Definition
Incrementality is the revenue a marketing channel or tactic actually caused, over and above what would have happened without it. It is the difference between the outcome with the activity and the outcome without it, and it is the only true measure of whether that activity worked.
The distinction matters because most reporting credits correlation, not causation. An attribution model sees that a shopper touched a channel before buying and assigns it credit, but it cannot tell whether the shopper would have bought anyway. Incrementality answers exactly that question.
How to measure incrementality
You measure incrementality with a controlled experiment, usually a holdout test:
- Split a comparable audience into a group that receives the channel or tactic and a group that is deliberately withheld from it.
- Run for long enough to gather meaningful data, sizing the test first so you are not fooled by noise, using a sample size calculator.
- Compare revenue between the two groups. The difference is the incremental revenue the activity caused.
- Confirm the result is real, not chance, with an A/B test significance check.
The result frequently differs sharply from what any attribution model reports, because channels like branded search and retargeting appear near purchases that would often have happened anyway, and therefore look far more valuable in a model than they are in reality.
Why incrementality matters for revenue
Incrementality protects you from the most expensive marketing mistakes: scaling a channel that gets credit for sales it did not cause, or cutting one that quietly drives demand. Because attribution rewards presence rather than causation, decisions made on attributed revenue alone can move budget in exactly the wrong direction.
This is why the honest discipline is to allocate with attribution but verify with incrementality. Use your attribution model as a working hypothesis for where revenue comes from, then test the big bets with a holdout before you commit. Our guide to revenue attribution covers this pairing in depth, and it becomes more important every year as privacy changes make attribution models lean harder on assumptions.
From a Revenue Intelligence perspective, incrementality is the confidence check that turns an estimate into a proven number. It is the same instinct as measuring a recovered leak against a baseline: do not trust that something worked until an experiment shows it did. The Revenue Intelligence framework treats this move, from estimated to measured, as the hinge of the whole discipline.
In short
Incrementality is the revenue an activity actually caused, measured with a holdout test rather than credited by a model. It is the honest check on attribution, it prevents costly budget mistakes, and it is how a marketing claim goes from plausible to proven.