A win-back email is a message to a customer who has stopped buying, usually sent after three to six months of inactivity and usually carrying an offer. Six examples below cover the patterns that work. The sharper point is that “lapsed” is a lagging indicator: by the time the segment fires, the signals that predicted the lapse are months old.
We build an agent that acts on those earlier signals, so read this as informed but interested. The examples are usable on any platform, Klaviyo’s timing advice is quoted in its own words, and the argument about lag is ours.
Klaviyo defines the audience as “existing but dormant subscribers”: “people on your email list who previously visited your website, clicked on an email, or made a purchase, but haven’t bought from your store or opened your emails in a long time.” On timing, it says “you can generally consider a subscriber or customer dormant after 3-6 months of inactivity,” and suggests finding “the timeframe where 75-85% of all customers would repurchase.” It recommends “2-5 emails” in the series, and on incentives it is candid: “if a customer hasn’t engaged with your brand in 3-6 months, it will likely take a discount to re-engage them.”
Shopify ships win-back as one of its “ready-to-use automations like welcome, winback, and upsell emails.” In both, the mechanism is the same: a rule waits for a period of inactivity, then a sequence starts.
Subject line: “Still here when you are.” No offer, one product the customer bought or browsed, one line acknowledging the gap without apologizing for it. Fits a considered-purchase brand where a discount would cheapen the relationship. Risk: it reads as filler if the product shown is generic.
Subject line: “About time for more [product]?” Sent when the typical use-up window for what the customer bought has passed. Fits consumables: coffee, skincare, supplements, pet food. This is the only win-back that is really a post-purchase email arriving on schedule, and it is often the strongest one a store has; see post-purchase email.
Subject line: “New since your last order.” Three new products or one meaningful change, chosen from the categories the customer showed interest in. Fits brands with real catalog turnover. Risk: showing newness the customer does not care about proves the brand has stopped paying attention.
Subject line: “A little something to come back.” A modest incentive, escalating only if ignored, across the two to five emails Klaviyo suggests. Fits brands whose margin can carry it. Risk: it is the pattern most stores default to, so it teaches the customer that lapsing earns a coupon.
Subject line: “Did we get something wrong?” A one-click reason: price, fit, no longer needed, went elsewhere. Fits any brand, because the answers are worth more than the reactivation rate. Risk: asking and then doing nothing with the answer.
Subject line: “Should we stop emailing you?” The last message before the address is suppressed, honest about what happens next. Fits every store, because deliverability depends on it: a list that keeps mailing the unresponsive drags down the sends that matter. This one is not really trying to win anyone back. It is protecting the inbox placement of the emails that can.
Mostly, before they leave. A lapsed customer is a customer who was, at some point, still buying and starting to drift, and the drift showed: sessions got shorter, an email went unopened, a category was browsed without a purchase, a delivery ran late, a support ticket was opened. Each of those was a moment when one right message might have held them. The win-back email arrives after all of them, addressed to a segment defined by absence, and it opens with a discount because, as Klaviyo says, by then it will likely take one.
So the useful order is: hold the customer while there is still a signal to act on; when a lapse happens anyway, send the check-in or the replenishment reminder before the offer; ask why; and sunset cleanly. The examples above are the second half of that sequence. The first half is not a flow at all.
A win-back segment is defined by the absence of an event over a window: no order in 90 days, no open in 120. Absence can only be measured after the window closes, which means the earliest possible win-back email is sent exactly one window late, and the brand’s own timing rule guarantees it. Make the window shorter and the segment fills with customers who were simply between purchases; make it longer and the customers in it are gone.
The signals that precede a lapse are events, not absences, and they arrive one shopper at a time, in real time. A rule cannot act on them well, because a rule is written for a segment and these signals are individual: this person’s session was short, this person’s delivery was late. Acting on them means deciding, per shopper, whether a message is warranted right now and whether it needs an offer at all. That is a decision problem, not a scheduling problem, and it is the reason we think win-back is the lifecycle email that most clearly shows the ceiling of the flow paradigm. A related timing point covers the hour after an abandoned checkout, made in abandoned cart vs abandoned checkout; the broader retention picture is in Shopify retention strategy.
Klaviyo’s advice is honest: after months of silence, a discount is probably required. The job, in our view, is to find the smallest one that works. The cost of that advice is invisible in a flow, because a flow cannot decide that this particular shopper would have come back for a reminder. An agent with a margin floor can. It can send the check-in to one customer, the replenishment nudge to another, and nothing at all to a third who is simply between orders, and it can hold the offer for the shopper whose signals say it is needed. Across our customers the result is 80% less spam and lower send costs, and the customer example that fits this post best is POV Beauty: 2X fewer emails, same revenue.
Relvino runs Observe → Decide → Act on the store’s owned channels. It watches live shopper signals, and for each shopper decides in under 80 milliseconds whether a message is warranted, and if so which offer, on which channel, at what moment, then executes across email, SMS and on-site. There is no win-back flow because there is no window: drift is a signal the agent recognizes for one shopper, not a segment a person defined. Guardrails are set once (margin floors, channels, brand voice), and inside them 100% of flows run without a human in the loop.
The priors come from a Large Retail Model trained on 7M+ data points across 10K+ retailers and 1.78M shoppers, so the agent knows what drift looks like in a category before it has seen the store. Brands replacing an incumbent see 2–6× ROI in 30 days and up to 10× revenue uplift year over year. Terra Kaffe saw 2X the revenue of standard flows. Migration from Klaviyo takes about 30 minutes, the proof is a 14-day pilot beside the incumbent, and the wider lifecycle argument is in autonomous email marketing. Pricing is on the pricing page.
For win-back, a good example names something true about this customer (what they bought, what they browsed, how long it has been) rather than announcing a discount to a segment, arrives before the customer has forgotten the brand, and makes the next step one click. Of the six patterns in this article, the replenishment reminder and the no-offer check-in meet that bar most often; the stepped discount meets it least, and it is the one most stores send by default.
Mostly before they leave. The signals that precede a lapse (shorter sessions, an unopened email, browsing without buying, a late delivery, a support ticket) arrive one shopper at a time, and one right message in that window can hold a customer no later offer will. When a lapse happens anyway: send a check-in or replenishment reminder before any discount, ask a one-click question about why, escalate an offer only if ignored, and sunset cleanly to protect deliverability.
For a Shopify store, the ten lifecycle emails most programs run are welcome, browse abandonment, abandoned cart, abandoned checkout, order confirmation, shipping and delivery, review request, replenishment, win-back and sunset. The first nine are behavior-triggered and earn revenue on days nobody sends a campaign. Our guides cover welcome email examples, post-purchase email, and abandoned cart vs abandoned checkout; this article covers win-back and sunset.
Klaviyo's own guidance is that a customer can generally be considered dormant after 3 to 6 months of inactivity, adjusted to the timeframe in which most of a store's customers would repurchase, with 2 to 5 emails in the series. The honest caveat is that any window-based rule fires one window late by definition. Acting earlier means responding to individual signals of drift rather than to the absence of an order, which is a per-shopper decision rather than a scheduled flow.
About 30 minutes for the technical cutover: connect the Shopify store, point the sending domain, connect SMS, and shopper data ingests automatically. Win-back flows are not migrated because Relvino has no windows or sequences; drift is a signal the agent recognizes per shopper and answers inside guardrails you set once, with any offer decided per shopper inside a margin floor. Revenue is proven in a 14-day pilot run beside the incumbent, and pricing is on the pricing page.