How incrementality works
Attribution assigns credit for conversions that already happened. Incrementality measures cause by comparing a group exposed to ads against a comparable group held back from them. The gap between the two is the lift, the conversions the advertising genuinely produced. This matters because some of the best-looking spend in any account is claiming sales it did not cause, especially branded search and retargeting, where the buyer was already on the way to purchase. A great reported ROAS can hide near-zero incremental value.
How incrementality is measured
The two workhorse methods are conversion lift tests, where a random slice of the audience is withheld from ads, and geo tests, where matched regions are turned on or off and outcomes compared. Both create a control group so you can see what happens without the spend. The output is an incremental cost per acquisition, which is almost always higher than the platform-reported number, because it strips out the conversions that would have arrived for free.
Where incrementality changes decisions
The usual surprises are branded search, retargeting, and broad Performance Max. Branded search often shows low incrementality when organic already owns the result, so some of that spend is buying clicks you would have won anyway. Retargeting frequently over-claims because it reaches people already close to buying. Prospecting, by contrast, often proves more incremental than last-click attribution gives it credit for, since it reaches people who were not searching yet. Incrementality is what corrects those distortions.
How iClick uses incrementality
iClick treats a suspiciously high reported ROAS on branded or retargeting campaigns as a hypothesis to test, not a result to celebrate, and pressure-tests it with a holdout before scaling budget into it. The working principle is that the metric which pays the business is incremental profit, not claimed conversions. When a lift test shows a channel is over-credited, budget shifts toward the campaigns that genuinely move the top line.
