How to Prioritize Lapsed Customers for Reactivation
If you gave me 100,000 lapsed customers tomorrow, I would probably ignore most of them first.
That sounds strange until you look at what “100,000 lapsed customers” actually means.
Inside that audience might be:
- former VIPs,
- recent repeat buyers,
- one-time discount shoppers,
- cancelled subscribers,
- three-year-old inactive addresses.
Treating all of those people as one opportunity is not strategy.
It is counting.
The first job is deciding:
Who deserves our attention?
RFM is the obvious starting point
RFM looks at:
- Recency
- Frequency
- Monetary value.
RetentionX describes these as core drivers of customer quality, while Shopify and Klaviyo both use RFM-based customer groups to identify valuable and at-risk customers. (help.retentionx.com)
It is a strong framework because it answers three useful questions:
How recently did they buy?
How often did they come back?
Were they ever worth much?
But RFM is not enough
Two customers can have similar RFM scores and completely different recovery potential.
Customer A:
Bought coffee.
Customer B:
Bought a sofa.
Same recency.
Same spend.
Different purchase logic.
This is where Altimax adds:
- product repeat logic,
- reachability,
- Return Reason.
Recovery Priority Score
A simple illustrative weighting:
| Factor | Weight |
|---|---|
| Recency relative to Lapse Window | 25% |
| Frequency | 20% |
| Historical value | 20% |
| Product repeat logic | 15% |
| Reachability | 10% |
| Return Reason strength | 10% |
Total:
100%.
This is not a universal algorithm.
It is a disciplined way to rank hypotheses.
Customer A
8 orders
$980 lifetime spend
25 days beyond normal reorder
replenishable product
recent email engagement
clear replenishment Return Reason
Possible score:
91
Customer B
2 orders
$180 spend
9 months lapsed
moderate repeat logic
reachable
possible score:
57
Customer C
1 order
$39 spend
3 years lapsed
no engagement
weak product logic
possible score:
14
Now you know where the first test belongs.
High value does not always mean high priority
A $2,000 historical customer gone for six years may be less recoverable than a $400 customer whose monthly habit broke last month.
Priority is not the same as historical value.
It is:
value × recoverability × contactability.
Expected value adds another layer
A rough expected recovery model:
Expected recovery value
=
Probability of recovery × AOV × expected future purchases × contribution margin
Example:
Recovery probability:
8%
AOV:
$100
Expected future purchase multiplier:
1.5
Contribution margin:
65%
Expected contribution per customer:
0.08 × 100 × 1.5 × 0.65
= $7.80
Across 5,000 customers:
$39,000 expected contribution before campaign costs.
Again:
scenario, not forecast.
But it makes the decision more economic.
RetentionX’s predictive tools show where this can go
RetentionX combines signals including:
- RFM,
- AOV,
- return behaviour,
- quantity,
- customer characteristics
to estimate churn risk and predicted LTV. (help.retentionx.com)
You do not need predictive modelling to begin.
But it shows that prioritization can move beyond static lists.
Shopify’s “Previously Loyal” segment is interesting
Shopify categorizes customers with low recency but a very strong history of frequency and spend as Previously Loyal. (help.shopify.com)
That is exactly the kind of group I would investigate.
But not automatically target.
Add:
- product,
- timing,
- marketing status.
Then decide.
First-time buyers need different scoring
They have no frequency history.
So prioritize using:
- first product,
- first-order value,
- acquisition channel,
- expected second-purchase window,
- cohort repeat rate.
Do not penalize them simply because F = 1.
Everyone begins there.
Cancelled subscribers need cancellation context
A former subscriber who cancelled because:
I have too much product
is different from:
I cannot afford this anymore
and:
I hated the product.
The cancellation reason can be more useful than an RFM score.
Reachability belongs inside value
A customer who looks economically attractive but is not appropriately marketable should not score as a practical recovery opportunity.
Opportunity must reflect execution reality.
Start with cohorts, not individual micro-scoring
Do not over-engineer.
You might begin with:
- Recent repeat lapse
- Former VIP
- Second-purchase overdue
- Cancelled subscriber
Then rank those groups.
That is often enough to identify the first test.
A worked prioritization
Lapsed customer base:
60,000.
Former VIP
2,000
Score: 86
Recent repeat lapse
5,500
Score: 82
Second-purchase overdue
18,000
Score: 63
Cancelled subscriber
4,500
Score: 60
Deep-lapsed one-time buyer
30,000
Score: 21
The largest segment is last.
Exactly as it should be.
Prioritization protects deliverability too
Sending to weaker, colder cohorts can create:
- complaints,
- unsubscribes,
- low engagement.
So Recovery Priority Score is not only about conversion.
It protects the asset you are using to recover revenue.
Frequently Asked Questions
Is RFM enough?
It is a strong starting point. Product and reachability context make it much better.
Should VIPs always go first?
No. A recent repeat buyer may be much more recoverable.
Should the biggest segment go first?
Almost never just because it is biggest.
Do I need machine learning?
No.
The best cohort is not the biggest one
It is the one where the story makes sense.
They used to buy.
They were valuable.
Something changed.
The product can still matter.
You can reach them.
And you have a believable reason for them to return.
That is where I would start.
Related reading
How to Identify Lapsed Customers
How to Measure Customer Reactivation
Want us to identify the cohort we would test first?
That is one of the core outputs of the Revenue Recapture Analysis.