Winback Campaigns
Date: 2026-08-16
Reaching customers who’ve stopped buying. The economics look excellent because you’re not paying acquisition costs — and they look excellent partly because a share of them were coming back anyway.
What it is
A winback campaign targets customers who have passed your churn threshold, aiming to produce another order.
The appeal is straightforward: no CAC, an existing relationship, and a known purchase history. Reaching a lapsed customer costs pennies where acquiring a new one costs £22.50.
Timing beats offer
The most useful finding in this area: response falls sharply with elapsed time, and it falls faster than most programmes assume.
months since last order relative response
1–3 (lapsing) high
4–6 moderate
7–12 low
12+ marginal
Which means the best winback campaign is one that fires before the customer is properly lapsed — in the lapsing stage, where the habit is interrupted rather than broken, and where a reminder still works without an incentive.
Most programmes wait too long and then compensate with a bigger discount. That’s the expensive order of operations — Lifecycle Stages, Time Between Orders.
The incrementality question
Central here, more than anywhere else in lifecycle work.
Some lapsed customers were going to return regardless — they’d drifted, not decided. Discounting them costs contribution for nothing:
1,000 lapsed customers receive 20% off
180 order
of those 180, how many would have returned anyway?
without a holdout, you cannot know
and you have discounted all 180
Always hold out a random slice. Compare order rate in the treated and untreated groups; the difference is what the campaign caused. This one test changes the economics of the whole programme, and it’s a one-line audience exclusion — Incrementality Testing, Holdout Groups.
Escalating properly
Don’t lead with money. A sequence that starts cheap and escalates only for non-responders discounts the fewest people:
1 "we've missed you" + what's new no incentive
2 relevant product, based on history no incentive
3 modest incentive for non-responders only
4 stop
Stopping matters. Repeatedly messaging non-responders damages deliverability and burns the list — and a customer who ignored four winbacks is telling you something.
Segment before spending
Not all lapsed customers are worth the same effort:
- High past value, recently lapsed — the priority. Proven buyers, recoverable — RFM Segmentation
- Low value, long lapsed — usually not worth an incentive at all
- Lapsed after one order — a different problem. They never formed a habit, and it’s closer to a failed onboarding than a winback — Repeat Purchase Rate
- Lapsed after a bad experience — a return, a delivery failure, a complaint. A discount is the wrong response; acknowledgement is
That last group is worth identifying from support and returns data, because messaging them a generic offer reads as tone-deaf.
Measuring it
- Incremental orders, from the holdout — not attributed orders
- Contribution after the discount, not revenue — Discount Impact on Margin
- Second-order rate among winbacks. A reactivated customer who buys once and lapses again is worth far less than the campaign report implies
- Unsubscribe rate, as a guardrail on how hard you’re pushing a disengaged audience
In plain terms: winback looks like free money because the cost side is small. The hidden cost is discounting people who needed no discount, and the only way to see it is to withhold the campaign from some of them.