Measuring how many sales a channel or campaign actually caused, by comparing a group that saw it against a similar group that did not. It separates credit from cause.
Before scaling a big channel, when platform-reported ROAS looks too good, for retargeting and branded search, and when budgets are being reallocated.
Geo holdout tests with spend paused in matched markets; platform conversion lift studies; turning off branded search in a test region; a holdout group excluded from an email series.
The most honest answer to 'is this working?' for any team spending real money. Flagged when attribution dashboards are treated as proof of cause.
