This is the summary of an article by Glen L. Urban, et al. You can get the pdf of the behavioral targeting article here: Morphing Banner Advertising.
An online ad morphs whent it dynamically changes depending on a consumer segments’ preferences. These preferences are determined from the online behavior of these consumers. A dynamic program helps a website determine how the online ad should morph by exploring and exploiting; that is, the program explores by trying out various morphs and seeing which is most effective, and exploits by using ads that are based on recent anticipated results.
This paper is the first random, field testing for morphing, with a relative large sample size used to make more accurate observations regarding steady-state behavior.
Banner advertisements are paid advertisements located on a web page. Around twenty four percent of the revenue from online advertising is credited to banner advertisements, and during 2010, that’s around 6.2 billion US dollars. Therefore, a lot of businesses are interested in making banner advertisement even more effective.
Many firms try to improve banner ad effectiveness by using behavioral targeting, and increasing click through rates (CTR) in the process. However, most research show that behavioral targeting techniques today are quite static. There’s no automatic and dynamic learning involved among these strategies and they do not improve depending on how successful or bad the banners are. In other words, there is no morphing in behavioral targeting strategies.
Brief Review of Online Banner Morphing
Banner morphing exists because customized banner ads are very effective in catering to the needs of a given consumer segment. A study involving morphing banners needs to be able to accomplish two things: first, based on clickstream data from a given consumer, that consumer must be assigned to a certain segment. Second, there must be a method of assigning the best banner advertisements to each segment.
Segments are identified using a priming study, wherein consumers explore a priming website, then a model is developed to be able to assign a set of consumers to certain segments related to their clickstream data.
CNET Field Experiment
The study is done in cooperation with CNET.com, a website containing reviews and news regarding the latest high tech products. Everyday, 8 million people visit the website, and the site is worth 1.8 billion dollars in the market. Advertising revenue from CNET.com has a high reliance on banner ads, and some of them are behaviorally targeted, demanding premium prices.
In the study, eight banners regarding smart phones from AT&T are used. Visiting consumers are randomly assigned to control and test cells. The banners differ in terms of characteristics that appeal to different consumers having varying cognitive styles. The control cell had randomly assigned banners, while that of the test cells contain banners that morph based on the cognitive style of the consumer.
Results of the Banner Morphing Test
More than 450,000 banners were seen by over 116,000 consumers in this study. There were behaviorally targeted banners, particularly those that are located in pages where there are pictured, discussed, priced, compared and rated phones. Around 27 percent of the consumers saw behaviorally targeted banner advertisements.
Results show that morphing is highly effective in combination with behavioral targeting. Here, banner morphing improves substantially and significantly, and clickthrough rates almost double. The control cell, on the other hand, shows an insignificant lift for consumers and banners even when behavioral targeting was used.
In addition, there was no significant lift for banners that morphed but were not behaviorally targeted. Indeed, there is an interaction between behavioral targeting and morphing.
Implications and Future Directions
Online morphing is currently at its developing stages, and this technology is going to continue to improve website designs and banner ads and make them more effective.
Further scientific studies about banner morphing will also lead to further knowledge regarding designing and developing banners that morph according to consumers’ needs. Indeed, implementing morphing technologies to come up with the best banner for each consumer segment will lead to developments in the creativity aspect of banner ads.