Customer retention cohort

Identity: unified customer key

Start: first valid purchase period

Window: 30/60/90 days or month

Behavior: repeat, revenue, category

Segment: channel, product, customer type

Fix the population before observing behavior

Overall repeat rate changes with new-customer share. Cohorts hold first-purchase groups constant.

Define first valid order, refunds, identity, window, and repeat event.

Identity determines results

Phone, member ID, platform account, and offline identity can duplicate or disappear. Map sources with confidence and isolate anonymous orders.

Do not force uncertain merges.

Recent cohorts have not completed 90 days. Show completed windows or label immature data.

Apply consistent cancellation and refund rules.

Compare acquisition channel, first product, store, campaign, and type with sample-size controls.

Combine repeat count with basket, margin, and interval; differences are not causal proof.

Test cross-channel identity, anonymous orders, full refunds, first-purchase restatement, and incomplete windows.

BI0 can organize customer and order analysis; identity, cohort templates, and attribution require verification.

Talk through a real business scenario and see how governed AI BI can fit your team.