Lift Study

A lift study is a randomized experiment that measures how much an advertising campaign changed a specific outcome, by comparing an exposed group against a control group that did not see the ads. The outcome can be a brand perception metric such as ad recall, or a business metric such as purchases. The gap between the two groups is the lift, and it represents the change the campaign caused rather than the change that merely followed it.

Why Lift Studies Matter

Platform conversion reports and brand surveys both measure results without proving cause. A campaign report showing 5,000 conversions includes buyers who would have purchased anyway. A brand tracker showing awareness rising from 20% to 24% cannot separate the campaign from a PR cycle, a competitor exit, or seasonality.

A lift study removes that ambiguity. Because the holdout is randomized, the only systematic difference between the groups is ad exposure.

The practical payoff is budget defensibility. Upper-funnel video and display spend rarely produces trackable clicks, so it loses budget arguments against last-click channels by default. A brand lift study gives that spend a measured number to argue with.

Types of Lift Studies

Lift studies are named after the outcome they measure, not the statistical method behind them.

• Brand lift study. Measures perception through survey questions served to exposed and control users. Standard metrics are ad recall, brand awareness, consideration, favorability, and purchase intent. Google Brand Lift and Meta’s poll-based brand lift both use this design.
• Conversion lift study. Measures purchases, signups, or other tracked events. Meta Conversion Lift and Google Ads Conversion Lift run this natively on a user-level holdout.
• Search lift study. Measures the increase in organic searches for the brand or product among exposed users. Google reports this alongside Brand Lift for YouTube campaigns.
• Sales lift study. Measures offline or retail sales, usually matched against loyalty card or point-of-sale panel data. Common in CPG, where the purchase never touches the advertiser’s site.
• Foot traffic lift study. Measures store visits from location data. Used by retail, QSR, and automotive advertisers.
• App install lift study. Measures installs and post-install events against a holdout, since install attribution windows overstate paid contribution.

How a Lift Study Works

1. Define one outcome and one campaign. A study measures a single question. Bundling three campaigns produces a number nobody can act on.
2. Randomize the audience. The platform splits eligible users into exposed and control before delivery starts. Splitting by geography, device, or audience segment instead of at random introduces differences that read as lift.
3. Suppress ads for the control group. Control users stay eligible to win the auction but see no campaign ad.
4. Collect the outcome identically for both groups. Brand studies serve the same survey question to both arms. Conversion studies read the same pixel or conversions API events for both arms.
5. Run to the planned end date. Two to four weeks is typical. Stopping at the first positive reading inflates false positives.
6. Read lift with its confidence interval. Platforms report a confidence level alongside the result. A study returning “no significant lift” has failed to detect an effect, which is not proof that none exists.

The design is one application of incrementality testing, which covers the broader family of holdout, geo, and on-off experiments.

How to Calculate Lift

Lift is reported in two forms, and confusing them overstates results.

Absolute lift = Exposed positive response rate - Control positive response rate

Relative lift = Absolute lift / Control positive response rate

Lifted users = Absolute lift x Exposed users

Cost per lifted user = Campaign cost / Lifted users

A worked example. Ad recall reaches 22% in the exposed group and 14% in the control group. Absolute lift is 8 percentage points. Relative lift is 57%. Across 500,000 exposed users, that is 40,000 lifted users, and a $20,000 campaign gives a cost per lifted user of $0.50.

Relative lift inflates on small baselines. A move from 1% to 2% is a 100% relative lift and a 1 point absolute lift. Report both, and compare cost per lifted user across campaigns rather than lift percentages.

Lift Study Requirements

Most lift studies are gated behind spend and volume thresholds, because small samples cannot detect small effects.

• Minimum budget. Google’s Brand Lift documentation sets a minimum spend per survey question over the study period, historically USD 5,000, and recommends at least 10 days of runtime. Meta Conversion Lift and Brand Lift studies are set up through a Meta representative and carry their own minimums.
• Sufficient conversion volume. Conversion lift needs enough events in both arms. Run a power calculation before launch to find the minimum detectable effect at your baseline.
• A real holdout cost. Control groups usually run 10% to 20% of the audience. That share of revenue is deliberately forgone for the study duration.
• Consistent measurement setup. Both arms need identical event definitions and campaign tagging, or the comparison breaks. Inconsistent UTM conventions between arms are a common cause of unreadable results.

TikTok, Amazon, Snap, and LinkedIn all offer managed brand or conversion lift products with similar gating.

Frequently Asked Questions

What is a lift study in marketing?

A lift study in marketing is a controlled experiment that isolates the effect of a campaign by withholding it from a random control group. Both groups are measured on the same outcome, and the difference is the lift. It is the standard method for valuing channels that generate few clicks, such as YouTube, connected TV, and display.

What is a brand lift study?

A brand lift study measures the change in brand perception an ad campaign caused, using survey questions served to both exposed and control users. Typical questions cover ad recall, awareness, consideration, and purchase intent. Google runs these as single-question surveys on YouTube and Display.

What is the difference between a brand lift study and a conversion lift study?

A brand lift study measures perception through surveys, and a conversion lift study measures tracked actions such as purchases or signups. Brand lift shows whether the campaign registered with people. Conversion lift shows whether it moved revenue. Running both on one campaign is common, since upper-funnel video often lifts recall well before it lifts sales.

Is a lift study the same as an A/B test?

No. An A/B test compares two versions of an experience, and both groups receive marketing. A lift study compares marketing against no marketing, so the absence of the ad is the variable being tested. That difference is why a lift study carries a real revenue cost and an A/B test usually does not.

How long should a lift study run?

Two to four weeks fits most digital lift studies, and Google recommends a minimum of 10 days for Brand Lift. Statistical power sets the real floor, not the calendar. Categories with long consideration cycles need a window that covers the full purchase path, otherwise conversions land after measurement closes.

To keep campaign delivery verifiable across both arms of a study, track your tagged links with linkutm’s link analytics.