How a winback campaign works
The starting point is a definition of lapsed, and it has to come from your own sales records rather than from a default in the software. Look at how long your customers normally leave between orders. Somebody who has gone well past that gap without buying has lapsed; the same gap for a different business might be perfectly normal.
Once the window is set, the sequence triggers automatically when a customer crosses it. A short run of messages follows: an acknowledgement that it has been a while, a reason to look again — what has changed, what is new, what they bought before — and finally a direct question about whether they want to keep hearing from you at all.
The last step is the one people skip. Anyone who does not respond to the sequence should be moved to a suppression list and stop receiving campaigns, because a contact who ignores everything is now doing harm rather than nothing.
Why it matters
A lapsed customer already knows the business, has trusted it with money once and needed no explaining to. Reaching them costs a fraction of finding somebody new, and a portion of them have simply drifted rather than defected.
The sequence also protects the rest of your email. Mailbox providers watch how recipients behave, so a list carrying a large tail of people who never open anything drags down sender reputation and pushes ordinary campaigns towards the spam folder. A winback gives that tail one last fair chance and then clears it, which usually improves results for everybody who stays.
Where it goes wrong
Using one lapse window for the whole catalogue is the first error. A customer who buys coffee is overdue after a few weeks; a customer who bought a mattress is not overdue for years. Applying the same rule to both means chasing people who are not lapsed and ignoring people who are.
Opening with the deepest discount you have is the second. It tells loyal customers that going quiet is rewarded, and it drags back the buyers least likely to return at full price. The third is a winback that never ends — a sequence that keeps running month after month against a group that has clearly stopped listening, which is precisely the behaviour the campaign was meant to cure.
How to get it right
Work out the normal gap between orders from your own data first, and set the window from that. Where the range varies by product, split the campaign so a consumable and a durable are treated differently. RFM analysis is a straightforward way to find which lapsed customers were valuable enough to be worth the effort.
Keep the sequence short and make it easy to answer honestly. Ask plainly whether they still want your emails, offer a lower frequency as well as an unsubscribe, and treat a clean unsubscribe as a good outcome rather than a loss. Then suppress the silent remainder and leave them alone.