Short answer: A Meta conversion lift test compares a test group that can see your ads with a control group that can't. Incremental conversions are the extra conversions the test group had. Lift percent is how much higher the test group converted than it would have without ads. Cost per incremental conversion is spend divided by incremental conversions. Incremental ROAS (iROAS) is incremental revenue divided by spend. Confidence tells you how sure the test is that the lift is real.
What are the test group and control group?
Meta randomly splits the people your campaign could reach into two groups:
- Test group: people who can see your ads.
- Control group (holdout): people Meta keeps from seeing your ads during the test.
Because the split is random, the two groups are alike in every way except ad exposure. Any difference in conversions comes from the ads. Meta's own training calls randomized experiments "the gold standard of measuring incrementality."
The control group is usually smaller than the test group, so Meta scales its results up to the test group's size before comparing. When you see "control conversions" in results, it is usually this scaled number.
Incremental conversions
The conversions your ads caused.
Incremental conversions = test group conversions minus scaled control group conversions
If the test group had 1,000 purchases and the scaled control group had 700, your ads caused 300 purchases. The other 700 would have happened anyway.
Conversion lift percent
How much higher conversions were because of ads, compared with what would have happened without them.
Lift % = incremental conversions ÷ scaled control conversions × 100
Using the same numbers: 300 ÷ 700 × 100 = about 43% lift.
Cost per incremental conversion
What each caused conversion really cost. Some tools call it cost per incremental acquisition (CPIA) or cost per conversion lift.
Cost per incremental conversion = spend ÷ incremental conversions
If you spent $30,000: $30,000 ÷ 300 = $100 per incremental purchase. Ads Manager might have shown $30 per purchase on the same spend. The gap is the money spent on sales you'd have gotten anyway.
Incremental ROAS (iROAS)
Return on ad spend counting only caused revenue.
iROAS = incremental revenue ÷ spend
If the 300 incremental purchases averaged $120, incremental revenue is $36,000, so iROAS is 1.2. Reported ROAS might have said 4.0.
Incrementality factor
The share of Meta's reported conversions that were truly incremental. Haus and other measurement teams use it to calibrate everyday reporting.
Incrementality factor = incremental conversions ÷ conversions Meta reported
If Meta reported 900 purchases and 300 were incremental, the factor is 0.33. Multiply your normal Ads Manager ROAS by 0.33 to estimate true ROAS until your next test.
Confidence
How sure the test is that the lift is real and not random noise. Meta shows this as a confidence or probability figure, and some tools show a confidence interval (a range the true lift probably falls in).
Low confidence doesn't mean "no lift." It usually means the test didn't have enough conversions. Run it longer, with a bigger holdout or more spend, before you act.
A worked example, all together
Made up, rounded numbers for a 4 week test.
| Metric | Value |
|---|---|
| Spend | $30,000 |
| Conversions Meta reported | 900 |
| Test group conversions | 1,000 |
| Scaled control conversions | 700 |
| Incremental conversions | 300 |
| Lift % | 43% |
| Cost per incremental conversion | $100 |
| Incremental revenue (at $120 AOV) | $36,000 |
| iROAS | 1.2 |
| Incrementality factor | 0.33 |
Which metric should change your spend?
Cost per incremental conversion and iROAS. Compare them with your margin. If you need a 2.0 ROAS to break even and your iROAS is 1.2, the campaign loses money on every caused sale, no matter what Ads Manager says. Then see what to change after a lift test.
Key takeaways
- Incremental conversions = test minus scaled control.
- Lift % compares caused conversions with what would have happened anyway.
- Cost per incremental conversion and iROAS are the numbers to budget on.
- The incrementality factor lets you adjust everyday reporting between tests.
- Low confidence means not enough data, not proof of no lift.
FAQ
What is a good conversion lift percent? There's no universal number. A campaign aimed at new customers usually shows more lift than one reaching past buyers. Judge it on cost per incremental conversion against your margin.
Why is my cost per incremental conversion so much higher than my CPA? Because CPA counts every credited conversion, including ones that would have happened anyway. The gap is your non incremental spend.
What does a negative lift mean? Usually noise in a small test. Check confidence before acting.
Is iROAS the same as ROAS? No. ROAS counts all credited revenue. iROAS counts only revenue the ads caused.
Want help reading a lift test? We do a free Meta audit: a full account audit plus net new reach and cumulative reach reports. Reach out on our contact page.
Sources
- Meta, "Measure What Matters with Incrementality" practitioner deck
- Haus, "Understanding Meta incrementality testing"
- Triple Whale, "Meta Conversion Lift Tests."
Check what you learned
Tap an answer to see if you got it right.
1. How do you calculate incremental conversions?
2. Which numbers should you budget on?
3. Low confidence on a lift test means what?
Ready for the full test? Take the Measurement exam and get certified. Want this run on your account? Get my free audit.
Module 3: Measurement · article 8 of 15
Previous: Spending $300K+ a Month on Meta? You're Probably Wasting 25% of It
Next: Why Your Meta Lift Test Results Don't Match Ads Manager (and Which One to Believe)
About the author
Jake Mehani, founder of Creative Analytics
Jake has spent 15 years running Meta and paid social ads for consumer brands. Meta published his RBX Active Advantage+ case study, he taught the DMAnc Meta and TikTok ads masterclass, and he spoke at Motion's 2024 Creative Strategy Summit. His senior team runs media buying, creative and measurement like an in house team.