Incremental ROAS (iROAS)
Also known as: Incremental ROAS, incremental return on ad spend
The revenue that would not have happened without ads, per dollar of ad spend.
Attribution reports credit; incrementality reports causality. iROAS is the only ROAS number that answers 'what did we actually get for the money?'.
Definition
iROAS measures the causal lift of ads: revenue that only exists because of ad exposure, divided by ad spend. Reported ROAS includes purchases that would have happened anyway; iROAS strips those out.
Why it matters
Half of what platforms call 'ROAS' on retargeting is often revenue you'd have earned from customers already headed to checkout. iROAS separates real growth from vanity.
Formula
iROAS = (Test Revenue – Control Revenue) ÷ Test Ad Spend
How to measure iROAS
Two common approaches: (1) user-level holdouts — a random % of the audience sees no ads, and you compare revenue between exposed and holdout groups; (2) geo-level tests — a subset of geographies runs a spend change while the rest stay flat, and you model the difference. Meta Lift Studies are the built-in version; Nielsen and Kantar sell more rigorous versions.
Test cell: $120k revenue at $30k spend. Control cell (no ads): $90k revenue. iROAS = ($120k – $90k) ÷ $30k = 1.0.
Common mistakes
- ✕Assuming reported ROAS equals iROAS.
- ✕Running incrementality tests too short to stabilise.
- ✕Confusing iROAS with MMM-modeled lift.