Short answer
Set win-back timing from your repurchase cycle, send up to three messages, then suppress unresponsive profiles. Measure clicks, orders and list health.
The answer in detail
Set inactivity from your own repurchase cycle, send no more than three win-back messages as a starting policy, and stop marketing to people who remain unresponsive. The sequence helps maintain a usable audience as much as it generates sales. It should not become an endless stream of discounts.
The benchmark
The original article cites Klaviyo's 2026 benchmark report, covering more than 110,000 accounts: average attributed revenue per recipient was $0.82 for win-back, against $6.77 for abandoned checkout. Those are vendor averages, not guaranteed incremental revenue.
An anonymized women's fashion brand reported 1,088 win-back sends over 90 days, one €135 order and ten unsubscribes. Its single promotional message went to everyone without a recent purchase. That case illustrates weak targeting; it does not establish a universal conversion rate.
Skincare may be replenished within weeks; a coat may be replaced after a year. Klaviyo recommends a first delay slightly longer than the normal buying cycle, with 180 days as an example, and up to three emails. The source's second-purchase example had a median of 139 days while the mean exceeded 280. A few very late buyers can distort the mean.
| Customer state | Operational interpretation | Starting action |
|---|---|---|
| Still within normal buying cycle | Not yet overdue | Usage or replenishment content |
| Beyond the cycle plus a buffer | Potentially inactive | Eligible win-back sequence |
| No meaningful engagement for months | Disengaged | Sunset sequence, then suppression |
| Retention limit reached | Legal/data-policy question | Delete, archive or otherwise handle under the applicable policy |
Retention is a separate purpose-specific decision from campaign suppression. Under the GDPR where applicable, personal data must not be kept longer than necessary for its purpose, taking account of legal recordkeeping requirements. Set review or deletion limits; do not treat a universal three-year clock as a global rule.
The rule
Cap the sequence and make the exit real. Continued messages to people who never respond can damage the effectiveness of other campaigns.
The editorial recommendation differs from Klaviyo's example: do not lead with a discount. Reminding someone of a useful product or a meaningful range update is a better first test than teaching them to wait for a promotion. If an offer is necessary, evaluate its margin cost and reserve it for a later message. Some recipients would have returned without it.
Apply the relevant marketing rules before entering anyone into the sequence. EU data protection, UK PECR and US CAN-SPAM should not be merged into one global consent or retention rule.
The steps
- Use the repeat-purchase analysis to calculate the median repurchase interval by product family, then add a reasonable buffer. If using Klaviyo's export of customers with at least two orders in two years, understand that taking the median of each customer's Average Time Between Orders differs from taking the median of all order intervals.
- Trigger from Placed Order with a delay matching your threshold. Use an exit condition such as Placed Order zero times since starting this flow so a returning buyer stops receiving win-back reminders.
- Send a first message without an offer. Refer to the earlier purchase, suggest a relevant complement and explain what has changed.
- Give a specific reason to return in the second message. If you add an incentive, state a genuine end date and check contribution.
- Use the third message to offer a clear frequency choice and an easy unsubscribe. Do not obscure the exit behind a preference-center maze.
- Move unresponsive people into a sunset process. The source's Klaviyo example uses an existing Sunset (Email) segment and an Unengaged property. It reports that a flow does not itself perform suppression: use the segment's suppression action or an appropriate API process. Verify current controls before relying on automation.
The pitfall
Do not judge reactivation by opens. Apple's Mail Privacy Protection can download content in the background, producing apparent opens without a person reading the email. Prefer clicks, site activity and orders, with tracking limitations acknowledged.
A textile-homewares business described by the contributors had 63% of profiles with no recorded opens. A campaign to engaged subscribers achieved a 5.7% click rate, versus 1.4–2.2% for whole-list campaigns. It moved to a 90-day engaged audience; the source explicitly says the deliverability effect remained unmeasured. Do not present the comparison as proof of that effect.
Google's current sender guidance requires keeping spam rates reported in Postmaster Tools below 0.3% when sending to personal Gmail accounts. Monitor that platform-specific measure alongside sales while cleaning the list; it is not a worldwide legal threshold.
What to tell your team
“In this fictional list, 14,000 of 40,000 subscribers have neither clicked nor visited for a year. We will send three eligible messages, retain people who respond and suppress the rest from campaigns. Our separate retention policy determines which records must be kept or deleted.”
