The Klaviyo winback flow that doesn't send everyone one coupon
Most winbacks send every lapsed customer the same 30 percent off, best buyer and worst alike. The fix is a ladder of treatments, built from the buckets you already have.
One list, three treatments, with customer value setting each offer.
The one-coupon winback is a default nobody chose
Nobody sat down and decided your best lapsed customer deserves your worst offer; the template did.
The ladder: three tiers, three treatments.
Concierge first, coupon last
These are your highest-value customers, quiet the longest. That history has earned serious effort: the restock, the new version of what they bought, a we-noticed note. Lead with a coupon here and you teach your best people that lapsing pays, so hold the strongest incentive for the later touches. Open with a discount and the lesson they take is to wait for the next one.
Graduated value, on their clock
Mid-value customers drifting past their usual gap. Give them graduated value inside their reorder rhythm: the bundle, the soft offer, a reason to come back that stops short of a markdown. Time the touches to their reorder cycle rather than a fixed calendar. If the silence continues, escalate a step; if it stops, so do you.
One deep offer, then quiet
Low on every score. This is the one place a deep discount leads, and it runs once. If it does not land, stop sending. Mailbox providers weight engagement, so continuing to mail the unengaged tail suppresses inbox placement for everyone who still reads you. Stopping is what pays here: a single deep offer, then silence that protects the rest of the program.
Start at day 60, while the window is open
Operator guides such as Hightouch and Finsi (linked in Sources) put the effective winback window at 60 to 90 days after the last order. Most calendars wait for the six-month mark and open the conversation after the decision to leave has already hardened.
The same guides report that once a customer enters At Risk, the response window is roughly 30 to 45 days. Both windows vary by category and purchase cycle, so calibrate them on your own reorder data rather than an outside benchmark.
The 60/80 check: is it winback or a sale?
Pull your last winback campaign and check what share of reactivations redeemed the discount code. The operator rule of thumb: above roughly 60 percent, the program is training people to wait for the sale; around 80 percent, you have a discount habit wearing a winback costume. Both thresholds vary by category, so run the check on your own last campaign before trusting either.
Grade the winback on recovered margin. Recovered revenue reads bigger because it counts the sale and ignores what the coupon gave away to get it.
Two numbers, neither of them opens.
Reactivated margin
A winback that only converts at 40 percent off can lose money at full success: every save arrives pre-discounted. Count what the reactivation kept after the offer comes out. The gross number hides what the coupon gave away. A smaller number of undiscounted returns can be worth more than a bigger number of bought ones.
Revenue per recipient
Judge winback sends the way the flow checklist on the hub does: revenue per recipient rather than per open, and one change at a time. Opens flatter a subject line; per-recipient revenue grades the whole decision, who got the email and what it asked. Change one variable per cycle so you know what moved.
The ladder only stands on scored buckets. If your list is not scored, run the RFM teardown first; the three properties it writes back are what the tiers read. Once the ladder is live, the segment migration view is how you check the saves are real: watch whether recovered customers actually climb back, month over month.
The most common finding: best and worst, same coupon
When we read a winback program cold, the most common finding is the simplest one: the best lapsed customers are getting the same coupon as the worst. That read is part of every Blueprint, your flows, your buckets, your winback economics, on your data. Book the fit call if you want it run for you. Build from this page if you would rather not; it stays yours either way.
What operators ask
Should a winback ever lead with a discount?
For the bottom tier, yes, once: a deep offer is the only lever left, and it either lands or the sending stops. For every tier above it, lead with a reason to return, the restock, the new arrival, the note. Save the discount for the last lever in the sequence.
How deep should the winback discount go?
There is no universal number worth quoting. Depth follows the ladder's law: inverse to customer value, proportional to how long they have been gone, and your margin sets the floor under all of it. A depth that works for a consumable would sink a furniture brand. Set it from your own numbers and validate on your own data.
When do I stop emailing a lapsed customer?
After the bottom-tier attempt fails. Two or three unanswered touches past that point and you are burning sender reputation for no return: mailbox providers weight engagement, so mail to people who never open works against the people who do. Stopping there protects your reach to every customer who still opens.
Build the ladder yourself, or bring us your buckets
Everything on this page is enough to build the ladder in your Klaviyo this week. Or book the fit call and we read your winback economics as part of the Blueprint: $2,500, credited, built in your account. It stays yours either way.
More from this series: the lifecycle playbook the RFM teardown segment migration our Klaviyo Composer review
Sources
- Hightouch: winback timing, targeting, and the discount trap
- Finsi: win-back guide and the 60 to 90 day window
- Klaviyo: win-back campaign examples
- Concomitant: the RFM scoring teardown (the buckets this ladder runs on)
Benchmarks above are third-party operator figures. They vary by category, price point, and purchase cycle; validate every one against your own data.