A day-thirty percentage compresses a month of behaviour into one cell. Two cohorts can arrive at the same number through very different paths: one loses most users immediately and then stabilises; another decays steadily; a third returns on a monthly schedule that daily retention represents poorly.
Start with the expected cadence
Daily use is natural for messaging and less meaningful for payroll, travel or a monthly membership task. Choose a return window that reflects when value can reasonably recur. Rolling retention answers whether someone came back on or after a date; bounded retention asks whether they returned inside a defined interval. Those are not interchangeable.
Look at where the slope changes
A sharp first-session fall can point toward acquisition fit, expectation setting or failure to reach initial value. A later bend may align with a trial ending, content exhaustion or a recurring task. Release markers and lifecycle messages belong beside the curve, not in a separate presentation.
Treat the plateau carefully
A flattening line may indicate a durable user group, but identity loss can create a false decline and sparse observation can create a false plateau. Check cohort size and confidence before telling a habit story.
The useful question is not simply “what is D30?” It is “which part of the curve can the team influence, and what evidence distinguishes the competing explanations?”