What is Incrementality?

Incrementality exists because attribution systematically over-credits paid ads, especially retargeting and branded search, which frequently get the click from buyers who were already going to purchase. A retargeting campaign can post a 6x reported ROAS while its true incremental ROAS (iROAS) is close to 1x, because most of those people would have come back on their own. You measure it by running a controlled experiment: a geo test splits markets into a treatment region that sees the ads and a matched control region that does not, then compares total sales between them; a conversion-lift or ghost-ads holdout does the same at the user level by withholding ads from a random control group. The gap between the two groups is the real lift. The case for incrementality is that it is the only metric that answers the question that actually matters: if I turn this spend off, how much revenue do I lose. The honest case against it is cost and patience. Clean tests need real scale and weeks of data, they cost you the control group's potential revenue while running, and small brands often cannot reach statistical significance at all, which is exactly when blended ROAS and MER become the more practical guardrails. For most DTC advertisers in 2026 the workable answer is to run periodic incrementality checks on the channels you most suspect of free-riding (retargeting, brand search) and let blended metrics govern the day to day.