Cohort retention
Cohort retention groups users by when they joined, then measures how many are still active a week, month, or quarter later. Instead of one blurred average, you see each starting group's curve: the January cohort, the February cohort, and so on. A healthy product shows the curve flattening (retention stabilises) rather than decaying to zero.
For a founder this is the truest signal of product-market fit. New sign-ups can hide churn underneath; cohort retention exposes whether people actually stick. It also tells you if recent changes are working, because a newer cohort retaining better than an older one is hard evidence you improved something real. Read the curve at the point it flattens, not week one, and compare like-for-like cohorts so seasonality or a one-off launch spike doesn't fool you.