If Klaviyo attributes $X of revenue to a win-back flow, how much of that revenue was actually incremental?

A question I’m pressure-testing:

If Klaviyo attributes $X of revenue to a win-back flow, how much of that revenue was actually incremental?

My current test structure is simple:

Eligible customers
→ randomized treatment / holdout
→ same measurement window
→ repeat purchase + revenue per eligible customer
→ compare the difference

The numbers in my example are synthetic.

I’m more interested in how merchants actually make this decision with smaller audiences, where statistical significance can take much longer.

CAMPAIGN

1,000 eligible customers

        ↓

Random split

900  Treatment
100  Holdout

        ↓

30-day observation

Purchase rate
Revenue / eligible customer
Complaints

        ↓

Incremental lift