Conversion lift studies
A conversion lift study randomizes users into a group eligible to see ads and a control group held out from them, then compares conversion rates between the two. The difference estimates incremental conversions — those caused by the ads rather than ones that would have occurred anyway. Major ad platforms offer lift studies as a counterfactual check on attributed conversion counts.
What this means
In a conversion lift study, the platform randomly assigns users to a test group (eligible to receive the campaign) and a control group (held out). Both groups are then observed for the conversion of interest. Because assignment is random, the only systematic difference between groups is exposure, so the difference in conversion rate estimates the campaign's causal effect.
This is fundamentally different from attribution, which distributes credit across recorded touches. Lift answers 'how many extra conversions did the ads cause', while attribution answers 'which touches do we credit for the conversions we saw'.
How platforms run them
Google Ads, Meta, and other walled gardens offer lift study products that manage the holdout internally — they can withhold ads from a randomized control even within their own inventory (see ghost ads and PSA control designs). Results report incremental conversions and incremental cost per conversion over the test window.
Caveats matter: lift studies need sufficient scale to detect an effect, run for a fixed window, and report aggregate results rather than per-user credit. A lift number and an attributed number measure different things and should not be summed or directly equated.
- Randomized exposed vs held-out control groups
- Measures incremental conversions, typically below attributed counts
- Offered as managed products inside major ad platforms
How it appears in analytics and logs
A positive lift means the exposed group converted more than the held-out control; the gap is the incremental conversions, which is usually smaller than the attributed count.
Diagnostic use case
Use a conversion lift study to find out how many of a campaign's attributed conversions were actually caused by it, rather than coincident with it.
What WebmasterID can help detect
WebmasterID's first-party conversion events can serve as the outcome signal a lift study compares across exposed and control groups.
Common mistakes
- Equating incremental lift with attributed conversion totals.
- Running a study without enough scale to detect an effect.
- Treating a single study window as a permanent constant.
Privacy and accuracy notes
Lift studies compare aggregate group conversion rates rather than re-identifying individuals. This page is educational and not statistical or legal advice.
Related pages
- Incrementality testing: what attribution cannot tell you
Incrementality testing measures the lift a channel actually causes by withholding it from a control group and comparing outcomes. It answers the question every attribution model dodges: would this conversion have happened anyway? It is causal where attribution is merely correlational, but it requires deliberate experiment design.
- Ghost ads methodology
Ghost ads are an experimental design for measuring ad effectiveness. Rather than showing a placebo ad to the control group, the system records which control users would have been served the test ad had they been in the treatment group, then compares only comparable users. This isolates the ad's incremental effect while avoiding the cost and bias of serving placebo creative.
- PSA control group testing
A PSA (public-service announcement) control test is an incrementality design where the control group is served unrelated placebo ads instead of the test campaign. Because both groups receive an ad impression, exposure conditions are similar, and the difference in conversions estimates the test campaign's incremental effect. It is an older alternative to ghost ads.
- Attribution analytics
Compare lift results against attributed conversions.
Sources and verification notes
- Google Ads Help — About Conversion liftDocuments randomized test/control conversion lift methodology.
Last reviewed 2026-06-24. Facts are checked against primary/official sources where available; uncertain specifics are marked “Data not yet verified” rather than guessed.