Win-Back Campaign: How to Bring Lapsed Customers Back

A win-back campaign targets customers who have already bought and have stopped. That makes it different from the two campaigns it gets confused with. Cart recovery reaches someone who is in the middle of buying right now. A newsletter reaches everyone who subscribed, whether they ordered last week or never. A win-back campaign sits between them: the relationship exists, but it has gone cold.

The economics are the reason it is worth the effort. Harvard Business Review put the range plainly in 2014: acquiring a new customer is anywhere from five to 25 times more expensive than retaining an existing one (Amy Gallo, The Value of Keeping the Right Customers, Harvard Business Review, 2014, retrieved 2026-09-24). Treat that as a wide published band rather than a figure for your shop — but the direction is not in doubt.

One honest limit before you start: a win-back campaign repairs a lapse, it does not prevent one. If most of your customers never place a second order, the fix sits earlier in the journey and no amount of reactivation messaging will cover for it.

Ninety days is the default everyone copies, and for most shops it is wrong. A coffee subscription lapses in weeks. Skincare runs on a two-to-three-month refill cycle. Furniture, luggage or a winter coat may have no meaningful lapse point at all inside a year. The window has to come out of your own order data.

A workable method, in three steps:

1. Take every customer with at least two orders and measure the days between their first and second order.
2. Use the median, not the average. A handful of customers who reordered after two years will drag an average somewhere useless.
3. Double it. If the median gap is 45 days, a customer who has not ordered in 90 days is genuinely off-cycle; at 30 days they are simply between orders and the message will read as nagging.

If you have never looked at these numbers, start with the two that sit underneath every retention decision — they are covered in our guide to ecommerce customer retention.

Not every lapsed customer is worth reaching, and sending to all of them is how a win-back campaign turns into an unsubscribe event. Rank by what they did before they went quiet:

• Two or more orders. A repeat buyer has demonstrated the habit once; a one-time buyer has demonstrated nothing yet.
• Average order value at or above your shop median. The same message earns more from this half of the list.
• Reachable and opted in. Consent decides who you may message at all, and it is independent of how valuable the contact is.
• No open complaint or recent refund. Those customers need a reply from a person, not a campaign.

If you already score customers by recency, frequency and monetary value, the win-back audience is the group that used to sit in your strong segments and has slipped out of them — the method is in our RFM analysis guide for Shopify. The often-quoted benchmark for why this group is worth the effort comes from Paul W. Farris and colleagues in Marketing Metrics (Pearson, 2nd edition, 2010): the probability of selling to an existing customer is put at 60–70%, against 5–20% for a new prospect. Again, a published band, not a promise about your catalogue.

One thing to be clear about first, because it changes the build: Keaz has no "this customer has lapsed" trigger. Flows start from events — a purchase, an abandoned checkout, a webhook, a chat-in — and time passing is not an event. A win-back campaign in Keaz is therefore a segment plus a send, which you repeat on a rhythm, rather than an automation that fires by itself.

1. Open Audience → Segments and create a new segment. Name it for the window, for example "Lapsed 90+ days, 2+ orders".
2. In Included Groups, add a Last purchase condition set before your lapse date, and an Order count condition of two or more. Conditions inside one group are combined with AND, so both must hold.
3. Add Average order value or Total revenue if you want to reach the stronger half of the group first, or RFM group if you already work with those scores.
4. In Excluded Groups, take out anyone who received a newsletter in the last few days. Excluded groups subtract from the included ones, which saves you maintaining a second, inverted segment.
5. Read the panel on the right before you save. Audience coverage tells you how many contacts you are about to message, and average time since last purchase tells you whether the window you set actually matches the people it caught.

The full condition list and the AND/OR logic are documented in the Segment Builder help article, and the wider picture of what the platform does with a segment once it exists is on the Keaz features page.

Send one message first. The three-part sequence that every template pack recommends is a good idea only once you know the single message works, and on a segment this small a sequence mostly produces unsubscribes from people who were never coming back.

• Name the gap without apologising for it. "It has been a while" is enough; a paragraph of regret is not.
• Lead with a reason that is not a discount: a restock of what they actually bought, a new size or shade, a change they asked about.
• Reference their own order history where you can. A win-back that could have been sent to anyone reads like a list purchase.
• Keep the next step to one tap — one link, one button, one product.
• Hold the discount for a second message, sent only to the part of the segment that did not respond. Opening with money trains the rest of your list to wait for it.

Which channel carries it is a secondary question. WhatsApp complements email here rather than replacing it: a short, personal message suits a lapsed-customer nudge, while email carries the longer catalogue case. How a send picks up a segment is described in the newsletter sending help article. Whichever channel you use, the contact has to have opted in to it — consent is per channel, not per customer.

Some lapsed customers come back on their own. Without a control group you will credit the campaign for all of them, decide win-backs are wildly profitable, and scale something you have not actually measured.

Two numbers are enough. The first is the reactivation rate: the share of the segment that placed an order inside a fixed window — 30 days is a reasonable default — minus the same share among the roughly 10% you deliberately did not message. The difference is what the campaign did. The second is revenue per recipient, which keeps you honest when a high reactivation rate turns out to be made of small orders bought with a discount.

If you want an estimate of what a reactivated customer is worth before you run anything, the Keaz Forecast models it from your own shop data rather than from a benchmark table.

• Borrowing the 90-day window. It is someone else's repeat cycle. Measure your own median gap and double it.
• Sending to every lapsed contact at once. The weakest half of the segment costs you unsubscribes and buys almost nothing.
• Opening with a discount. It converts the customers who would have returned anyway, at a lower margin, and teaches everyone else to wait.
• Skipping the holdout. Without it every returning customer looks like a campaign win.
• Expecting an automation to do it. There is no lapse trigger — the segment has to be rebuilt and sent on a schedule you keep.
• Running win-backs instead of fixing the second order. If most customers never reorder, the problem is upstream of this campaign.