What is a winback email?

A winback email is an automated message that fires when a customer has gone longer than usual without ordering. It is the last of the standard flows and the one most stores never build.

It is not an abandoned cart email, which goes to someone who is in the middle of buying right now. It is not a post-purchase email, which goes to someone who has just bought. It is not a replenishment reminder either, though the two are often confused. A replenishment message says your product is probably running low, and it goes out while the relationship is still normal. A winback says you have not bought in a while, and it goes out after that point has passed.

The distinction matters because the messages are completely different in tone and content. A reminder can assume the customer still intends to buy. A winback cannot. Something changed: they bought elsewhere, the product did not suit them, the reason to buy went away, or they simply forgot the store exists. A good winback acknowledges the gap rather than pretending nothing happened.

The audience is small compared with your main list, and the revenue per recipient is usually high, because these are people who have already proved they will buy from you once.

Set the trigger at a point comfortably past your own typical gap between orders, worked out from your order history rather than from a default someone suggested.

Start by finding the usual gap between a customer's orders for each product group. If a product is reordered around every two months, a customer who has not ordered in five is genuinely lapsed and the winback should have fired. If your product is bought once a year, sending a winback at ninety days will annoy people who are behaving perfectly normally.

Run the flow off the date of the last order, per customer, and set separate versions per product group where the cycles differ a lot. A store selling both a fast consumable and a durable item needs two, and running one timing across both means half the sends land at the wrong moment.

Also exclude the people who should not be in it. Anyone who has ordered since the trigger was set, anyone currently in an active subscription, anyone who already received a winback in the last few months, and anyone who has asked not to be emailed. Lapsed lists get stale fast, and repeatedly mailing addresses that never engage is the quickest way to damage how your sends are delivered.

Lead with a reason to come back that is not money, and hold the discount for the final message if you use one at all.

The reasons that work are specific. What has changed since they bought: new products, a reformulated version, a colour or size they asked about, a shipping option that is faster than the one they used. What they are missing: the accessory, the refill, the next size. Or a straightforward acknowledgement that they have not been back, with a short question about why, which sometimes produces the single most useful feedback a store gets all year.

Show the product they actually bought. A lapsed customer who sees the exact item they own recognises themselves in the email, and a generic bestsellers grid does not do that.

If you use a discount, use it once, at the end, with an expiry, and keep it modest. A large code sent to everyone who lapses teaches customers that going quiet is rewarded, and the people most likely to notice that pattern are the ones who buy from you most often. Some stores find that free shipping or a free sample outperforms a percentage off, because it costs less margin and feels like a gesture rather than a fire sale.

Two or three messages, spaced over a few weeks, is enough. Beyond that the returns fall away sharply and the cost to your deliverability starts to matter.

A workable shape: the first message is the reason to come back with no offer. The second, a week or two later, adds proof, which might be a review of the product they own, a use case, or what other customers pair it with. The third, if you use one, carries the offer and an expiry.

Then stop, and mean it. If someone has not opened three messages, they are not going to open a fourth, and continuing to send to them drags down how your whole list is treated by inbox providers. Move non-openers to a suppressed segment, and try again in six months with something genuinely new rather than the same sequence.

One more thing worth building at the same time. Add a preference option so people can choose fewer emails instead of none. A customer who drops to a monthly send is worth far more than one who unsubscribes, and the winback message is a natural place to offer that choice.

Hold a portion of the lapsed segment back from the flow and compare, because a winback will always take credit for orders that were coming anyway.

This is the flow where that matters most. Some proportion of lapsed customers return on their own, and if you send everyone the email, every one of those returns is attributed to it. A holdout of a tenth of the segment, left alone for a few months, tells you the difference between orders the flow caused and orders it merely witnessed.

Watch three numbers over that period. Orders from the treated group against the holdout, the margin on those orders after any discount, and whether the returning customers go on to buy a third time. That last one decides whether the winback is producing customers or just producing one-off transactions bought with a code.

If the flow shows no lift over the holdout, the usual causes are timing that is too early, an offer that was not the barrier, or a lapsed list that had already been mailed too often to respond. Change one of those at a time rather than all three, and keep the holdout in place while you do.