What’s the Deal With Facebook’s Modeled Conversions?
By Rebekah Rauckman
Facebook announced earlier this year that they would be modeling conversions for iOS14 users to give a more accurate read on how conversions actually look.
In Facebook’s words:
“Modeling uses aggregated data from different sources to account for conversions we’re unable to count directly due to various factors, including cross-site tracking limitations.
Because each user’s ad interactions can vary tremendously, depending on factors such as browser, device, demographics, and the nature of the ads themselves, our models are dynamic and adaptable.”
In practice what we have been seeing is a larger and larger discrepancy between where Facebook is placing conversions on the ad set and ad level and where Shopify/Google Analytics is saying these conversions came from.
- Ad Set 1: 10 sales in Facebook, 2 sales in GA
- Ad Set 2: 4 sales in Facebook, 14 sales in GA
The question becomes, which ad set is actually driving sales. Because Facebook is modeling their data, GA becomes the real source of truth on what is driving real sales vs estimated sales.
Recommendations for how to be get Facebook to behave how you want it to.
- Leave ads/ad sets on that have strong GA/Shopify sales even if Facebook isn’t reporting these as sales yet. Eventually Facebook will figure out these ads/ad sets are working.
- Launch less ads per ad set so there is less room for error with Facebook’s modeling.