Brand Lift

A business runs video ads to make more people recognise its name, but sales may not change during the campaign, so it surveys people about awareness. The study compares a group eligible to see the ad with a comparable group held out from exposure.

Google Ads describes Brand Lift studies and survey measures, and a Google research paper discusses methods for measuring online ad lift, and both make clear that design and response patterns matter because a raw before-and-after change does not isolate the campaign. Choose the question first, because awareness, ad recall, consideration and purchase intent are different outcomes, and select the one that matches the campaign goal.

Define the population as well, since a campaign aimed at first-time buyers should not be judged only by existing customers. Construct a control that resembles the exposed group except for the ad opportunity, because poor matching can create false lift.

Distinguish exposure from targeting, since being in a targeted audience is not proof an ad was actually seen and platforms define eligibility differently. Keep the survey wording stable across groups, as small changes in the question or answer options can move results, and avoid leading questions that reveal the advertised brand too clearly.

Choose follow-up timing for the goal, because surveying immediately after one impression may capture recall but not lasting preference. Calculate absolute lift and relative lift separately and say which one is reported.

An awareness rate of 38% versus 30% is an eight-percentage-point gap, while eight points over a 30% control rate is about 26.7% relative lift, not an eight-per-cent increase. A small sample can produce a noisy estimate, so review confidence intervals or the platform's statistical thresholds.

Watch for bias and confounding factors that can distort the comparison. People who answer surveys can differ from those who ignore them, and randomisation helps but may not solve every issue, while email, outdoor ads and earned publicity may affect both groups differently.

Compare exposed users with a proper holdout rather than ad clickers with non-clickers, which reflects prior interest rather than ad impact, and remember that a famous brand may have high baseline awareness and little room to move. Look beneath the average by separating creative variants and checking frequency, because a single average can hide a weak message and more exposures help only up to a point before fatigue sets in.

Segment cautiously and choose the primary metric, audience and timing before results arrive, since tiny subgroups are unreliable and hunting for the best-looking one overstates confidence. Intent is not purchase, a large lift among a tiny audience may have little market impact, and no detectable lift may mean the ad had little effect or the study lacked power, so inspect the sample and uncertainty before concluding anything.