Brand lift is the causally-measured incremental change in upper-funnel brand outcomes — ad recall (aided and unaided), brand awareness, consideration, message association, purchase intent, and favorability — attributable to a media campaign. It is typically measured by a randomized exposed-vs-control survey, either embedded in the ad platform running the campaign or fielded off-platform by a third-party panel provider.
Mechanically, users are randomized into an exposed arm that receives the campaign and a control arm that does not, with the specific implementation varying by platform: some withhold delivery to the holdout, others substitute a non-campaign creative such as a public-service spot, and the exact mechanics shift over time and across markets. After the exposure window closes, the survey fires against both arms. Lift is reported as exposed_metric - control_metric in percentage points, not relative percent. That makes brand lift the upper-funnel counterpart to a conversion-lift test: same randomized causal design, but the outcome class is attitudinal movement that does not surface in attribution reports on a short cycle.
The credibility caveat is statistical power. Attitudinal shifts of this kind are small, and resolving them takes a large exposed cell — which is why sub-scale media budgets often produce a platform-reported number without a defensible read behind it. Brand lift is a planning instrument for upper-funnel media against outcomes the conversion pipeline cannot see; it is not a substitute for conversion measurement, and a low-sample platform result is directional at best.