Thursday, October 18, 2007
Predictive Banners
During the last week I've had several conversations about behavioral and predictive banner serving. I've come to realize there are two specific areas or players that can attempt to deliver predictive banners: The ad network and the advertiser. The idea is difficult to execute yet based on the simple concept that past behavior trends across a critical mass of users can be used to predict future success.
The ad network is in the position to collect the broad data required to reach a critical mass of users. Since the ad network touches viewers across multiple sites and can track previous 'successes' for each viewer, the ad network is also in the position to deliver 'the next' banner from their stable of diverse advertisers. Therefore, the goal is to build a data processing system that can decide to serve, at run time, one banner over another. The problem, however, is not, in my opinion, the data system, although this is a big hurtle, it is rather the constant variable changes that occur between past successes and current presentations. Specifically, banner creative, context of publisher site, relationship between the brand and the viewer and the motivating factor for past success. And because these variables are more controlled from within a single advertiser, the advertiser may actually be in a better position to execute a predictive banner system.
I'll detail each of these areas in my next submissions and follow-up with a look at predictive banners from an Advertiser's standpoint.
1. How Banner Creative effects results
2. How viewers' "Viewing Goals" effect results
3. How "Viewer-Brand" relationship effect results
4. How the viewer's motivation or "Immediacy Factor" effect results
The ad network is in the position to collect the broad data required to reach a critical mass of users. Since the ad network touches viewers across multiple sites and can track previous 'successes' for each viewer, the ad network is also in the position to deliver 'the next' banner from their stable of diverse advertisers. Therefore, the goal is to build a data processing system that can decide to serve, at run time, one banner over another. The problem, however, is not, in my opinion, the data system, although this is a big hurtle, it is rather the constant variable changes that occur between past successes and current presentations. Specifically, banner creative, context of publisher site, relationship between the brand and the viewer and the motivating factor for past success. And because these variables are more controlled from within a single advertiser, the advertiser may actually be in a better position to execute a predictive banner system.
I'll detail each of these areas in my next submissions and follow-up with a look at predictive banners from an Advertiser's standpoint.
1. How Banner Creative effects results
2. How viewers' "Viewing Goals" effect results
3. How "Viewer-Brand" relationship effect results
4. How the viewer's motivation or "Immediacy Factor" effect results
Labels: advertising 2.0, Internet Secured Application Technology, ISAT, online advertising, Online Marketing, Performance Communications Group, predictive advertising
