A lapsed customer already knows you, already trusted you once, and already has your product in their life somewhere. Winning them back costs a fraction of finding a new one. Most win-back campaigns fail because they sound desperate — 'We miss you!' with a discount — rather than useful. The customer left for a reason; the campaign should address it.

Define lapsed by the customer's own rhythm

Not 'no purchase in ninety days' for everyone. Calculate each customer's typical interval between orders or logins and flag them when they pass twice that interval. A monthly buyer is lapsed after two months; a twice-a-year buyer is not. This alone removes most false positives and most wasted messages.

Three stages, three different messages

First: something new — a product, a feature, a change — that they have not seen, with no ask beyond looking. Second: a direct question — 'what would have made you stay?' — from a person, with replies read by a person. Third: an offer, if your margins allow, tied to something they bought or used before. Two weeks between stages, on the channel they responded to historically.

Use what you know

Their last product, their last support ticket, their stated reason if they cancelled. A message that references the actual last order beats any generic one. If they left after a bad experience, the first message should acknowledge it and say what changed.

Know when to stop

Customers who ignore all three stages go to a quarterly, low-frequency list at most, or are suppressed entirely. Continuing to message the uninterested damages deliverability and brand for the customers who are still listening. Measure win-back as returning revenue over the following quarter against a holdout.