AdPalette: an algorithm for customizing online advertisements on the fly

摘要

In this paper, we address customization and dynamic optimization of online advertisements. For online ads that attract click-throughs, we use click through rates to develop a methodology for customizing advertisements on the fly by changing content, copy, placement, animation and other attributes. We use techniques from optimization, conjoint analysis and genetic algorithms. Ads are reconstituted on the fly using graphic files for each level of each attribute, much like a painter would use a palette. We show that this approach improves response rates, reduces server storage requirements and improves ad efficiency.