Somebody is going to present a win-back slide in a quarterly review this month. Reactivation email, 4x on spend, best number on the deck.
It will get a nod and no questions, because nobody argues with the good number.
That slide is the one I’d argue with. Not because win-back doesn’t work, it does. Because the figure on it is almost certainly two or three times larger than the real one, and the gap is where the budget decisions get made.
So, the short answer. Win-back works, considerably less well than your dashboard says, and it works best when you stop treating “lapsed” as one audience. Segment by why somebody left rather than how long ago. And lead with a discount only if you’ve decided you don’t want to know why they came back.
The benchmark spread that should end the argument
Go and read the 2026 win-back benchmarks. I did, properly, and they are a mess.
One set of published figures puts automated win-back open rates at 42.5% and conversion at 10.3%. Another puts open rates at 25% to 32% and conversion at 1% to 3%, with top performers reaching 5%. Both are presented as industry data for the same flow in the same year.
A tenfold spread on conversion. Those are not variations around a number, that’s two different realities.
There is no win-back benchmark. There is only your holdout, and most teams don’t have one.
That’s the part I’d push on, and it’s unpopular because it turns a reporting line into an experiment.
Why the number on the slide is inflated
Mechanically, not because anybody is lying.
Every retention platform dashboard, whichever one you use, assigns revenue by last touch within an attribution window. Somebody lapsed opens the email, clicks, buys. Email gets the sale. That’s correct bookkeeping and a terrible causal claim, because the dashboard cannot see the counterfactual where the email never arrived.
Now think about who sits in a lapsed segment. People who bought once and drifted, yes. Also people on a natural repurchase cycle who were going to come back in week eleven anyway, and people who had already decided to reorder and were waiting for a reason to do it today.
That third group is why the number looks so good. You sent a 15% code to somebody who was already in the cart in their head, and booked the revenue as recovered.
The size of that distortion is measurable, and somebody has measured it. One published holdout analysis of a print catalog program found 14% incremental revenue lift among recipients against the 40% contribution the vendor’s own attribution reporting claimed. Different channel, identical mechanism.
I’d expect win-back email to sit somewhere similar. Real, worth running, roughly a third of what the report says.
The argument: this needs an owner, and their first job isn’t a campaign
Here’s the position, and it’s the one I’d defend in the room.
Win-back is usually nobody’s job. It’s a thing the email person does in a slow week, which means it gets a flow, a 10% code and no scrutiny. Give it a named owner with a target, and the first deliverable from that person should not be a sequence.
It should be a permanent 10% holdout inside every lapsed segment. Never mailed, tracked forever.
Within one quarter you’ll know your actual natural return rate, which is the only number that makes the win-back figure mean anything. I’ve never seen a team regret building this and I’ve seen several discover their flagship reactivation flow was beating the holdout by a few points rather than carrying the retention story.
That’s not a reason to stop. It’s a reason to stop over-investing in the easy half and under-investing in the half that’s actually broken.
Segment by why they left, not when
Recency segmentation is the default because recency is the field you already have. It’s also close to useless, because four completely different situations sit inside “no purchase in 180 days”.
The card failed. Payment processors and subscription billing vendors put involuntary churn somewhere between a fifth and two fifths of subscription cancellations, and some put failed payments at close to half. These people didn’t leave. Their Visa expired. Sending them a “we miss you” email is slightly insulting and definitely the wrong tool. The fix is dunning logic, card account updater, retry timing and a readable in-app prompt. Fix this before you write a single word of copy, because it is the cheapest recovery in the building and it isn’t marketing’s budget.
They finished. They bought the mattress, the water filter, the eighteen month thing. They’re not lapsed, they’re mid-cycle. What they need is replenishment timing, which I’ve written about as the post purchase flow where margin actually lives.
They went somewhere else on price or product. This is the group everybody counts in the segment and nobody wants to talk about. You’re not winning them back, you’re asking them to switch a second time, and nobody switches twice for 10% off. The only opener that works here is something that materially changed. New pricing tier, the missing integration shipped, next day delivery now covers their state. If nothing changed, a win-back campaign is a faster way to be told no.
Something went wrong and nobody fixed it. Damaged delivery, a support ticket that died, a bad first experience. These are the highest intent and the most dangerous. They’ll tell you what happened if you ask, and they will react badly to a promo.
Four reasons, four different programs. One flow for all four is why the aggregate numbers are mediocre.
The discount is the wrong opener
Three reasons, and the third is the one people miss.
It teaches lapsing. If the reliable way to get 20% off is to stop buying for five months, a portion of your base learns the trick. Not consciously, but behavior follows incentives.
It prices your worst margin segment worst. Lapsed buyers have the lowest expected lifetime value in your file, and you’re handing them your deepest discount. Run that math before you set the code.
And it destroys the information. A customer who returns for a discount tells you only that they like discounts. A customer who returns because you told them the thing that changed tells you which change mattered, which is the single most useful signal retention work produces.
Open with what changed. Hold the discount for email three, if at all.
What I’d say in that quarterly review
I’d ask for the holdout number before I looked at the slide. If there isn’t one, the slide is a measure of how many of your customers come back on their own, printed next to a spend figure.
That’s not nothing. It’s just not 4x, and knowing the difference is worth more than the campaign.
If you want to think through where your own lapsed file actually sits, get in touch by email, WhatsApp, phone or LinkedIn for a quick chat. My notes on how I approach the acquisition side cover the rest.