What is brand lift measurement?
Measuring whether advertising changed what people think — awareness, recall, consideration, favourability — rather than what they immediately bought. It exists because a great deal of advertising works on outcomes no click can capture.
The problem. Digital measurement is built around conversions, so it systematically over-credits activity that produces immediate clicks and under-credits everything else. Advertising that makes someone more likely to consider a brand in six months registers as a failure by that standard, which biases spend toward short-term response — a distortion documented at length in the marketing effectiveness literature.
How brand lift studies work. They are randomised controlled experiments, which is what distinguishes them from ordinary tracking:
The audience is randomly split into an exposed group and a control group that is withheld from the campaign.
Both groups are surveyed — typically a single-question poll served in the same environment — asking about recall, awareness or brand association.
The difference between groups is the lift attributable to the advertising. Randomisation is what makes the comparison causal.
What is measured: aided and unaided recall, awareness, message association, consideration, favourability and purchase intent. These sit at different points in a funnel and move at different rates — recall shifts fastest, consideration and preference far more slowly.
The limitations, which are real:
Survey response bias. People who respond to in-feed polls are not representative, and response rates are low.
Stated intent is a weak proxy for behaviour, and the gap between saying you would consider a brand and buying it is large.
Small effects need large samples, so studies require substantial spend to be conclusive.
Platform-run studies grade the platform's own work, which is an obvious conflict, and methodologies are frequently not fully disclosed.
Where it fits. Brand lift for short-term perception effects, marketing mix modelling for aggregate long-run contribution, and incrementality tests for causal sales effect. No single method covers all three, and using only conversion tracking is what the whole apparatus exists to correct.