iROAS (incremental return on ad spend)

iROAS is incremental sales divided by media cost: the revenue a campaign caused, as estimated against a baseline from an accepted incrementality method, per unit of spend.

The difference from ROAS is the numerator. ROAS credits every sale the attribution rules allow; iROAS credits only the sales above what the control group, synthetic control or model says would have happened anyway. The IAB/MRC guidelines list the accepted methods (randomised controlled trials, synthetic controls, matched-market tests, machine-learning models) and require disclosure of the design (IAB/MRC, January 2024, chapter 4). IAB Europe's V2.1 makes incrementality optional but, if offered, bound to the joint November 2025 guidelines (IAB Europe, V2.1; IAB, 3 November 2025).

iROAS is usually a fraction of ROAS, which is why networks resist reporting it and brands insist. It is also noisy: a holdout of a few thousand shoppers on a low-frequency category produces wide confidence intervals, and the guidelines ask for those intervals to be disclosed (IAB/MRC, January 2024, section 4.4).

Example: the campaign above is credited with $400,000 attributed sales (ROAS 8.0). A 10% randomised holdout shows exposed shoppers bought 6% more than controls; incremental sales are $24,000 on $50,000 spend: iROAS 0.48. The brand now knows the campaign mostly reached people who were buying anyway and can change targeting.

Related: ROAS, incrementality, closed-loop measurement, retail media network.

Source: IAB/MRC, "Retail Media Measurement Guidelines" (January 2024), chapter 4; IAB and IAB Europe incremental measurement guidelines (3 November 2025).

1. IAB/MRC, January 2024, section 4.4
2. IAB Europe, V2.1
3. IAB, 3 November 2025