Skip to main content
michael_lifecycle
Problem Solver I
Problem Solver I
September 27, 2026
Solved

Is “45 days since last purchase” enough for a win-back segment?

  • September 27, 2026
  • 3 replies
  • 47 views

I’m testing a simple question around win-back segmentation:

If a product normally has a 30–45 day reorder cycle, is “45 days since last purchase” enough to decide that a customer should enter a win-back flow?

I’m finding that the same lapsed window can represent very different situations:

• genuinely overdue
• naturally low-frequency
• stockpiling after a large order
• seasonal purchasing
• recently engaged but not ready
• previous support issue
• little current buying intent

So I’m considering combining:

1. Product-specific reorder behavior
2. Purchase frequency
3. Recent clicks / site activity
4. Previous support interactions
5. Order-size anomalies

Then testing different actions by group rather than giving everyone the same discount.

Curious how others handle this in Klaviyo:

Do you use a fixed “days since last purchase” rule, or combine it with product/customer behavior?

Best answer by ArpitBanjara

Hey ​@michael_lifecycle 

A fixed 45 days won't work for everyone, because the same gap can mean something different for each customer. Klaviyo has a better option built in called Expected Date of Next Order. It looks at each customer's own past orders instead of using one number for the whole list, and Klaviyo's own team suggests using it over Churn Risk for win back flows, since Churn Risk tends to flag brand new customers as "at risk" right after their very first order. You can read more on how it works here: Klaviyo's guide to predictive analytics.

This feature only turns on once your account has at least 500 customers who've ordered, 180 days of order history, and some repeat buyers. If your account isn't there yet, try Average Days Between Orders instead of a flat 45. It's a simple calculation based on real order history, so it works even before predictive analytics kicks in.

Clicks and site visits are easy to add in. Support tickets and unusual order sizes aren't something Klaviyo tracks by default, so if you want to use those too, your helpdesk tool would need to send that info into Klaviyo first. Until then, it's easier to just handle those cases by hand, separate from the main segment.

I hope this helps and thank you for sharing your question here in the community.

Cheers,

Arpit.

3 replies

ArpitBanjara
Principal User II
Principal User II
September 27, 2026

Hey ​@michael_lifecycle 

A fixed 45 days won't work for everyone, because the same gap can mean something different for each customer. Klaviyo has a better option built in called Expected Date of Next Order. It looks at each customer's own past orders instead of using one number for the whole list, and Klaviyo's own team suggests using it over Churn Risk for win back flows, since Churn Risk tends to flag brand new customers as "at risk" right after their very first order. You can read more on how it works here: Klaviyo's guide to predictive analytics.

This feature only turns on once your account has at least 500 customers who've ordered, 180 days of order history, and some repeat buyers. If your account isn't there yet, try Average Days Between Orders instead of a flat 45. It's a simple calculation based on real order history, so it works even before predictive analytics kicks in.

Clicks and site visits are easy to add in. Support tickets and unusual order sizes aren't something Klaviyo tracks by default, so if you want to use those too, your helpdesk tool would need to send that info into Klaviyo first. Until then, it's easier to just handle those cases by hand, separate from the main segment.

I hope this helps and thank you for sharing your question here in the community.

Cheers,

Arpit.

Arpit Banjara - Lead Email Marketing Specialist at Flowium
michael_lifecycle
Problem Solver I
Problem Solver I
September 28, 2026

This is very helpful — Expected Date of Next Order is exactly the kind of built-in capability I wanted to understand before assuming another tool was needed.

It also changes the question I’m testing.

If Klaviyo can already estimate when someone is likely to reorder, the remaining question is less about prediction and more about decision + incrementality:

  1. Why is this customer not reordering?

  2. What intervention should we try?

  3. Would they have reordered without it?

I’m particularly interested in the third question. In your experience, do merchants actually use a clean holdout to measure incremental revenue from win-back flows, or is attributed revenue still the usual measure?