Geo-Spatial Modeling of Online Ad Distributions

Abstract

The purpose of this document is to demonstrate how spatial models can be integrated into purchasing decisions for real-time bidding on advertising exchanges to improve ad selection and performance. Historical data makes it very apparent that some neighborhoods are much more interested in some ads than others. Similarly, some neighborhoods are also much more interested in some online domains than others, meaning viewing habits across domains are not equal. Basic data analysis shows that neighborhoods behave in predictable ways that can be exploited using observed performance information. This paper demonstrates how it is possible to use spatially correlated information to better optimize advertising resources.