column splits the search by value.
How to act
Treat the threshold as an experiment hypothesis, not proof that driving the action raises retention. Validate with an onboarding change or an event switch, and let A/B or incrementality carry the causal claim.
Frequently asked questions
If I drive this action, will retention improve?
Not necessarily. The relationship is observational—users who would have stayed anyway may simply perform the action more. Causality requires an experiment.
What if I have many event types?
More combinations mean more candidates that look good by chance. Read sample size and lift stability together instead of trusting the single top row.
What data is required?
A row-level CSV with user id, event name, and timestamp. A long-term outcome column makes the lift calculation more reliable.