Analyzing Political Advertisers’ Use of Facebook’s Targeting Features

The popularity of online advertising—now with aggregate revenues in the hundreds of billions of dollars each year— is strongly driven by targeting, or the ability of an advertising platform to help an advertiser select exactly which users should see their ad. To enable such targeting, advertising platforms routinely collect detailed data on users, a practice which has led to a raucous debate over privacy. Unfortunately, regulators and the public at large often have little visibility into how these advertising platforms implement their targeting features or how they are being used. Recent events, however, have caused platforms to start offering limited transparency for political ads. In this paper, we focus on Facebook and collect data from Facebook’s Political Ad Archive on all political ads from a large list of political advertisers. We then cross-reference this data set with the crowdsourced ProPublica Political Ad Database (which contains information about ads’ targeting parameters), allowing us to have ground truth data on advertising campaigns’ impressions, money spent, and use of targeting features. Analyzing the resulting data set, we find that well-funded advertisers tend to use privacy-sensitive targeting features more frequently and that less-well-funded advertisers tend to more narrowly target their audiences geographically. We make the resulting data sets available to the community to enable further study.