Customer Segmentation for Win-Back Campaigns: Who Should Actually Get the Message?
The customer who bought from you eight times should not get the same win-back email as the person who bought once on Black Friday.
That one sentence is almost the entire segmentation argument.
The only reason generic win-back flows exist is because:
“Has not purchased in 90 days”
is easy to automate.
Easy is not the same as intelligent.
Segment 1: Recent repeat lapse
Strong prior frequency.
Recent behaviour break.
Usually one of the best places to start.
Segment 2: Former VIP
High historical value.
May justify:
- recognition,
- high-touch service,
- relevant newness.
Segment 3: One-time second-purchase overdue
A different problem.
They never established loyalty.
Focus on:
- product success,
- next purchase,
- education.
Segment 4: Replenishment overdue
The Return Reason may simply be:
You should need more.
Timing and quantity matter.
Segment 5: Cancelled subscriber
Use cancellation reason.
Do not treat:
price
the same as:
excess inventory.
Segment 6: Deep lapse
Higher risk.
Lower brand memory.
Needs a stronger reason to exist.
Carhartt shows what richer segmentation looks like
Bluecore describes Carhartt creating different inactive-customer strategies based on:
- lapse duration,
- purchase history,
- colour affinity,
- discount affinity,
- browsing behaviour. (bluecore.com)
That is a customer strategy.
Not merely a date filter.
Tibi uses relationship-state changes
Klaviyo reports that Tibi triggers win-back campaigns when high-value recency, frequency and monetary-value groups move into weaker states, helping the brand distinguish deterioration among previously valuable customers. (klaviyo.com)
Again:
the change matters.
Plank+Beam used predicted churn risk
RetentionX reports that furniture brand Plank+Beam separated customers into different predicted churn-risk tiers and varied incentives accordingly.
The test group reportedly generated an 82% higher order rate than the control group, and the program recovered 1.17% of customers within ten days. These are vendor-reported figures. (retentionx.com)
The strategic point is stronger than the case-study number:
Different levels of customer risk can justify different interventions.
Discount intensity should follow segment economics
Former VIP:
20% may be justified.
Low-value one-time buyer:
perhaps not.
Do not give the most expensive offer to the largest audience simply because automation makes it easy.
Use holdouts by segment
If possible, test:
Recent repeat lapse treatment vs holdout.
Then:
Former VIP treatment vs holdout.
This tells you which segments actually produce incremental response.
Build the win-back matrix
| Segment | Return Reason | Incentive |
|---|---|---|
| Recent repeat | Resume habit | Often light |
| Former VIP | Recognition/newness | Tailored |
| One-time | Product progression | Test |
| Replenishment | Need more | Often none |
| Cancelled subscriber | Reason-specific | Depends |
| Deep lapse | New relevance | Stronger |
Now copy becomes much easier because the strategy is clear.
The win-back flow should be the output of segmentation
Not the starting point.
Start with:
Who are they?
Then:
What happened?
Then:
Why should they come back?
Then write.
Related reading
How to Prioritize Lapsed Customers