Cohorts for loyalty managers who are not analysts

Cohort work sounds technical, yet the core idea is ordinary: people who joined under the same conditions tend to behave more alike than people who joined years apart under different rules. Loyalty managers already think this way when they remember a festive campaign that brought in bargain hunters.

Pick one framing at a time. Join month, acquisition partner, or store cluster is enough. Mixing every label into one chart creates noise that invites false confidence. Once the cohort is fixed, track three measures for the first ninety days: first redemption, second visit or earn, and any support contact.

Compare cohorts only after you confirm the programme rules were stable. A mid-period change to earn rates or tier thresholds can make an older cohort look “worse” for reasons that have nothing to do with member quality. Note rule changes beside the cohort labels so stakeholders do not misread the story.

Bring the comparison into a briefing, not a slide of charts alone. Say what the better-performing cohort received in the welcome journey, what the quieter cohort missed, and one change you will test next. Membership behaviour analysis earns its keep when it changes a decision, not when it fills a deck.