The short answer
Attributed ROAS uses revenue credited to advertising under a reporting rule. Incremental ROAS asks how much additional revenue advertising produced relative to what would have happened without that advertising. The second question requires a credible comparison, not just a different division of the same platform totals.
Define the missing comparison
The counterfactual is the outcome without the advertising being evaluated. You cannot observe both outcomes for the same person at the same moment. A well-designed experiment can help estimate the comparison using treatment and control groups. Ordinary before-and-after totals may also reflect seasonality, prices, stock changes, or other marketing.
Incremental ROAS = estimated additional revenue ÷ additional ad spend
Illustrate the distinction
Illustrative inputs, not an experiment result: an ad report credits $20,000 of revenue to $4,000 of spend, producing $20,000 ÷ $4,000 = 5× attributed ROAS. Suppose a valid experiment separately estimates that the advertising produced $8,000 of additional revenue for that additional $4,000 of spend. Estimated incremental ROAS would be $8,000 ÷ $4,000 = 2×. The assumed $8,000 must come from an actual measurement design in real use; it cannot be inferred from the 5× figure alone.
Do not turn the example into a discount factor
The example does not establish that a fixed portion of attributed revenue is incremental. Effects can differ by audience, channel, brand demand, and time. Applying the example’s relationship to another campaign would invent information the report does not contain.
Check economics after measuring lift
Additional revenue still has associated costs. With an illustrative 50% contribution margin, the assumed $8,000 creates $8,000 × 0.50 = $4,000 before the additional ad spend. Subtracting $4,000 leaves zero contribution from the change, before excluded expenses. This calculation assumes that the incremental revenue has the stated margin.
Report uncertainty alongside the estimate
Describe the group assignment, test dates, exclusions, spending difference, and how uncertainty was estimated. A result too imprecise to distinguish meaningful gains from losses should remain inconclusive. Use attributed reporting for its stated purpose, and do not relabel it as an incrementality study.
Specify the missing comparison
- Gather: Record the attribution rule and the outcome an experiment would measure without the tested ads.
- Produce: Keep attributed ROAS separate from any experimentally estimated incremental return.
Decision boundary: A fixed percentage discount applied to attributed revenue is not an incrementality study.
Next reading
Examples are illustrative, not client results or benchmarks. The calculations use the inputs shown; real-world decisions depend on your records and assumptions. Report a correction.