Optimizing Efficiency On Ad Delivery

Abstract

Digital advertising has grown into one of the biggest budget items however, the process and determination of serving ads has been obscured from much of the public. This means that we cannot know why and how user data is being used in the advertising revenue model. This thesis aims to expose some of the obfuscation in the digital advertising space and present ways to recreate and optimize advertising heuristics to lower the cost of advertising in aspects such as network bandwidth. The task comes down to a lot of data as much of advertising efficiency involves matching and analyzing user data. As such, we focus on both the process of how data is acquired to give a background on what kind of data we have as well as the data analysis and prediction models.