What is Brand Lift?

Brand lift is the measured change in brand metrics among people who saw an ad versus those who did not. The object of measurement is usually a survey, not sales: whether the respondent recalls the ad, knows the brand, or would consider buying. The result is reported as lift — the gap between the exposed group and the control, in percentage points or as a percent of the control baseline.

The term spread with self-serve ad accounts and video. It is often confused with incrementality. Both compare “saw the ad” with “did not,” but brand lift almost always rests on a survey, while incrementality rests on a holdout and observed actions (order, install, deposit).

What the survey asks

The metric set depends on the platform and the brief; the core is stable:

Ad recall — remembering the ad (“seen an ad for brand X in the last N days”).
Aided / unaided awareness — brand knowledge with and without a prompt.
Consideration — willingness to include the brand in the choice set.
Favorability — attitude toward the brand.
Purchase intent — stated intention to buy.

Some studies add message association: whether the respondent linked the creative to the right brand. That is not a conversion and not ROAS. Consideration can rise in a survey while nobody buys.

How a study is built

Typical design: random split into exposed and control (or a survey sample drawn from both). The campaign runs; after a set window the platform or a research vendor shows a short questionnaire. Lift is the difference in “yes” shares (or mean scores) between groups.

Who counts as exposed matters. In an ad account that is usually an impression under platform rules — not always an MRC viewable impression. If the control still saw adjacent placements of the same brand, or branded search, lift shrinks. Sample size decides whether +2 pp is noise or an effect: weak campaigns often come back “not significant” even when the creative is not zero.

The report is not a continuous CPA dashboard. It is a slice over the study window, with a confidence interval. Comparing lift across two accounts without the same questions, window, and impression definition is meaningless.

VK, Direct, DV360, Meta

In practice a buyer meets brand lift as an in-platform product, not a Kantar field panel.

VK Ads — studies on VKontakte and related placements: post-impression surveys, typically recall, knowledge, consideration. Useful on CIS reach and video when the goal is not only a lead.
Yandex Direct — brand lift for media and video (including the Yandex ad network and video inventory): respondents from exposed and control. It does not replace a conversion counter in Metrica.
DV360 and YouTube — Brand Lift Studies in Google’s measurement stack: questions on a viewer sample, often with video and CTV. Impression and survey live in Google measurement, not in an offer postback.
Meta Ads — Brand Lift (surveys) is separate from Conversion Lift. Conversion Lift is closer to incrementality: a holdout and events. Brand Lift answers whether perception moved, not how many extra purchases the auction produced.

Third parties (Nielsen, Kantar, YouGov) run the same surveys outside the walled garden — slower and more expensive, but one questionnaire across channels.

Brand lift vs incrementality

	Brand lift	Incrementality
Evidence	Survey (attitudinal)	Behavior: orders, revenue, installs
Control	Did not see / not sampled as exposed	Holdout: the ad was deliberately withheld
Question	Did brand perception change?	Did extra conversions appear that would not have happened without media?
Typical use	Video, CTV, reach, upper funnel	Performance, where the fight is “ads vs organic”

Survey lift does not prove sales incrementality. A respondent may recall the brand because they are already a customer; the control may have bought with no ad. The reverse happens too: promo and retargeting produce measurable orders with zero consideration shift — demand was already there.

For performance marketing, brand lift is an extra layer, not a substitute for CPA or the LTV / CAC pair. For brand work the opposite holds: last-click CPA alone does not show whether the upper funnel moved.

How to read the numbers

Platforms report lift versus control (“+18% ad recall”) or in points (“+4.2 pp”). Those are different: +4 pp on a 20% base is +20% relative to control. Check significance: if the interval crosses zero, the “plus” is not established.

Do not compare brand lift with view-through: VTC is modeled attribution of an event after an impression, not a survey. Do not confuse lift with a rise in CTR: click-through is not brand recall.

Limits

A survey captures social desirability and recency of the impression, not the till. Samples are thin on narrow GEOs and small budgets; accounts often require an impression floor. A study “inside” Meta or YouTube does not see that the same user also saw search ads. The control is contaminated by competitor reach and offline.

Brand lift does not answer whether buying can scale on unit economics. That is LTV, CAC, and incrementality. Do not confuse it with the separate question of brand versus performance as a buying style: lift measures communication effect, not the payment model.