Loyalty glossary · 7. Data and analytics (31)
Retention Rate
Retention rate is the percentage of members from a defined starting cohort who remain active after a chosen period, usually a month or a year. It measures whether a loyalty programme is keeping its members engaged over time, independent of new joins or one-off activity.
The convention is to report retention over a rolling month or a year, but the choice of window changes the answer enough to make cross programme comparisons meaningless. A monthly figure looks stronger than an annual one, and a programme that hides behind a monthly number is not being honest about the shape of its member base over time.
Acquisition is loud; retention is quiet. A programme that adds many new members while losing a similar number looks successful on a dashboard, but if its retention rate is falling the base is eroding underneath the growth. The right question is not how many joined last quarter but how many of last quarter's joiners are still active this quarter.
Programmes that rely on activity-based qualification to define retention are measuring effort, not loyalty. A member who earns a tier through a single burst of purchases may be counted as retained for the year, but that classification hides an inactive future. Retention should be measured on recurring engagement, not on accumulated thresholds.
The actuarial model should treat retention as the primary driver of future liability. Every accrual for unredeemed points assumes a pattern of future redemptions, and that pattern depends on how many members stay engaged. Understate retention and the programme overstates breakage; overstate retention and the deferred revenue is a fiction.
Work the arithmetic because compounding is not intuitive. A programme that retains 90 percent of its members each month keeps 90 percent from one month to the next. Over 12 months, the 90 percent compounds to 28 percent of the original cohort still active. Drop the monthly rate to 80 percent and the 12 month survival falls to 7 percent. A difference of five points each month produces a fourfold difference in annual survival.
Retention rate is not a single number; it is a curve. Programmes that report one headline figure are hiding the fact that retention decays differently by tenure, by tier, and by channel. The honest metric is a cohort retention table, and any programme that cannot produce one is guessing.