Targeted, Not Tracked: Client-Side Solutions for Privacy-Friendly Behavioral Advertising

The current discussion of potential Do Not Track regulation for online advertising is worrisome for the advertising industry, as it may signif cantly limit the capability for targeted advertising, a key revenue source for online content. The present discourse conates the behavior tracking and ad targeting processes, leading to the presumption that providing privacy must come at the cost of eliminating advertisers' targeting capability. This paper focuses on a family of methods that facilitate behavioral targeting while providing consumer privacy protections. This is achieved by di erentiating between client-side and server-side tracking. Client-side solutions provide for mechanisms and policies that address the privacy concerns over lack of user control over data while providing advertising platforms with the ability to target users. We compare and contrast several client-side methods along several dimensions of user privacy, adoption e ort, and trust. A novel client-side pro ling method is proposed that di ers from prior work in not requiring installation of additional software by the user and providing compatibility with existing ad serving infrastructure. Empirical evaluation of the method on large-scale real-world datasets demonstrates the potential for high targeting performance of client-side techniques. We hope that by considering such middle-ground approaches, the present debate will converge towards solutions that satisfy both advertisers' desire for targeting and users' desire for privacy.