A win-back email is worth testing when a past customer is late to buy again, you have a useful reason for them to return, and you have permission to email them. Decide how the test will end before you send it. If someone remains inactive after that planned attempt, apply your suppression rule instead of adding another reminder.
But a missing order is not enough to call someone inactive. A customer who still clicks through to your store may need more time. Someone who has ignored your emails and stopped visiting for a sustained period needs a different decision.
The job is to tell those people apart. “Send everyone a discount” does not do that job particularly well.
First, decide what you are trying to win back
Two problems often get bundled together.
One customer still reads about your products but has not bought again. Another customer has stopped responding altogether. The first problem is a missing purchase. The second is a fading email relationship.
For this plan, use win-back for a bounded attempt to bring a past buyer back. Use sunset for the final decision about subscribers with sustained inactivity. That distinction helps you avoid suppressing interested people just because they did not buy on your preferred timetable.
Klaviyo’s win-back guidance ties the starting point to the normal buying cycle. Its sunset guidance deals with sustained inactivity and removing unresponsive profiles from marketing sends. Those are related decisions, but they need different evidence.
Work out whether the customer is actually late
Start with order history for the relevant product or customer group. Look at the time between repeat purchases. Separate products that run out from products people expect to keep for years.
A customer who bought a dining table may be perfectly happy with it. Six months without another dining table is not a retention emergency.
Our recommendation is to write a plain eligibility statement before building the flow:
This test is for past buyers of this product group who are beyond our usual repeat-purchase window, can receive marketing email, and have not already bought again.
Then check that the data can support each part. If you have very few repeat orders, treat the timing as a hypothesis. Start with a small, relevant group and record what you learn. Do not turn a guess into a permanent account-wide rule.
Recent orders, unresolved service issues and overlapping promotions deserve a check too. A person waiting for a replacement order probably does not need a cheerful invitation to spend more.
Qualify people before writing the offer
Use buying behavior and email engagement together. This is the decision table we would start with, then adapt to the business.
| What you can see | What to do next |
|---|---|
| The customer is still within the normal buying cycle | Leave them out of this win-back test. Keep any useful, relevant communication. |
| They are late to buy, still show interest, and can receive marketing email | Test a useful reason to return with a clear limit on follow-ups. |
| They have shown sustained inactivity | Review them against your sunset criteria. Do not put them back into every routine promotion. |
| They bought again | Stop the win-back messages and move to the appropriate post-purchase experience. |
| They cannot receive marketing email | Exclude them. Do not change that status to make the test bigger. |
There is also an account-level check. If you are repairing sender reputation, pause this idea. Klaviyo advises against running win-back, re-engagement and sunset flows during reputation repair. A quiet audience is not the place to prove that sending more will fix delivery.
Give the test a finish line
Before the first message, agree on five things:
- Who enters. Define the eligible customer group and exclusions.
- What they receive. Set the message limit and gaps between sends.
- What stops the messages. A new purchase should stop a win-back pitch. Other changes in eligibility should stop it too.
- When you judge the result. Allow a reasonable response window after the final message.
- What happens next. Decide how buyers, interested non-buyers and unresponsive people will be treated.
Here is an example to make that concrete. A brand might test two messages, one week apart, then review the group one week after the second message. That is an illustration, not a recommended timing for every store. A considered purchase may need much longer. A consumable may need a different approach entirely.
The useful part is that the team chose the endpoint before seeing the results. Nobody gets a surprise third “last chance” email because the first two were disappointing.
In Klaviyo, the win-back setup guide uses a profile filter for Placed Order zero times since starting this flow. Check the relevant order events from every sales integration you use. Test the exclusion with real profile histories before making messages live.
Give the customer a reason to care
A good message answers a customer question: Why would I want to come back now?
That could be a useful product update, a relevant new option, help choosing the right version, or a reminder tied to a genuine repeat need. Use what they bought to make the message useful. You do not need to announce that you have been watching the calendar.
An incentive can be part of a deliberate test. Decide what it costs and who needs it. If a customer already showed strong buying intent, a discount may reduce the value of an order you would have received anyway.
Keep the email focused on one reason to return. Make changing preferences or leaving easy. Then respect the response.
Measure the group, not just the nicest number
Our recommendation is to judge the test on both the buying result and the cost of getting it.
Record the eligible audience size before sending. Review orders and net revenue per eligible customer over the same observation window. Include discount cost and margin where the data allows. Check useful clicks, unsubscribes and complaints alongside sales.
If the audience is large enough for a useful comparison, randomly hold out a group from the win-back messages. Keep other treatment comparable. That can help separate orders encouraged by the sequence from orders that would have happened anyway. Without that comparison, report the observed result without calling every attributed order an extra order caused by email.
Do not use an open alone as proof that the relationship recovered. Klaviyo notes that Apple Mail Privacy Protection can inflate opens. Look for stronger activity and check the quality of your tracking before changing someone’s treatment.
Close the loop when the test ends
At the review date, make the next action explicit.
Bought again: end the win-back sequence. Check that they are receiving the right help after the order.
Clicked or otherwise showed useful interest without buying: look at the product, price, timing and landing page. Keep them out of an automatic “no order equals suppress” rule. They may need a more relevant message or a longer gap.
Still inactive, with no useful response: if they meet your defined long-term inactivity criteria, apply your suppression process. Do not quietly put them back into broad campaigns.
Klaviyo’s sunset example uses a profile property to identify unengaged people, followed by a separate suppression step. A label is not the same thing as stopping marketing emails. Check that the final action actually happened.
Keep the decision and the reason in your records. You want to know who left the sequence, who needs a different approach, and who should stop receiving marketing email. That is more useful than keeping every profile active because the list looks bigger that way.
Start with the rule you will actually use
Before editing the subject line, complete this sentence:
We will send this group a limited win-back sequence. We will stop when they buy or become ineligible. At the review date, we will separate useful interest from sustained inactivity and act on that result.
If the team cannot agree on that sentence, the flow needs another decision before it needs another email.
For the broader process, read our guide to improving Klaviyo flows. It covers how to decide which flow problems deserve attention first.
Common questions
When should a win-back email start?
After the customer is meaningfully beyond the buying window for the relevant product or customer group. Use your own purchase history. There is no single day count that fits both durable goods and products people use up.
Should everyone who does not buy be suppressed?
No. A missing purchase and sustained inactivity are different things. Check useful engagement, buying context and your defined sunset criteria before deciding to suppress.
Does a win-back sequence need a discount?
No. Start with a relevant reason to return. If you test a discount, measure its cost as well as the orders it receives.
What is the stop rule?
It defines when follow-ups end and what happens after the review window. It should cover new purchases, changes in eligibility and the next step for people who remain inactive.