Customer Retention Playbook

A 5% increase in retention can increase profits by 25-95%. Those numbers get quoted endlessly and acted on almost nowhere, because retention requires a system — segmentation, triggered flows, programme design — not a hope and a monthly newsletter. The playbook is that system.

Segmentation is RFM-based: recency, frequency, monetary value. Behaviour, not demographics. A 25-year-old who buys monthly is more valuable than a 45-year-old who bought once. The seven priority segments — champions, loyal customers, potential loyalists, at-risk VIPs, one-timers, lapsed, lost high-value — each get a different action, because sending the same campaign to a VIP and a one-timer wastes both.

The single highest-impact play is converting one-timers to two-timers. The probability of a third purchase jumps from roughly 30% after order one to roughly 50% after order two — the second purchase is the hardest and most valuable transition. The playbook concentrates disproportionate effort on it: a three-email sequence starting Day 14, with content matched to product type. Consumables get a replenishment frame ("running low?"). Durables get a complementary cross-sell. Experiences get a "here's how to get more from your [product]" frame. Not the same email rebranded.

Win-back flows have four stages and only one carries a discount. Stage 1 is soft ("we miss you" plus a recommendation). Stage 2 surfaces what's new since they last bought. Stage 3 introduces a meaningful offer. Stage 4 is the explicit "is this goodbye?" — and after that, you stop, because the probability of reactivation past 3x the average purchase cycle is below 5% and you're better off spending the budget elsewhere.

The loyalty programme is matched to the model. Points work for frequent lower-AOV purchases. Tiers work for higher-AOV aspirational brands. Hybrid works best for most. The skill won't recommend a 4-tier programme with 12 perk levels for a brand with five SKUs — a programme too complex to maintain does more harm than no programme at all. VIPs get a different treatment entirely: personal outreach, surprise gifts, early access, founder contact — the things automated emails can't replicate and competitors can't discount away.

Honest about new stores. If you have under 500 customers or no purchase history, the skill won't pretend to segment. You get benchmark targets (27% repeat rate for general e-commerce, 40%+ for consumables) and a focused plan: build the data collection, build the second-purchase campaign, come back for the segmentation architecture once you have 1,000+ data points. Subscription and SaaS businesses get a different framework — RFM's Recency becomes Engagement Recency (last login, last feature use), because their churn is continuous, not purchase-gap based.