Customer Win Back Email Examples That Work

A win back email targets customers who bought before and have gone quiet past your normal repurchase window. The campaign works when the lapse window is set from your own repeat-purchase data rather than a round number, and when the offer escalates across three sends instead of leading with the deepest discount you are willing to give.

Set the lapse window from your own data, not 90 days

Ninety days is the default in most ESP templates and it is wrong for most catalogues. A coffee brand with a 30-day consumption cycle has a customer who is genuinely lapsed at day 60. A furniture brand does not have a lapsed customer at day 90, or at day 900. Using the same window for both produces a campaign that nags one audience and misses the other entirely.

The number you need is the median days between first and second order, taken from your own order history. Set the first win back send at roughly 1.5 times that median, and the final send at 3 times. If your median gap between orders is 45 days, the sequence runs at day 68, day 90 and day 135. Recalculate it once a quarter, and split it by category if your catalogue spans consumables and durables, because a single blended median will be wrong for both.

One more split is worth the effort. One-time buyers and repeat buyers respond to different messages. Someone who ordered once and vanished may never have found the product that fit. Someone who ordered six times and stopped has a specific reason, and a discount rarely addresses it. Treat them as two campaigns.

The three-email ladder, with the offer last

Email one carries no offer. It asks whether anything went wrong and restates what the brand does. This email exists partly to recover customers cheaply and partly to clean the segment, because the people who open and click here are worth the offer in email two and the people who do not are candidates for suppression.

Email two carries a modest, specific incentive tied to the product they actually bought. Free shipping on a reorder, or a fixed amount off the category they browsed, beats a sitewide percentage because it reads as targeted rather than desperate. Keep the reorder path to one click if your platform supports it.

Email three is the last contact before suppression and it should say so plainly. A clear “this is the last email in this series” line outperforms a fourth and fifth attempt, and it protects deliverability by ending the sequence rather than letting it grind on against an unresponsive segment. If your deepest offer is going anywhere, it goes here.

Subject lines across the ladder should stay literal. “Did something go wrong?”, “Your usual, with shipping on us”, and “Last one from us” test well because none of them pretend to be a newsletter. Avoid guilt framing and avoid fake urgency, both of which raise complaint rates on a segment that is already disengaged.

What to measure, and when to stop

A win back campaign is judged on incremental orders, not open rate. The honest measure is a holdout: withhold the sequence from 10 percent of each lapsed cohort and compare order rates over the following 60 days. Without a holdout you are counting customers who would have returned anyway and calling it recovery.

Sixty days is the right window because a lapsed customer who returns does not always return immediately, and a 14-day read will undercount the campaign while a 180-day read will credit it with orders it had nothing to do with.

Watch complaint rate and unsubscribe rate per send, and cut any email whose complaint rate runs above your provider’s warning threshold. A win back sequence that damages your sending reputation costs more than the orders it recovers, because the cost lands on every campaign you send afterwards.

When we scope reactivation work we quote [BMO NUMBER: insert the incremental order-rate lift you measured against a holdout, with the cohort size and window] as the realistic ceiling, because it is measured rather than modelled. Our email marketing agency  [link to: /services/email-marketing] team builds the cohort logic, the holdout and the suppression rules together, and pairs the sequence with a sunset flow  [link to: /blog/klaviyo-sunset-flow] so the people who do not come back stop costing you deliverability.