Abstract

Advertising is one of the primary means for revenue generation for millions of websites and mobile apps. While the majority of online advertising revenues are based on pay-per-click, alternative forms such as impression-based display and video advertising have been growing rapidly. In this talk, we present two perspectives on measuring and analyzing online advertising and discuss our findings from both perspectives. The first is the is web publisher perspective, where we focus on the problem of fraudulent traffic that aims to inflate advertising impressions on websites. To assess this problem we deploy a set of honeypot sites, which provides a window into the basic mechanisms used for impression fraud and in particular enables us to identify pay-per-view (PPV) networks. Our results show that PPV networks deliver hundreds of millions of fraudulent impressions per day, resulting in hundreds of millions of lost advertising dollars annually. The second perspective is that of users, who are the targets for online ads. Our goal is to understand the features and dynamics of the mechanisms that are used to place ads on web sites. For this study, we develop a scalable crawling capability that enables us to broadly gather the details of ads on websites including creatives and landing pages. Of critical importance in this work is the recognition that a user’s profile can have a significant impact on ads that are shown. Using our profile-enabled system, we gathered ads from a large number of websites using diverse profiles. We report the details of the findings from measurements that highlight how and when targeting is used in online advertising.

Bio: Paul Barford is a professor of Computer Sciences at the University of Wisconsin – Madison. He was a visiting professor at the University of Cambridge and an EPSRC Visiting Fellow in 2011. His research interests are in networking, security and large data analysis.