AI Assistant (Leat MCP)
July 20
→
Retention and cohort analysis
Find out whether the program is getting better at keeping customers, not just how many it has. Follow every group from the month they joined, and compare how they're retaining against the groups before them. Because cohorts are fixed while segments move, this is the one view that shows whether a change to the program actually worked.
Group customers by when they joined, when they first bought, which source brought them in, or which version of the program they met. Follow each group forward and see the share still active at each month since. The curve that results is retention as it happens: steep where members leave early, flat where they stay. Laid out cohort by cohort, it's the shape of the program's ability to keep people.
Time from joining to second purchase. Visits per member by month of tenure. Where in the first ninety days the drop-off happens. How many members of a cohort ever redeem, and how soon. Retention is the outcome; these are the behaviors that produce it, and seeing them by cohort shows whether members who joined this quarter are moving toward loyalty faster or slower than the ones before.
The cohorts who joined before the new welcome journey against the cohorts who joined after. Members who met the old stamp card against those who met the new one. Customers from one signup source against another, or one location against another. When the retention curve for later cohorts sits above the curve for earlier ones, the program got better at keeping people, and you can see when. When it doesn't, you know that too, before another year passes.
The retention curve is what a lifetime value forecast is built on. The point where cohorts drop off is where a lapse journey should trigger, and it's read from the data rather than guessed at ninety days. The behaviors that separate retained cohorts from lost ones are the signals propensity scoring learns from. And whether a change caused the curve to shift, or merely coincided with it, hands off to incrementality measurement and the holdouts that answer it.
Lapsed customer recovery
Spot fading visit patterns across locations before a customer quietly stops coming back
New customer acquisition offers
Reach first-time buyers with an offer that can't leak to existing customers
Program change management
Change a rule once and have every campaign, channel, and decision honor it
Customer lifetime value tracking
Measure real customer value using spend, visits, and redemptions from every channel
Loyalty program setup
Define how the program works before the first customer joins
