Short answer: Ads Manager counts every conversion that happened after someone saw or clicked your ad within the attribution window. A conversion lift test counts only the conversions your ads caused, by comparing people who could see your ads with a random group who couldn't. Ads Manager will almost always show more. For budget decisions, believe the lift test, then use it to adjust how you read Ads Manager between tests.
Two different questions
- Ads Manager (attribution) answers: "Which conversions happened after someone interacted with our ad?"
- A lift test answers: "Which conversions would not have happened without our ads?"
Meta's training uses a simple example. John saw an ad and bought because of it. Pedro saw the same ad but would have bought anyway. Attribution credits both. A lift test credits only John.
Why are the numbers different?
1. Attribution counts people who would have bought anyway. Past customers, email subscribers and people searching your brand often convert at high rates whether or not they see an ad. Attribution counts them. Lift subtracts them through the control group.
2. Different time windows. Ads Manager uses an attribution window, like 7 day click and 1 day view. A lift test counts conversions during the test period and its own measurement window. See Meta attribution windows.
3. Different scope. A lift test measures every conversion in the test and control groups, including ones Ads Manager would credit to another channel or miss entirely, such as a view that led to a brand search days later. That means lift can sometimes find impact attribution missed.
4. Different counting. Ads Manager can credit one purchase to more than one ad or Ad Set. A lift test counts each person's conversions once.
5. Modeled conversions. Ads Manager includes modeled conversions to fill gaps from privacy limits. A lift test compares actual measured outcomes between two random groups.
Does lift always come in lower?
Usually, but not always. For upper funnel campaigns, a lift test can show impact Ads Manager missed, because people saw an ad, didn't click, and bought later through search or a store. Meta's incrementality training cites a study in which a share of conversions caused by Meta ads were credited to other channels.
Which one should you believe?
Believe the lift test for decisions about how much to spend and where. Keep Ads Manager for day to day optimization, because it's fast and granular. Connect the two with an incrementality factor:
Incrementality factor = incremental conversions from the lift test ÷ conversions Ads Manager reported for the same campaign and period
Multiply Ads Manager ROAS by that factor to get a closer estimate of true ROAS until your next test. The Kellogg researchers who studied thousands of Meta experiments found this kind of calibration makes everyday metrics far more useful than they are raw.
A worked example
Made up, rounded numbers.
| Ads Manager | Lift test | |
|---|---|---|
| Purchases | 900 | 300 incremental |
| Spend | $30,000 | $30,000 |
| Cost per purchase | $33 | $100 per incremental purchase |
| ROAS (at $120 AOV) | 3.6 | 1.2 iROAS |
Incrementality factor: 300 ÷ 900 = 0.33. Next month, when Ads Manager shows a 4.0 ROAS, your estimate of true ROAS is about 1.3.
Key takeaways
- Attribution counts conversions near your ads; lift counts conversions caused by them.
- Lift is usually lower, but can be higher for upper funnel campaigns.
- Budget on lift results; optimize day to day in Ads Manager.
- Use an incrementality factor to connect the two between tests.
- Retest regularly, because the factor changes.
FAQ
Is Ads Manager wrong? No, it's answering a different question. It just isn't the right number for deciding how much spend is worth it.
Why does GA4 show something different again? GA4 uses its own attribution rules and only sees site sessions. It's a third view, not the truth either.
How close should lift and Ads Manager be? There's no standard. Retargeting heavy accounts tend to show the biggest gap.
Can I just use Meta's Incremental Attribution setting instead of a test? It's useful, but it's a model, not an experiment. See Meta Incremental Attribution.
Want help reconciling your numbers? 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
- Kellogg Insight, "Here's a cost effective way to tell if your digital ads are working"
- Haus, "Understanding Meta incrementality testing."
Check what you learned
Tap an answer to see if you got it right.
1. What's the difference between lift and Ads Manager?
2. Is lift always lower than Ads Manager?
3. How do the two work together?
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 9 of 15
Next: Conversion Lift vs Brand Lift on Meta, Plus Single Cell vs Multi Cell Tests
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.