The standard ecommerce win-back campaign looks like this: pull everyone who has not ordered in some round number of days, send them 20% off, call the resulting orders "recovered revenue."
It reliably produces orders. It also reliably discounts a large group of people who were going to buy anyway, contacts a lot of people who left for reasons a coupon cannot address, and reports a number that overstates its own impact. Here is a better version.
Start by admitting "lapsed" is not one group
The customers in your win-back list left for at least four distinct reasons, and they need four different messages:
- They drifted. No complaint, no incident — life happened and the habit broke. This is the largest group and the most winnable. They need a reminder and a reason to act now.
- They had a bad experience. A late delivery, a damaged item, a support interaction that went badly. Sending this group a discount without acknowledging what happened reads as tone-deaf, and it is the fastest way to convert a lapsed customer into a detractor.
- Their payment failed. Especially on subscriptions. They did not choose to leave at all — a card expired. This is not a win-back problem, it is a billing problem, and treating it as a win-back both wastes margin and delays the fix.
- They were never going to repeat. One-time gift purchasers, deal hunters who came in on a heavy promotion, people the product genuinely did not suit. Some of your list is not recoverable, and spending on it drags down every average you report.
Any system that cannot separate these is going to send one message to all four.
Timing: relative, not absolute
Fixed-day win-back triggers — 30, 60, 90 — are convenient and wrong, for the same reason a fixed churn definition is wrong. They ignore the customer.
The right trigger is relative to each customer's own established interval. Telltale targets win-back at customers whose churn risk has crossed the high threshold and who are meaningfully past their personal reorder gap — around 1.2 times their own interval and beyond. A three-week buyer gets contacted in week four. A six-month buyer is not touched until month seven.
Two practical rules follow:
Earlier is better, within reason. Reactivation gets harder the longer someone has been gone, because the habit decays and a competitor usually fills the gap. Most brands wait far too long. If you can only make one change to your win-back program, move the trigger earlier.
Respect a cooldown. Win-back should not re-fire on the same person for a couple of weeks minimum. Telltale uses a fourteen-day cooldown for this campaign type, on top of the global per-customer limits.
The offer: earn it, do not default to it
Discounting is the reflex, and it is where win-back programs quietly lose money. Every point of discount you give to a customer who would have returned anyway is pure margin transfer.
A more disciplined approach makes the offer conditional on evidence:
- Does this audience actually respond to discounts? You can measure this. Compute what share of each customer's past orders were placed on promotion. If the segment skews heavily discount-driven, an offer is justified — those customers have told you what moves them. If it does not, lead with the product.
- Is price even the blocker? For failed payments, no. For a shipping complaint, no. Diagnose before you discount.
- Would free shipping do the job? On lower-value baskets, shipping is often the real friction and it costs you less than a percentage discount on a larger order.
Telltale's win-back logic follows exactly this: a discount is applied when the segment's discount sensitivity supports it, sized accordingly, and withheld when it does not. Roughly speaking, the goal is that the offer is the smallest thing that changes the decision.
Channel: decide per customer
Email and SMS should not be a campaign-level choice. Some of your lapsed customers read email; some have not opened one in a year but reply to texts.
Route on engagement and consent, per person. In practice a win-back audience splits across both channels, with a residue of customers who are unreachable on either — and knowing the size of that unreachable group is itself useful, because it caps what the campaign can possibly deliver.
If you use SMS, quiet hours are not optional. Messages must land inside the recipient's local 8am–8pm window, computed in their timezone.
Copy that does not pretend
Three things separate win-back copy that works from the template everyone sends:
Acknowledge the gap without guilt. "We miss you" is fine. "You have not shopped with us in 137 days" is creepy and slightly accusatory.
Give a reason beyond the discount. A new product, a restock of what they used to buy, an improvement to the thing that went wrong. The offer should be the tiebreaker, not the entire argument.
Never invent personalisation. If you are not certain what they bought, do not reference it. A wrong product name destroys more trust than generic copy ever would.
Measure it honestly
This is the part almost everyone skips. If you send a win-back to 5,000 lapsed customers and 300 buy, you did not recover 300 customers. Some fraction were coming back regardless.
The only way to know your real contribution is to hold a randomly chosen slice of the eligible audience back and never contact them, then compare purchase rates between the two groups. The difference is your actual lift. It is almost always smaller than the attributed number — and it is the only number worth optimising against. We wrote about this in attribution is not impact.
A win-back that is not a win-back
The best win-back campaign is the one you did not need to send. Most of the customers in your lapsed list passed through a window, weeks or months earlier, when their reorder gap was stretching and a well-timed, discount-free nudge would have kept the habit intact. Catching people there is cheaper, more effective and less corrosive to margin than reactivating them later.
That is the case for reorder-window timing and for predictive risk scoring generally: win-back should be your safety net, not your retention strategy.
Telltale runs win-back alongside twelve other campaign types, each with its own targeting, timing and offer logic — and measures all of them against a held-back control group. Install from the Shopify App Store.