Brand lift is the measured change in what people think about a brand after seeing its advertising. The same survey question goes to two groups drawn from one audience: a group served the ad, and a control group deliberately held back from it. The gap between the two sets of answers, reported in percentage points or as a proportion of the control group's starting level, is the lift.

The metric exists because the standard currencies of digital advertising describe the wrong outcome for brand campaigns. Clicks, viewable impressions and last-click conversions record what happened on a screen, not whether a viewer now remembers the advertiser or would put it on a shortlist. Brand lift borrows its design from clinical trials rather than from ad servers.

How a study is constructed

Randomisation comes first. Users eligible for the campaign are split into an exposed cell and a control cell before bidding begins, so the only systematic difference between them is whether the ad was allowed to reach them. Targeting, inventory and flight dates are held constant. That design is what separates lift from a correlation between exposure and attitude.

Survey delivery follows exposure. On YouTube, the questionnaire appears as a single question ahead of a video, served to both cells roughly a day after the ad was shown or withheld. Google Ads documentation lists ad recall, brand awareness and consideration as outcomes, reported alongside lifted users, cost per lifted user, absolute, headroom and relative brand lift, and the baseline and exposed positive response rates. Display & Video 360 widens the choice to association, awareness, consideration, favourability, purchase intent and video ad recall, capped at three metrics per study.

The three lift figures answer different questions. Absolute lift is the raw percentage-point gap between exposed and control. Relative lift divides that gap by the control group's baseline, which flatters small brands with little existing awareness. Headroom lift divides it by the distance between the baseline and a perfect score, the fairer reading for a brand already recognised by most of its category.

Sample size governs everything. Display & Video 360 documentation indicates that strong campaigns start producing readable results at roughly 2,000 responses per metric, and Google publishes budget floors to get there: over a ten-day window on YouTube and partner inventory, $5,000 to $20,000 for one to three questions in the cheapest country tier, rising to $15,000 to $60,000 in the most expensive. Cross-exchange connected TV studies carry no formal minimum, though $15,000 per question is recommended.

Confidence is reported rather than assumed. Google's certainty of lift metric is one minus the p-value, a frequentist read of how likely the observed gap would be if the advertising had done nothing. The platform targets 90%, publishes results between 50% and 90% as directional, and labels anything below 50% as no lift.

Who operates it

On the buy side, the study is configured by the advertiser or its agency inside the demand-side platform. Amazon opened a self-service workflow in May 2026 covering more than 50 third-party measurement products across 18 countries, with DISQO and Kantar supplying brand lift. The Trade Desk routes equivalent studies through its Measurement Marketplace, where from September 2, 2026 LoopMe set a floor of one million impressions against the three to four million demanded by rivals, matching control groups on more than 100 variables across 35 markets and refreshing reports every 24 hours.

On the sell side, the platform or publisher supplies both the media and the survey slot. That is the structural oddity of the category: the seller of the impression frequently fields the questionnaire that judges it. Independent panel vendors, among them Cint, Kantar, Dynata, DISQO and Upwave, exist largely to break that circle.

Origin and evolution

Survey-based effectiveness research predates programmatic by two decades. Dynamic Logic was founded in 1999 by Nick Nyhan and Ronit Aviv to standardise it, and WPP announced on June 20, 2005 that Millward Brown had acquired the firm. Nielsen bought Vizu on July 2, 2012, folding intercept surveys into what became Nielsen Brand Effect. An IAB report in 2010 had already flagged the weakness that still dogs the method, according to Digiday: too few studies used equivalent control groups.

Google entered on March 11, 2013, when product manager Paul McDonald announced Brand Lift surveys inside AdWords, powered by Google Consumer Surveys and requiring no extra tagging. The tool was rebuilt in October 2018, when a partnership with IRI brought surveys fielded continuously through a campaign rather than in an opening burst, plus lifted users, cost per lifted user and Maximize Lift bidding. That last feature was supported until March 31, 2022 before being dropped in favour of skippable in-stream formats, according to Adweek.

Independence and reach arrived next. Kantar moved its Brand Lift Insights product onto Google's Ads Data Hub on February 4, 2021, measuring YouTube without cookies or pixels in the United States and 15 pilot markets. TikTok introduced its own Brand Lift Studies in June 2021 with Kantar verifying and questionnaires styled to match the in-feed experience. In December 2023, Google extended cross-exchange Brand Lift Studies in Display & Video 360 to connected TV, letting one study span Google, YouTube and third-party publishers.

Why it matters to marketers

Brand lift is the number upper-funnel budgets are defended with, because awareness and consideration sit at the top of the marketing funnel, where performance dashboards show almost nothing.

Platform research leans on it. Amazon's analysis of more than 3,000 brand lift measurements taken across five markets over the eight quarters to the third quarter of 2024 reported that display combined with video produced 2.2 times the brand awareness of video alone, and that plans running upper and lower funnel together doubled brand lift. Those figures come from the seller of the media.

Meta placed randomised brand lift and conversion lift at the top of what it calls a ladder of incrementality in a framework published on May 28, 2025 and drawn from 307 studies across 54 advertisers. It also reported that the average advertiser runs 3.8 measurement solutions and that 55% get contradictory results, which is why brand lift is rarely read alone.

Limitations and disputes

Response rates are the oldest complaint. Digiday reported on July 2, 2014 that pop-up brand surveys drew responses from under 1% of those served, against roughly 9% for panel-based work, according to then InsightExpress co-chief executive Marc Ryan. Ben Winkler, then chief digital officer at OMD, said respondent demographics frequently failed to match the target audience even after professional survey takers were filtered out. Panel response rates have kept falling since.

