Cohort Analysis

A cohort is a group of users who share a starting point in time (typically the same signup week or month); cohort analysis tracks how their behavior evolves afterward, compared cohort-to-cohort.

Cohort analysis exists because blended, all-time averages hide trends. If overall retention looks flat, it's often because a declining trend in recent cohorts is being offset by strong historical cohorts still in the denominator — cohort analysis separates them so you can see whether the product or acquisition quality is actually getting better or worse over time, not just what the current blended snapshot shows.

The classic output is a cohort retention table or curve: rows are signup cohorts (by week or month), columns are time since signup (Day 1, Day 7, Day 30), and each cell shows what percentage of that cohort was still active at that point.