Platform Lift Studies

Platform lift studies are controlled experiments built into ad platforms such as Meta, Google, TikTok, and LinkedIn. The platform randomly splits your audience into a group that sees your ads and a group that does not, then compares outcomes. Attribution assigns credit after the fact. A lift study answers a direct question: did seeing this ad change behavior? For most teams it is the easiest way to measure incrementality.

How it works

A lift study is a randomized controlled trial run inside the platform. The test group sees your ad. The control group sees nothing or a placeholder ad. After the campaign, the platform compares an outcome between the groups: awareness (brand lift), searches (search lift), or conversions (conversion lift). Because assignment is random, other factors balance out. That makes lift studies more reliable than last-click or multi-touch attribution for judging whether a campaign worked.

How to do it

Choose the study type. Use brand lift for awareness campaigns (recall, favorability, consideration). Use conversion lift for performance campaigns (purchases, signups, leads).
Check the platform’s minimum spend, duration, and audience size, and budget to meet them.
Set up the study in the platform’s measurement tools before launch. Most studies cannot be added afterward.
Run the full study period without major changes to targeting, creative, or budget.
Compare the outcome between test and control. Calculate cost per incremental conversion and compare it to your targets.

What to measure

Incremental lift: percent increase in the outcome caused by seeing the ad.
Cost per incremental result: spend divided by incremental outcomes. Expect it to be higher than cost per attributed result.
Statistical significance: the confidence that the lift is real, as reported by the platform.

Best practices

Start with your largest campaigns, where wrong measurement costs the most.
Compare lift results to the platform’s attributed results. The gap shows how much attribution overstates.
Use the gap to adjust ongoing reports. If attribution shows twice the conversions the study confirms, discount attributed numbers by half.

Common pitfalls

Stopping early because early numbers look clear. Underpowered studies give unreliable results.
Changing creative or targeting mid-study, which muddies what was measured.
Running studies on small audiences or low budgets where the expected lift is too small to detect.