Proscube
 
All articles

CRO

Win-back campaigns: who is worth going after and who is not

Segment by why they stopped, not by how long ago.

By Manpreet Singh·October 6, 2026·8 min read
Win-back campaigns: who is worth going after and who is not

The standard win-back is a 20% code sent to everyone who has not bought in 120 days. It recovers some revenue, costs margin on buyers who were coming back anyway, and tells you nothing.

The useful version segments by why they stopped.

The segments that matter

They bought once and never returned. Largest group, hardest to move. Something about the first experience did not land. A discount does not address that; a message that acknowledges it and explains how to get the result does better.

They bought repeatedly and then stopped. Smallest group, most valuable. Something changed — a bad delivery, a formula change, a competitor, or life. Worth a genuine question rather than an offer. The replies tell you things no survey will.

They bought seasonally. Not lapsed at all. A gift buyer in November is not churned in March. Sending them win-back emails misreads the behaviour entirely; time the contact to the season instead.

They ran out and forgot. Pure replenishment timing. No discount required, just a reminder calibrated to actual consumption rather than a flat interval.

Separating the seasonal buyers out is usually worth more than anything else here. In gift-heavy categories they can be a third of the list, and treating them as churned both wastes discount and makes your retention numbers look worse than they are.

The sequence for the once-and-never segment

Email 1 — the question. What stopped you? Genuinely asked, replies read by a person. The response rate is low and the information is worth more than the revenue.

Email 2 — the thing they probably missed. Usage, timeline, the result that takes longer than they expected.

Related service

List full of lapsed buyers?

See how we approach marketing

Email 3 — what has changed. New products, reformulation, faster shipping. A reason that is not a discount.

Email 4 — the offer, if you want to make one. Last, not first.

For the repeat-then-stopped segment

One email. From a person, not the brand. Asking what happened. No offer at all.

This is the highest-value email in retention marketing and almost nobody sends it.

WHERE SUBSCRIBERS ACTUALLY LEAVE100%M162%M249%M343%M439%M537%M6MOST CHURN IS THE SECOND ORDERWIN MONTH TWO AND THE COHORT FLATTENS
Illustrative cohort decay. Win-back works on the shallow part of this curve; past a certain point the cost of recovery exceeds the value.

When to stop

There is a point past which a contact is not a lapsed customer but a dead address hurting your deliverability. Suppress after a defined window of no opens and no clicks. A smaller list that lands in the inbox is worth more than a large one that does not.

Talk it through

Send us your lapsed segmentation.

Usually the first finding is that a large slice of it is seasonal buyers who never churned at all.

Get a read

A real read within one business day — not a sales call.

About the author

Manpreet Singh

Founder of Proscube. Leads the studio's Shopify, headless and CRO engineering, and works directly with founders rather than through account managers.

More about Manpreet

Work with us

Want help with this? Here’s how we do Digital Marketing.