Action survival and drop-off analysis

See when a key action drops off over time and which group should be tested first.

Loading…

End of analysisNext steps and reference material below. Not part of the analysis result.

Next step

Proof

Common next stepPrepare a new dataset
Aha-moment finder
Full journey

How long does a defined key action persist, and when does drop-off risk rise?

Estimate survival, hazard, and observed recurring value from censored action episodes.

Method and FAQOpen only when you need the methodology
See the calculation conditions and limits

When to use it

Use it to determine when an activation, return, repeat purchase, or key-feature action drops off and where to place an intervention. Each row must be one entity episode from action start until exit or observation end.

What it calculates

It calculates Kaplan–Meier survival, Greenwood 95% intervals, interval hazard, median survival, and restricted mean survival time within the observed horizon. Periodic recurring value and gross margin optionally produce observed-horizon value.

How to act

Form an onboarding, reminder, product, or pricing hypothesis before the risk interval, then validate it with an A/B test or holdout. Segment differences are associations, and neither survival nor value is extrapolated beyond observed data.

Frequently asked questions

Should cases still active at observation end be included?

Yes. They are censored observations, not cases that never exit. Excluding them biases survival downward.

Does a segment difference prove a cause?

No. Channel, plan, customer mix, and timing can differ together. Treat it as a directional observation until an intervention is tested.

Does this predict drop-off for individual entities?

No. It summarizes time-based survival and hazard for action episodes. Individual scoring is a separate modeling problem.