Abstract: In one embodiment, a selection process executing in an optimizing engine is invoked as users are presented with web pages that include ad content. The selection process provides an initial population of sets of characteristics that specify an initial generation of layouts of ads for the pages. The characteristics can include anything that affects an ads appearance on the display. For example, ad type, position within a web page, color, size, text font, animation, etc. can be specified. Statistics are obtained as to each ad or page layout's performance or impact on the user, such as revenue obtained from ads in the layout, how many user's click an ad, how long users view an ad, etc. Once a large enough sample of user impacts is obtained a second population of sets of characteristics is created to improve the overall performance of the initial population.

Abstract: In one embodiment, a selection process executing in an optimizing engine is invoked as users are presented with web pages that include ad content. The selection process provides an initial population of sets of characteristics that specify an initial generation of layouts of ads for the pages. The characteristics can include anything that affects an ads appearance on the display. For example, ad type, position within a web page, color, size, text font, animation, etc. can be specified. Statistics are obtained as to each ad or page layout's performance or impact on the user, such as revenue obtained from ads in the layout, how many user's click an ad, how long users view an ad, etc. Once a large enough sample of user impacts is obtained a second population of sets of characteristics is created to improve the overall performance of the initial population.