Conversion lift tests are Meta experiments that measure how many extra conversions your ads caused, by comparing people who could see the ads with a similar group who could not. They answer a question Ads Manager can't: how many of the sales it credits would have happened anyway.
How a lift test splits the audience
Meta's Experiments page says conversion lift "enables you to measure the incremental performance of your Meta ads." It creates two audiences: a test group that has the opportunity to see your ads, and a control group that does not. Meta also says there is no additional cost to create a test.
Because the groups are formed up front, the only planned difference between them is the ads. Switching a campaign off for a week is a weaker test, since seasonality and promotions change at the same time.
Let Madgicx's AI do the work
Its AI marketer finds what's working and prepares the next move. You approve.
Try for Free ($0 Trial)How it differs from attribution reports
An attribution model decides which ad gets credit for a conversion that already happened. It can't tell whether the person would have bought without the ad. A lift test measures that directly, which is the idea behind incrementality.
| Attribution report | Conversion lift test | |
|---|---|---|
| Question | Which ad gets credit? | Did the ads cause the sale? |
| Comparison group | None | A control group that doesn't see the ads |
| Timing | Every day, on every campaign | A planned test, for a set period |
| Result | Credited conversions | Incremental conversions with a range |
What you need before you run one
- Conversions Meta can see. A lift test can only count conversions that reach Meta, so check conversion tracking first. Sending the same events from your server through the Conversions API as well as the browser helps.
- One clear question. For example: do prospecting ads add purchases?
- Eligibility for your account. Where the test appears and who can run it can change, so look for it in Experiments in your account.
How to read the result
Meta's Lift Metrics Glossary defines the numbers. These are the ones to look at first:
| Metric | What it means |
|---|---|
| Incremental conversions | Additional conversions that would not have occurred without the ads being tested |
| Conversion lift | The gap between test and control conversions, adjusted for the difference in group size |
| 90% credible interval | A lower and an upper bound around the incremental conversions estimate |
| Confidence | The probability that the ads generated more than zero incremental conversions |
| Incrementality share | Incremental conversions divided by conversions from the exposed accounts |
| Cost per incremental conversion | Amount spent in the test divided by estimated incremental conversions |
Cost per incremental conversion = ad spend ÷ incremental conversions
Read the range, not just the middle. A wide interval means the test can't separate a strong effect from a weak one. More conversions narrow it. Then compare cost per incremental conversion with the number you can afford, such as your break-even cost per purchase.
Example. A skincare brand spends $20,000 on a prospecting campaign in a lift test. Among the accounts the ads reached, 1,600 convert. Meta estimates 800 of them are incremental, with a 90% interval of 560 to 1,040. That is an incrementality share of 800 ÷ 1,600 = 50%, and a cost per incremental conversion of $20,000 ÷ 800 = $25 (about $19 to $36 across the interval). The average order brings $30 of gross profit, so even the high end of that range loses little and the middle of it pays. The brand keeps the campaign.
Common mistakes
- Testing with too few conversions. A campaign with a handful of sales a week produces an interval too wide to act on.
- Changing things mid-test. A new offer or a budget jump muddies the comparison.
- Reading credited results as lift. The incremental figure is usually smaller than the conversions Ads Manager credits.
- Running one when nothing would change. If you would keep the budget whatever the answer, skip the test. A lift test earns its place when a decision hangs on it, such as moving budget between prospecting and retargeting. For everyday optimization, see how to optimize Facebook ads for conversions.
In Madgicx. A lift test can only measure the conversions Meta receives. Madgicx Cloud Tracking sends conversions to Meta from the server, so Meta sees more of them. Copilot, the AI chat in Madgicx, checks dataset quality and Event Match Quality in your account.
FAQ
What is a conversion lift test?
It is a Meta experiment that splits people into a test group that can see your ads and a control group that can't. The difference in conversions between the groups is the lift.
What is conversion lift?
It is the extra conversions your ads caused. Meta calculates it from the difference between test and control conversions, adjusted for the difference in group size.
How is conversion lift different from attribution?
Attribution decides which ad gets credit for a conversion. Lift tests whether the ads caused it at all. Attribution runs every day. A lift test is a planned check.
What does the confidence number mean in a lift test?
Meta defines confidence as the probability that the ads generated incremental conversions greater than zero. It is not the share of conversions the ads caused. That is the incrementality share.
Does a conversion lift test cost extra?
Meta says there is no additional cost to create a test. The ad spend in the test is what the cost per incremental conversion is built on.
Related
- Related terms: Incrementality, Brand lift, Attribution model
- In depth: How to optimize Facebook ads for conversions
Sources
- Facebook Lift Metrics Glossary, Meta Business Help Center
- About Experiments, Meta Business Help Center
