by Head of Search
Andrew Girdwood
In August 2006 Digg was used as a case study to examine Microsoft's adCenter technology. AdCenter is Microsoft's shiny new pay-per-click bidding technology, their attempt to take on Google and has the demographic targeting edge over Google's current AdSense. The adCenter labs are used to showcase the technology and whet our appetite for more. Andrew Girdwood
Back in August Microsoft claimed that less than %10 of Digg's visitors where in the under 18 age bracket. Their prediction program forecast that the under 18 population on the site would rise dramatically to over %20. This would double the "credit card-less" audience of the site. The biggest reductions would be in the 18~24 and 24~34 age brackets - typically where many marketers try and target their messages.
Has it happened? No.
A look at Microsoft's adLabs today shows that the demographics have shifted but not in the way that was predicted. The under 18 bracket has slid up, a small fraction, to only %9.80. Despite this, Microsoft is predicting an even higher increase is still to come and states that the future audience for Digg will see nearly %22 of the viewers on the website 18 years old or younger.
There have been bigger changes. In August Microsoft adCenter Labs predicted a drop in the number of 24~34 year olds on the site. October's figures show that it has increased to over %27.
The adCenter labs also show that the gender gap on Digg is closing and there is now only a 12 point split between male and female web users on the site.
The August case study revealed a 56/44 split in favour of male viewers on the site.
Does this mean that Microsoft's demographic tracking and prediction technology is no good? No. We need to zoom in on the small print. Microsoft does make it clear that the adLab tools are not yet in full production. The demographic prediction tool is a prototype to demonstrate what adCenter tools might be like. An even more important caveat is that the data used to process the results is a sample only. The adLabs tools do use actual audience data but not nearly as much data as Microsoft has available to it. When it comes to demographic predictions then volume in data does help with accurate predictions.
