Most winback sequences fail by leading with discounts. The customer who lapsed isn't waiting for a coupon — they made an active choice to walk away. Discounting first reads as 'we don't know you and we're trying to bribe you back.'
The patterns that actually work treat winback as a conversation, not a sales pitch.
A typical lapsed-customer winback in most ESPs:
Conversion on this sequence is mediocre because the framing is wrong. You're treating a churn signal (the customer stopped buying) as a pricing signal (they need a deal). Sometimes that's true; usually it isn't.
What works better in our scoring data:
Day 30 — Specific service question. (Template)
"Still using your [last purchase]? If you're using it, you might be about to run out."
The framing is service-flavored, not sales-flavored. Customers who reply tell you whether they're still in the active relationship or have walked away.
Day 60 — Open question. (Template)
"What changed? Reply 'happy' / 'tried X' / 'issues' and we'll respond accordingly."
The pre-formed reply options do the heavy lifting. Customers who'd never write a 300-word email will reply with one of three words. That signal becomes your churn intelligence.
Day 90 — Permission pass. (Template)
"We're going to stop emailing next week. If you'd like to stay, click below."
Counterintuitively, telling customers you're about to remove them re-engages a portion of them. The framing is honest, not manipulative.
Day 120 — Discount, only if needed. (Template)
"Here's 20% — for old times' sake. No deadline tricks."
If you're going to discount, do it last in the sequence and be honest about it. 'No deadline tricks' is doing trust work; most discount-led winbacks ruin it with fake countdowns.
Three structural moves:
Run this sequence for three months and you'll have classified your churned customers into segments you didn't have before:
Each segment deserves a different follow-up. Generic winback can't do this; question-led winback can.
Same structural moves, different cadence:
For SaaS, the post-cancel survey is the highest-value email. The honest answers about what made the customer leave become product roadmap data. Most teams ignore this; the ones who don't compound advantage over years.
If you run a discount-led winback sequence today, the test:
In our data, the recovery rate goes up about 30-50% on the new sequence (relative, not absolute). The bigger win is the reply data — you'll learn things about your churn that you didn't know.
For more, the winback templates section has the full set of patterns by segment. And our predicted-A/B-winner tool can score your two variants before you send.
Free anonymous run, no signup.
Try the editor