Incrementality

Incrementality measures how many extra conversions a channel or campaign actually created that would not have happened without it, not merely how many conversions were credited to it.

Incrementality asks a more fundamental question than attribution: attribution asks which channel a conversion's credit should go to, incrementality asks whether that conversion happened because of the channel at all or would have happened anyway. The difference becomes critical when a channel claims conversions that would have occurred on their own — the classic case being an ad on branded search that grabs an already-decided user. The only valid way to measure it is an experiment: a group exposed to the channel and a control group that does not see it, and the gap in conversion rate between the two is the incremental effect. So incrementality is a matter of experiment and experimental design, not of an observational report; no attribution model, however sophisticated, replaces a controlled test.