Advertising · Large-scale randomized advertising experiment
Estimating the Value of Offsite Tracking Data to Advertisers: Evidence from Meta
Marketing Science · 2025
Summary
The median cost per additional customer was estimated at $38.16 at baseline and $49.93 under the median loss in effectiveness when offsite data could not be used, a 31% increase.
For your marketing
Possible founder application
Useful for thinking about privacy-related measurement loss and incrementality, especially for small advertisers. The Meta campaign economics and targeting system are platform-specific and should not be treated as a universal CAC benchmark.
Reported result
What the study found
The median cost per incremental customer was estimated at $38.16 at baseline and $49.93 under the median loss in effectiveness when offsite data could not be used, a 31% increase. The estimated loss was larger for small advertisers; purchase-optimized ads generated more long-term customers per dollar than click-optimized ads.
Research question
What the authors investigated
How much does offsite tracking data improve ad delivery effectiveness for advertisers?
Design
Randomized experiment across live campaigns, with deconvolution methods used to estimate distributions of campaign-level effects.
Sample and setting
More than 70,000 Facebook and Instagram advertiser experiments across over 160 countries.
Scope and transfer
Limitations to keep in view
The experiment is partial-equilibrium and does not measure broader market adjustment or consumer privacy welfare. Offsite purchases and opted-out users are not fully observed, so some effects may be missed.
Research sources
INFORMS publisher abstract retrieved with Firecrawl; reported experiment scope, sample, results and limitations are summarized at abstract level.