How to Identify Lapsed Customers in Ecommerce
If you want the fastest possible way to find lapsed customers, create a segment:
Has not purchased in 90 days.
Done.
If you want the right customers, keep going.
Because the 90-day filter is only a starting point.
A serious lapse analysis should ask:
- what they bought,
- how often they bought,
- how much they bought,
- where they are in the customer journey,
- whether the current gap is unusual.
Identifying lapse is a detection problem.
Not a date filter.
Layer 1: Store-wide time between orders
Start simple.
For repeat customers, calculate:
- average days between orders,
- median days between orders,
- distribution.
RetentionX explicitly reports both average and median time between purchases because the two can tell different stories. (help.retentionx.com)
This gives you the baseline.
Layer 2: Break the data down by order number
Order one to two may take:
70 days.
Order two to three:
Order four to five:
If you use one store-wide average, you lose that progression.
The customer who has bought six times may have a much more predictable cadence than someone who bought once.
Layer 3: Break the data down by product
A replenishable product may have:
30-day consumption.
Another:
A durable product:
years.
The product often tells you more than the customer’s last-order date.
Layer 4: Use the individual customer history
For customers with enough history, ask:
What did this person usually do?
A customer who ordered every 40 days five times in a row has given you a strong baseline.
The sixth gap is now meaningful.
Layer 5: Check whether they are even reachable
A customer can be commercially lapsed but unsuitable for marketing.
Look at:
- unsubscribe,
- hard bounce,
- permission,
- engagement.
Identification and activation are separate steps.
Ometria’s order-gap logic is a useful example
Ometria acknowledges that its default at-risk and lapsed windows may not fit every business and explicitly offers order-gap analysis to help merchants understand the actual length of time between customer purchases. (success.ometria.com)
That is exactly the right mindset.
Your own data should help define your lifecycle.
Bluecore shows why category context matters
Bluecore defines inactive buyers in its benchmark framework as customers who purchased previously but not within the last year, then splits them into longer inactivity bands.
But its own category benchmarks show meaningful differences in purchase frequency across verticals. (support.bluecore.com)
So even a well-designed benchmark should not replace store-specific logic.
Build a lapse matrix
A simple working model:
| Order history | Current timing | Interpretation |
|---|---|---|
| 1 order | inside typical order-2 window | Active |
| 1 order | beyond typical order-2 window | Second-purchase overdue |
| 2–3 orders | slightly beyond cadence | At risk |
| 4+ orders | well beyond personal cadence | High-confidence lapse |
| Any | deep inactivity + no engagement | Deep lapse / cautious |
This is much more actionable than a single inactivity field.
Use quantity where possible
If the customer bought multiple units, adjust expected timing.
A replenishment algorithm that ignores quantity will generate bad triggers.
Use category transitions too
Sometimes the customer is not late buying the same thing.
They may be late making the next logical purchase.
RetentionX’s product intelligence includes follow-up purchase and cross-selling analysis precisely because what customers buy next can matter as much as whether they repurchase the same SKU. (help.retentionx.com)
This is especially important for:
- apparel,
- durable goods,
- beauty routines,
- multi-product wellness.
Detect deterioration before disappearance
The ideal system does not only identify:
customer gone.
It identifies:
customer behaviour weakening.
Watch for:
- order gaps stretching,
- order value declining,
- fewer categories,
- cancellation.
That creates an opportunity to intervene earlier.
A worked example
A coffee brand has:
80,000 customers.
Median second purchase:
37 days.
Median later reorder:
29 days.
Customer A:
1 purchase, day 25.
Not lapsed.
Customer B:
1 purchase, day 75.
Second-purchase overdue.
Customer C:
8 purchases, historical cadence 27–32 days, now day 55.
Strong lapse signal.
Customer D:
4 purchases, last order included four bags, day 55.
Probably not overdue yet.
One store.
Four customers.
Four answers.
Create useful operational segments
You might end with:
Second-purchase overdue
Recent repeat lapse
Lapsed VIP
Replenishment overdue
Deep lapse
Those become the inputs to prioritization.
Frequently Asked Questions
Can I identify lapse from Shopify?
Yes, at a basic level. Shopify provides order history, cohorts, RFM and customer segmentation.
Do I need predictive AI?
No. Good purchase-history logic can take you a long way.
What is the minimum data needed?
Last purchase date, order count and order value are a start. Product and quantity data make the analysis much stronger.
Finding lapsed customers is not the hard part
Any system can find people who have not ordered recently.
The hard part is distinguishing:
someone who is genuinely behaving differently
from:
someone who simply has not needed to purchase again yet.
That is what makes the segment useful.
Related reading
How to Prioritize Lapsed Customers