Segment definition, trigger timing, sequence design, suppression, and measurement. The unglamorous machinery that decides whether reactivation makes money, laid out step by step.
Win-back advice is nearly always about copy, which is the least consequential part of the campaign. The results are decided earlier, in the machinery: how you define lapsed, when the automation fires, how the sequence is ordered, and what happens to the people it fails to recover. Get the machinery right and mediocre copy still earns its keep. Get it wrong and the best copy in the world mails the wrong people at the wrong time, then keeps mailing them forever.
Here is the whole machine, decision by decision.
A fixed lapse definition treats a coffee subscriber and a mattress buyer identically, and gets both wrong. The workable approach is to measure your store’s typical time between orders and set the lapse threshold as a multiple of it, so a customer counts as lapsed when they have been quiet meaningfully longer than their cohort normally is. Omnisend’s published tiers, reminder messaging at 30-60 days inactive, value-led at 60-90, incentives at 90+, are sensible defaults for stores with monthly-ish cycles, and explicit calendar guidance you should stretch proportionally if your product is bought twice a year. The definition also needs a floor: a one-time buyer from a gift-heavy December cohort is a different reactivation bet than a four-time regular who went quiet, and your segment should know the difference.
The structural advantage of a flow over a quarterly “come back” blast is that the flow meets every customer at the same lapse depth. A blast hits someone at day 20, where the email is noise, and someone at day 300, where it is futile, with the same message on the same Tuesday. Automation fixes the alignment for free. Klaviyo’s 2026 benchmark across 183,000+ brands illustrates the general gap: flow emails convert at a 2.11% placed order rate against 0.16% for campaigns, a difference driven substantially by this kind of timing fit. Build win-back as an automation that enters each customer when they individually cross your lapse threshold, and exit them the moment they purchase.
The sequence walks Omnisend’s ladder in order: a plain reminder first, a what-changed message second, an incentive third if your margin supports one, and a final-chance email last. Two mechanical rules matter more than any subject line. Escalate only downward, meaning the discount never appears before the reminder has failed, because early coupons pay for returns the cheaper email would have earned. And walk the ladder exactly once per lapse: a customer who ignores the full sequence does not re-enter the flow next month for another round. The archetypes themselves, and what separates them from begging, are covered in our win-back examples piece.
This is the stage most stores skip, and the one that quietly pays for the whole campaign. Mailbox providers score senders on engagement, so a list padded with addresses that never open anything drags inbox placement down for every campaign and flow you send. The win-back sequence doubles as the fairest possible sunset test: the customer received a reminder, a reason, an offer, and a warning, and acted on none of them. After the final-chance email, non-responders move to a suppression segment and stop receiving regular sends. The reactivation campaign becomes, in effect, the gatekeeper of list hygiene, and that deliverability protection is worth more than the direct revenue for many stores.
Win-back economics are per-recipient economics. In Omnisend’s report data, the lapsed purchase automation averages a 33.00% open rate, a 1.96% click-to-send rate, and a 0.52% conversion rate, about $0.49 in revenue per email at a $94 average order value. Those numbers look small next to cart recovery, and they should, because the audience has already drifted. The correct comparisons are the cost of the flow, which after setup is nearly zero, and the acquisition cost of replacing a customer instead of recovering one. Track reactivation rate by lapse depth so you learn where your recoverable window actually ends, revenue per recipient by sequence position so you know which rung earns its place, and the size of your suppression flow-through so the hygiene benefit is visible on a dashboard rather than taken on faith. As a sanity baseline for engagement, Mailchimp’s benchmark page (data updated December 2023) puts average ecommerce campaign opens at 29.81%, so a lapsed segment opening at 33.00% tells you attention survives the lapse. The machine’s job is converting that attention before it fades for good.
Build the five stages in order, and revisit the definition in stage one once a year as your catalog and cycles change. Everything else runs itself, which is the point of a machine.
Enough to walk Omnisend's ladder once: a reminder, a value or product message, an incentive if your margins allow one, and a final-chance email. That is three to four sends. Beyond the final-chance message, further emails mostly generate spam complaints from people who have already declined several times.
Low single digits at best. Omnisend's report data puts the lapsed purchase automation at a 0.52% conversion rate with a 33.00% open rate. The flow makes sense because it runs automatically at near-zero cost, not because it converts like cart recovery.
Structurally yes, because a flow catches each customer at the same lapse depth, while a blast hits everyone at random depths, too early for some and hopeless for others. Klaviyo's 2026 benchmark shows flows in general converting at 2.11% placed order rate versus 0.16% for campaigns, which is the same logic playing out across all automation.
Suppress them from regular sending once they have ignored the full sequence. Mailbox providers judge senders on engagement, so continuing to mail dead addresses taxes inbox placement for your active subscribers. Suppression is not admitting defeat, it is paying for the deliverability of every other email you send.