The win-back playbook
Winning back a lapsed customer costs a fraction of acquiring a new one — but only if "lapsed" is defined by your data, the message gives a reason, and you know when to stop.
Most win-back programmes fire a "we miss you" discount at day ninety and call it strategy. The version that works starts earlier and thinks harder: derive when customers actually lapse per category, understand why before writing anything, match the re-engagement to the reason, and enforce the sunset that protects deliverability for everyone still listening. This playbook runs that sequence.
Stage 1: Define lapsed from your data
The lapse point is where an individual’s gap since last purchase materially exceeds the repurchase distribution for what they buy — weeks for consumables, quarters for durables, and different again per first-order value. Compute it per category, then check why: if lapse correlates with a product defect, a delivery failure, or a support incident, that cohort needs a fix acknowledged, not a coupon.
Repurchase interval distributions per category — not one blended number
Individual lapse flags relative to their own category’s curve
Lapse-reason correlation: returns, support contacts, delivery failures
Recoverability scoring: recency of lapse, engagement pulse, purchase depth
Stage 2: Re-engage with reasons, not sentiment
"We miss you" gives no reason to act. What does: what changed since they left — new arrivals in their category, restocks of what they bought, improvements to what disappointed. Sequence from reason-led to offer-led: the first touches sell what is new; the incentive enters late and only for segments where holdout testing shows it changes the outcome rather than subsidising it.
Stage 3: Sunset with discipline
A contact who ignores the full sequence is not asleep, they are gone — and continuing to mail them degrades deliverability for the customers who remain. Move exhausted contacts to a suppressed state with a rare, low-frequency re-permission touch. Shrinking the active list this way routinely improves the programme’s aggregate metrics, which is the point: sending less to the wrong people is a win.
How it works
Derive windows and reasons
Per-category lapse points, reason correlation, and recoverability scores from your own data.
Sequence reason-led re-engagement
What-changed messaging first, incentives late and holdout-proven, per segment.
Enforce the sunset
Exhausted contacts suppressed, deliverability protected, re-permission rare and deliberate.
What you get
Lapse defined by category-level evidence, not a round number
Product-caused churn routed to fixes instead of coupons
Incentive spend confined to where it changes outcomes
A cleaner active list that raises every send’s effectiveness
Handled by these agents
Works with
Frequently asked questions
When exactly is a customer lapsed?
When their gap since last purchase clearly exceeds the repurchase curve for their category and value band. For a monthly consumable that might be week ten; for furniture, year two. Deriving it per category is Stage 1 for a reason.
Do win-back discounts train bad behaviour?
They can — customers who learn that leaving earns a coupon will leave. That is why the sequence leads with reasons and the incentive is late, targeted, and holdout-tested rather than the opening move.
What win-back rate is realistic?
Single-digit to low-teens conversion of genuinely lapsed contacts is solid; claims far above that usually mean the "lapsed" definition was catching customers who were coming back anyway.
Which agents run this?
Analyst derives windows, reasons, and recoverability; Retention runs the sequences and the sunset. The holdout measurement lives in shared memory, so the incentive debate gets settled by data once instead of argued quarterly.
Keep reading
Win-back campaigns
Re-engage lapsed customers using your real churn window instead of a round number.
Customer segmentation
Segments built from purchase behaviour, not from generic engagement windows.
Design a loyalty programme
A loyalty programme designed from repeat-purchase economics — not points for their own sake.
Localise a store
Enter a new market with real localisation: structure, vocabulary, and permanent sync.
Put the team to work on your store
Connect your storefront, analytics, and email stack, set a goal, and let the agents run the work end to end. Start on the free plan with your own model key.