A conversion lift study measures whether your ads cause sales, rather than just being present when sales happen. Meta randomly divides the people your campaign could reach into a test group, who can see your ads, and a control group (the holdout), who can't. After the test, Meta compares conversion rates. The extra conversions in the test group are the incrementality of your ads.
Why it matters
Attribution, including view-through conversions, tells you which ads were near a purchase. It can't tell you which purchases would have happened anyway. A lift study can, because the control group shows what happens without ads. It is the closest thing to a controlled experiment most advertisers can run on Meta.
How a lift study works
- Choose what to test. A campaign, a group of campaigns, or all your Meta advertising.
- Choose the outcome. Usually purchases measured by the pixel and Conversions API.
- Set the holdout. A share of the audience, often around 10% to 20%, is kept from seeing ads.
- Run it. Campaigns run as normal for the test period, often two to four weeks.
- Read the results. Meta reports incremental conversions, lift percentage and a confidence level.
In Ads Manager, lift studies sit in the Experiments area. Availability, minimum spend and set-up options vary by account, and some studies are run with help from a Meta representative.
Worked example
Example: a four-week study.
| Test group | Control group | |
|---|---|---|
| People | 900,000 | 100,000 |
| Purchases | 1,350 | 120 |
| Conversion rate | 0.15% | 0.12% |
Scale the control to the test size: 120 × 9 = 1,080 purchases expected without ads. Incremental purchases = 1,350 − 1,080 = 270. Lift = 270 ÷ 1,080 = 25%.
If Meta's attribution claimed 800 purchases for the same campaigns, the study suggests only about a third of them were caused by the ads. Ad spend ÷ 270 gives the cost per incremental purchase, the true price of growth.
Lift study vs A/B test
| Conversion lift study | A/B test | |
|---|---|---|
| Question | Do my ads cause sales? | Which version performs better? |
| Control group | People who see no ads | None, every group sees an ad |
| Output | Incremental conversions | A winning variant |
What makes a study trustworthy
- Enough conversions. Small accounts may not get a clear result. Check the confidence level before acting.
- No big changes mid-test. Avoid launching sales or major budget swings in the middle.
- Clean tracking. Lift is measured through your pixel and Conversions API, so tracking gaps affect both groups but still add noise.
Fly ROAS's analytics compares Meta-attributed results with your store's totals, which helps you decide where a lift study is worth running. For how lift fits with attribution settings, read Meta ads attribution settings.
Frequently asked questions
What is a conversion lift study on Meta?
It is an experiment where Meta randomly holds back a share of your audience from seeing ads, then compares their conversions with the group that could see ads. The difference shows how many conversions your ads actually caused.
How long should a conversion lift test run?
Long enough to collect a meaningful number of conversions in both groups, often two to four weeks. Smaller accounts need longer. Avoid big changes, such as sales or large budget jumps, during the test.
What is the difference between lift and ROAS?
ROAS uses attributed revenue, which includes some sales that would have happened anyway. Lift measures only the extra conversions caused by ads. Incremental ROAS, using lift-based revenue, is usually lower than platform ROAS.
Can small brands run conversion lift studies?
Availability and minimum requirements vary by account, and small accounts may struggle to get a statistically clear result. A simpler alternative is a geographic holdout, pausing ads in some regions and comparing sales, read with care.
