In one line
Incrementality is the pure additional outcome an ad actually caused — not every conversion among people who saw it, but the share that wouldn't have happened otherwise.
How it's measured
The most reliable method is a holdout test: randomly split users into an exposed group and a holdout group, then compare conversion rates. The difference is uplift.
Why CPA/ROAS alone isn't enough
CPA and ROAS only describe people who saw the ad — they can't tell you whether the ad caused the conversion. If incrementality is near zero, CPA can look great while the ad is doing almost nothing.
Go deeper
The three practical ways to measure incrementality (holdout, ramp-up, ramp-down) are covered in Measuring Incrementality.