Incremental ROAS is the lift in revenue specifically caused by a campaign, not just the revenue that happened to occur during it. Last-click ROAS gives every conversion to whichever ad got the final click — even if that customer was going to buy anyway. iROAS strips out the would-have-converted-anyway and shows the actual marketing-driven lift.
How to measure it
- Geo-holdout tests: turn ads off in one region, run normally in a comparable region, compare revenue. The cleanest experiment.
- Conversion lift studies: Meta + Google can run randomized holdout audiences if you have enough volume
- Audience-level holdouts: serve a control 'no ads' group within your normal audience, compare conversion rates
- Time-based: pause ads for a defined period and watch organic+direct revenue. Noisy but cheap.
Why it matters
Most brands' platform-reported ROAS overstates incrementality by 30-60%. Branded search is the worst offender — those clicks were going to convert anyway. Retargeting often inflates against people who'd repurchase organically. iROAS forces you to confront whether you're earning revenue or just claiming credit.
Operator note
iROAS is more rigorous than ROAS but harder to measure cleanly without volume. The pragmatic path: track ROAS daily, run a quarterly geo-holdout or pause-test, and discount your platform ROAS by the gap you find. If your platform reports 4× and your geo-holdout suggests 2.5×, plan budgets against the 2.5×.