Incremental attribution is a Meta Ads attribution setting that counts, and optimises towards, only the conversions Meta estimates would not have happened without your ad. Standard attribution credits the ad with every conversion that follows a click or a view within a set window, whether or not the ad made any difference.
How incremental attribution works
Under a standard setting such as 7-day click, 1-day view, Meta claims any purchase made within seven days of someone clicking your ad, or within a day of them seeing it. That includes people who were going to buy anyway: loyal customers, people who had already searched for your brand, people with an item sitting in their basket. A view-through conversion can be credited even when the person barely noticed the ad.
Incremental attribution tries to remove that background. Meta says its model draws on lift experiments, which compare people who were shown ads with similar people who were held back, to estimate the share of conversions the ads actually caused. When you choose the setting on an ad set, delivery is steered towards people whose behaviour the ads are likely to change, and reporting shows the estimated incremental conversions.
At the time of writing (October 2026), the setting is only offered for certain objectives and optimisation events, and Meta continues to change where it appears. The figures are modelled estimates, not a list of individual buyers you can check one by one.
Why it matters
Standard attribution flatters some campaigns more than others. Retargeting and campaigns aimed at existing customers reach people already close to buying, so they collect a lot of credit for sales that would have happened anyway. An online shop with a large returning customer base can see an impressive return on ad spend in Ads Manager while total sales barely move.
Incremental attribution gives a smaller but more honest number, and it pushes spend towards finding new demand. That is the question a business owner actually cares about: what did the money add? The idea comes from incrementality testing, which is the most reliable way to measure advertising, applied inside the platform.
There is a UK angle too. Under PECR, many visitors decline tracking cookies, so fewer conversions are observed directly and more depend on modelling and the Conversions API. Good server-side data gives any attribution model, incremental or standard, better material to work with.
Common mistakes
- Comparing incremental results with last month’s standard results and concluding performance fell. The numbers are measured differently and will be lower.
- Switching the setting on a live ad set. Treat it as a new test in a new ad set so before and after stay clean.
- Treating modelled estimates as exact counts.
- Trying it on an ad set with very few conversions, where the model has little to go on.
- Judging any attribution setting by platform figures alone, without checking sales in your own shop or CRM.
How to act on it
If you spend a meaningful amount on Meta and suspect your retargeting or brand campaigns are taking credit for sales they did not drive, test incremental attribution in a separate ad set alongside your existing one. Run both long enough to gather a fair number of conversions.
Then step outside Ads Manager. Compare total sales or qualified enquiries in your own systems over the same period, and look at new customers separately from returning ones. If the incremental ad set brings in more new customers at an acceptable cost, it deserves more budget even if its reported figures look modest.
Choosing attribution settings and reconciling them with real sales is part of the Facebook ads management I provide.