In most stores, the customers who quietly stopped buying outnumber the prospects who never bought at all. They already trusted you once, they know the product, and reaching them costs nothing. That makes win-back the cheapest revenue in email marketing - and also the most dangerous send you own, because by definition you are mailing your coldest addresses. Done carelessly, a win-back blast damages the deliverability that carries everything else you send. This guide covers both halves: the flow that brings customers back, and the guardrails that keep it from hurting you.
Define lapsed from your own data, not a calendar
The classic mistake is an arbitrary threshold: "no order in 90 days" applied to every store alike. A coffee subscriber who has not ordered in 60 days is gone; a furniture customer who has not ordered in a year is normal. Look at the typical gap between repeat orders in your store and set the lapsed threshold at roughly one and a half to two and a half times that gap. If your customers usually reorder within six weeks, someone at week twelve is lapsed. If they reorder yearly, twelve weeks means nothing.
The flow: three emails, escalating slowly
Email 1: reconnect, no discount
Reference the relationship - what they bought, or what they liked - and show what is worth coming back for. Do not open with a discount. Leading with one trains customers to lapse on purpose and wait for the coupon, and it gives margin away to people who only needed a nudge. A meaningful share of lapsed customers come back on this email alone.
Email 2: show what changed (4 to 5 days later)
New arrivals since their last order, an improvement you shipped, a bestseller they have not seen. The silent assumption of a lapsed customer is that nothing changed; this email exists to break it. If your products get questions, one line of social proof belongs here.
Email 3: the incentive, with an expiry
Now the offer, if you want to make one at all. Keep it controlled: free shipping or a threshold offer ("save on orders over €75") protects margin better than a flat percentage. Make the code unique per recipient and give it a real expiry - a shared code leaks to coupon sites, and an offer that never ends creates no reason to act. Customers who ignore all three emails move toward a sunset segment, not back into your regular sends.
The deliverability guardrails
This is the part most win-back guides skip, and it is the part that can hurt you. Cold addresses are where spam complaints, dead mailboxes and spam traps live, and mailbox providers watch exactly the signals a careless win-back inflates. With Gmail and Yahoo enforcing hard complaint-rate ceilings, three rules are non-negotiable:
- Never blast every inactive contact at once. Send in slices, starting with the most recently active, and widen only when the previous slice has settled without complaint spikes.
- Exclude the truly dead. Someone who has not opened anything in a year does not belong in a win-back flow; they belong in a sunset policy. Win-back targets lapsed buyers, not ghosts.
- Stop on any signal of life. A purchase exits the customer immediately. An unsubscribe is honored instantly and permanently.
Measure what actually came back
Some lapsed customers would have returned anyway, and a flow that gives them a discount is losing money invisibly. If your volume allows it, hold back a small random slice of the lapsed segment from the flow and compare return rates between the two groups. The difference is what the flow actually earns. Even without a formal holdout, judge the flow on returned customers and the margin after discounts, not on opens.
Setting it up
In MailingPlatform, build the lapsed group as a segment on purchase behaviour - for example "has ordered before, but not in the last 120 days" - and start the flow when someone enters it. The three emails, waits and exits are drag-and-drop on the automation canvas, unique per-recipient discount codes are minted on your Shopify or WooCommerce store at send time, and a new order removes the customer from the flow automatically.
Win-back is the last line of defence. The cheaper fix is upstream: a post-purchase flow that makes the second order happen before anyone lapses at all.