Attribution answers "which channel gets credit for a conversion". Incrementality asks something more fundamental: how many conversions wouldn't have happened without the ad at all? That's the difference between what advertising measures and what it really adds.
Why attribution isn't enough
Attribution credits the conversion to a channel that was on the journey — but doesn't tell you whether the customer would have bought anyway. Example: brand ads to people who already buy show a great ROAS but really add little. Incrementality exposes that illusion.
How incrementality is measured
- Holdout test. You withhold ads from part of the audience (a control group) and compare conversions with those who saw the ad. The difference is the incremental lift.
- Geo experiments. You run ads only in some regions and compare with similar regions where they didn't run.
- Conversion Lift tools. Some platforms offer their own lift tests.
Where measurement comes in
Incrementality is only as good as the data behind it. If client-side loses conversions unevenly (e.g. more on iOS), the experiment is skewed. Reliable, complete measurement via server-side gives experiments a clean basis — without it you're comparing patchy numbers.
What to watch
- Incrementality is harder than attribution — it requires an experiment, not just a report.
- Treat it as a complement to attribution, not a replacement — see conversion attribution.
- It needs enough volume for the result to be statistically meaningful.
Summary
Incrementality measures the real impact of advertising — how much wouldn't have happened without it — and exposes channels that just collect credit for conversions that would have happened anyway. It rests on an experiment and on clean data that reliable server-side measurement provides. More in the complete guide.