Short answer: Incremental Attribution is a setting in Meta Ads Manager that reports and optimizes for conversions Meta predicts your ads caused, instead of every conversion that happened after an ad. It's built on what Meta learned from years of lift experiments, and it can push delivery toward people who wouldn't have bought otherwise. But it's a model, not an experiment, and it's Meta measuring Meta. Use it as a directional signal and keep running real conversion lift tests to check it.
What is Incremental Attribution?
Standard attribution counts conversions that happened within a window after someone clicked or viewed your ad. Incremental Attribution tries to count only the conversions your ads caused. Meta's practitioner training describes it as "a setting in Ads Manager to enable optimization and reporting based on incremental conversions," using machine learning "to predict and optimize for incrementality in real time, without the need to wait for experimental results."
It's tied to Meta's research on predicting incrementality from experiments. Researchers from Meta and Kellogg showed that a model trained on thousands of randomized tests could predict incremental results far better than 7 day last click attribution.
How does it change optimization?
When a campaign optimizes for incremental conversions, Meta tries to deliver ads to people who are less likely to buy without them, instead of people who would have bought anyway. Meta's training cites a 46% lift in performance when campaigns used incremental attribution, based on 37 conversion lift studies. That is Meta's own figure, from studies where lift was detected, so treat it as a best case.
What do independent tests show?
Seer Interactive tested the setting across 6 accounts and about $1.05 million in spend. Meta's numbers said 87% of conversions were incremental. GA4, using its own method, put it at 67%. Broad targeting and narrow retargeting showed the lowest incremental performance; refined mid funnel audiences showed the clearest impact. Seer's advice: treat it as directional and cross check it.
Should you trust it?
Use it, but don't let it replace a test. Three reasons:
- It's a prediction. A lift test measures what happened with a real holdout. The setting estimates it.
- It's self graded. Meta is reporting on Meta.
- Your account isn't average. Models learn from many advertisers. A lift test tells you about yours.
Our view stays the same: you can't believe any single number. Turn on Incremental Attribution as a column to compare, test it as an optimization setting on part of your spend, and confirm with a conversion lift study.
How to use it alongside lift tests
- Add incremental conversions as a reporting column next to your standard attribution, and watch the gap by campaign.
- Test it as an optimization setting in one campaign, ideally inside a multi cell lift test against your standard setup. See conversion lift vs brand lift, and single vs multi cell.
- Compare it to your lift results. If the setting says 87% incremental and your lift test says 40%, trust the test.
- Expect fewer reported conversions. That's the point: it stops counting sales that would have happened anyway.
Setting names and locations in Ads Manager change, so check Meta's current help pages for where to find it.
Key takeaways
- Incremental Attribution reports and optimizes for predicted caused conversions.
- It's a real step forward from last click, but it's still a model.
- Independent testing found Meta's incremental share much higher than GA4's.
- Use it as a column and a test, not as the final answer.
- Keep running conversion lift tests to check it.
FAQ
Is Incremental Attribution the same as a conversion lift test? No. A lift test uses a real holdout group. Incremental Attribution predicts incrementality with a model.
Will turning it on lower my reported results? Often, yes. It counts fewer conversions because it leaves out the ones it thinks would have happened anyway.
Should every campaign use it? Test it first on part of your spend and compare results with a lift test.
Where does the 46% figure come from? Meta's practitioner deck, based on 37 lift studies run between July and October 2024. It's Meta's figure.
Want help testing Incremental Attribution properly? 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
- Seer Interactive, "We tested Meta's new Incremental Attribution setting on $1M in ad spend" (7/30/25)
- Gordon, Moakler and Zettelmeyer, "Predicted Incrementality by Experimentation (PIE) for Ad Measurement," arXiv 2304.06828
Check what you learned
Tap an answer to see if you got it right.
1. What does Meta's Incremental Attribution report?
2. Is it the final answer?
3. How should you check it?
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 12 of 15
Previous: After the Lift Test: What a Great Media Buyer Changes in Your Meta Account
Next: Creative Metrics Formulas: Stop Rate, Hold Rate, Thumb Stop Click Rate and More
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.