Analysis
Retention rate
Also called repeat rate.
Retention rate is the share of a cohort that comes back or stays in a later period. It is the mirror of churn.
How it is measured
Take the starting cohort, count those active again in the later period, and divide by the starting size. You must define active: a visit, a login, a purchase. Keep the rule fixed.
Plot it by period since start, not by calendar date. A curve that flattens shows a base of loyal users; one that keeps falling shows a leak.
Worked example
A fitness-class booking site had 1,640 people make a first booking in January. In February, 410 booked again, a month-one retention of 25 percent. By April 262 are still booking, 16 percent.
The flattening between March and April suggests a core group. The studio reaches out to the 410 who stayed to ask what drew them back, and moves the free trial class from week one to week two for new people.
How it differs
Retention rate counts who stayed. Churn rate counts who left. Over the same cohort and window they sum to 100 percent. Retention gets attention in growth reports, churn in finance.
Common errors
Counting returning visits rather than returning people. Using a loose definition of active. Comparing different-length windows. Mixing cohorts. Judging from a single cohort.
In practice
Pick one repeat action as the definition and plot a retention curve for the last six start weeks. Look for where the curve flattens, then check what the users there have in common.