Lifecycle Stages
Date: 2026-08-16
New, active, lapsing, lapsed, reactivated. The stages themselves are obvious; the value is in the transitions, because that’s where an intervention can still change the outcome.
What it is
Lifecycle stages classify each customer by where they are in their relationship with you, using rules derived from order recency and count.
┌──────────┐ first order ┌──────────┐
│ prospect ├───────────────▶│ new │
└──────────┘ └────┬─────┘
2nd order │ no 2nd order
┌───────────────┴────────┐
▼ ▼
┌──────────┐ ┌───────────┐
│ active │───────────▶│ lapsing │
└──────────┘ no order └─────┬─────┘
▲ for 1 cycle │ still nothing
│ ▼
│ reactivated ┌───────────┐
└─────────────────┤ lapsed │
└───────────┘
The arrows are the point. Reporting how many customers sit in each box is inventory; acting on who’s moving between boxes is a programme.
Defining the boundaries
The thresholds come from your own Time Between Orders distribution, not from convention:
median interval 120 days
active ordered within 120 days
lapsing 120–240 days ← one cycle missed, still reachable
lapsed 240+ days ← two cycles missed
Lapsing is the stage that matters. The customer hasn’t decided to leave — they’ve drifted past their usual point. Reaching them here costs a fraction of reaching them once they’ve lapsed, and the message can be a reminder rather than a bribe.
Most programmes have no lapsing stage at all: they have “active” and “we should win them back with 20% off”, and the gap between those is where the cheap intervention lived.
New is the other critical stage
A first-time customer hasn’t formed a habit and is at maximum risk. On the running model they’ve also lost money — £15 contribution against £22.50 CAC — so the second order is what makes them profitable at all — it’s the retail version of activation.
The new-customer window deserves disproportionate attention: post-purchase content, replenishment timing, and a second-order prompt targeted only at people who haven’t ordered again. Unlike blanket promotion there’s no cannibalisation, because they aren’t buying — Repeat Purchase Rate, Payback Period.
Why transitions beat states
A stage count is a lagging report. A transition is a trigger:
STATE "4,200 customers are lapsing" → a number in a deck
TRANSITION "312 customers moved into
lapsing this week" → a campaign audience,
today, while it's cheap
Every lifecycle campaign is really a transition trigger — Lifecycle Messaging.
Building it
- Recompute daily. A weekly refresh means up to seven days of delay on the one stage where timing matters most
- Store the transition, not just the current stage, so you can trigger on the change and measure whether the intervention worked
- Segment stages by value. A lapsing customer worth £200 warrants different effort from one worth £30 — combine with RFM Segmentation
- Vary thresholds by category where your ranges have genuinely different cycles
- Handle subscriptions separately. They have an explicit cancellation event and different mechanics — Subscription Metrics
Where it goes wrong
- Thresholds chosen for neatness. 30/60/90 days is a convention, not a finding, and it’s wrong for most categories
- Identity failures. Guest checkout means a returning customer appears as a new one, so they’re in the wrong stage and get the wrong message — Identity Stitching
- Stages nobody acts on. A classification with no campaign attached is a report, not a programme
- Discounting every transition. Reaching a lapsing customer with money off trains them to lapse deliberately — Discounting Strategy