Analyzing user behavior by breaking them into groups (cohorts) based on a shared characteristic, typically the month they signed up.
Why it matters: It is the only way to accurately measure retention over time. Looking at blended averages hides whether recent product updates are improving or hurting retention.
In practice: Tracking the 'January Signups' cohort to see what percentage remain active after 3 months, compared to the 'February Signups' cohort.