Statistical power is the sharper constraint. A standard Google Brand Lift Study detects lifts of 2% or more; an Enhanced option released on July 20, 2026 detects lifts as low as 1.2% and raises the chance of detecting positive lift by 60%, at roughly three times the budget, according to Search Engine Land. Most campaigns shift brand metrics by low single-digit points, so a flat result frequently reflects an underpowered test rather than an ineffective campaign. The three-metric ceiling compounds this: adding the newer association metric means dropping another question.

Contamination undermines the control group. Holdout members still meet the brand on other channels, in stores and in search results, which narrows the gap a study can detect.

Stated answers are also not behaviour. A FreeWheel and MediaScience study of 436 participants published in September 2026 found unaided recall of 33% for a streaming environment against 12% for the same advertisement in a social feed, with heavier ad loads cutting recall from 41% to 26%. Effects of that size rarely surface in a platform-run study confined to one environment.

The newest dispute concerns whether surveys are needed at all. Upwave opened a closed beta of Predictive Brand Lift on September 9, 2026, using Bayesian modelling trained on measurements from more than 3,000 brands to estimate what a survey would have found. Chief executive Chris Kelly said constraints meant "advertisers could only measure a fraction of their campaigns". The trade is coverage against evidence: a modelled result carries no observation from the campaign it describes.

Not the same as

Conversion lift applies the identical randomised design to a behavioural outcome, counting purchases or sign-ups rather than survey answers.

Incrementality is the broader principle that only outcomes an advertisement caused should be credited to it. Brand lift is one application; Google cut the budget floor for its incrementality tests to $5,000 in 2025.

Holdout study names the experimental method rather than the metric. Brand lift studies are holdout studies whose dependent variable is a survey response.

Sales lift measures transactions instead of perception. Cint merged the two readings on June 17, 2026, pairing survey results with card transaction data from Affinity Solutions.

Attention metrics score the quality of an exposure, such as time in view. They describe the input to a brand effect, not the effect itself.

Recent developments

Measurement vendors spent 2026 moving studies into the campaign setup screen. illumin embedded Cint's brand lift into its programmatic workflow on May 14, 2026, claiming a 90% reduction in study launch time. Basis followed on June 16, 2026, wiring Cint's Lucid Measurement into its platform with automated tag creation and daily readings once response thresholds are met.

Programmatic access widened over the same period. Google Ads API v25.1, released on August 19, 2026, added read-only lift resources and 24 lift metrics segmented by age range, gender, device, campaign and video, all restricted to allowlisted accounts.

Timeline

  • 1999: Dynamic Logic founded by Nick Nyhan and Ronit Aviv to standardise online advertising effectiveness research
  • June 20, 2005: WPP announces Millward Brown's acquisition of Dynamic Logic
  • 2010: IAB report criticises the use of non-equivalent control groups in online brand surveys
  • July 2, 2012: Nielsen acquires Vizu, the intercept-survey business behind Nielsen Brand Effect
  • March 11, 2013: Google announces Brand Lift surveys in AdWords, powered by Google Consumer Surveys
  • July 2, 2014: Digiday documents sub-1% response rates for pop-up brand surveys
  • October 2018: Google partners with IRI, moves to continuous in-flight surveys, and adds lifted users, cost per lifted user and Maximize Lift bidding
  • February 4, 2021: Kantar moves Brand Lift Insights onto Google's Ads Data Hub for cookieless YouTube measurement
  • June 2021: TikTok launches Brand Lift Studies with Kantar as verification partner
  • March 31, 2022: Google ends support for Maximize Lift bidding on YouTube
  • December 2023: Google extends cross-exchange Brand Lift Studies in Display & Video 360 to connected TV
  • December 2024: Amazon publishes analysis of over 3,000 brand lift measurements across five markets
  • May 28, 2025: Meta publishes its incrementality framework ranking randomised lift tests above attribution
  • May 14, 2026: illumin embeds Cint brand lift into programmatic campaign setup
  • May 15, 2026: Amazon DSP opens self-service third-party measurement across 18 countries
  • June 16, 2026: Basis adds self-serve Cint brand lift studies to live campaigns
  • June 17, 2026: Cint combines brand lift and sales lift in a single in-flight dashboard
  • July 20, 2026: Google releases Enhanced Brand Lift Studies, detecting lifts as low as 1.2%
  • August 19, 2026: Google Ads API v25.1 exposes 24 lift metrics to allowlisted accounts
  • September 2, 2026: LoopMe brings brand lift studies to The Trade Desk at a one million impression floor
  • September 9, 2026: Upwave opens a closed beta of surveyless Predictive Brand Lift

Summary

Who: Advertisers and agencies commission brand lift studies; platforms including Google, Meta, Amazon, TikTok and LinkedIn run them on their own inventory, while independent vendors such as Cint, Kantar, Dynata, DISQO, LoopMe and Upwave field surveys across exchanges.

What: A randomised experiment that compares survey answers from an exposed group against a withheld control group to quantify how far advertising shifted metrics such as ad recall, awareness, consideration, favourability and purchase intent.

When: Survey-based effectiveness research dates to Dynamic Logic in 1999; Google brought it into AdWords on March 11, 2013, and rebuilt it around continuous in-flight surveys in October 2018. The current wave of in-platform and modelled variants arrived through 2026.

Where: Inside demand-side platforms and ad platforms, with surveys served in publisher inventory such as YouTube pre-roll, in-feed social placements and voluntary mobile ad breaks, and results delivered through platform dashboards, measurement marketplaces and reporting APIs.

Why: Clicks and conversions cannot show whether advertising changed what people think, which leaves upper-funnel budgets unaccounted for. Brand lift supplies a causal estimate of that change, though low response rates, limited statistical power, control group contamination and the conflict of a seller grading its own media all constrain what a single study proves.