Attribution tells you which touchpoints were present before a conversion. Incrementality tells you which actually caused it. The difference can be millions in misallocated budget.
The Attribution Problem
Attribution assumes correlation equals causation. If a user saw your ad and converted, the ad gets credit. But would they have converted anyway? Studies show 20-50% of attributed conversions would have happened organically.
Types of Incrementality Tests
- Holdout Tests: Split audience, show ads to test group only
- Geo-Based Tests: Use geography as test/control groups
- Ghost Ads: Show PSAs to control group
- Platform Lift Studies: Meta, Google built-in measurement
Sample Size Matters
You need roughly 400 conversions per group to detect a 20% lift with 80% statistical power. Underpowered tests waste time and money.
Calculating True ROI
If attribution shows 3x ROAS but only 40% is incremental, true ROAS = 3.0 × 0.4 = 1.2x. That "profitable" campaign barely breaks even.
Measure What Matters
ClicksFlyer helps you understand true marketing impact by combining attribution with incrementality insights.