Microsoft adCenter Labs - 'Demographics Prediction' Test
The blogosphere has been discussing Microsoft's adCenter Labs lately, so Website Services Magazine decided to run a test. Today, we test the Demographics Prediction feature. Look for further tests in the days to come.
According to the site, Demographics Prediction aims to "Predict a user's demographic information such as age and gender based on his/her past behaviors." A user enters a query or URL and gets back data. The idea is that an advertiser can use this information of predicted search habits to better target their advertisements to the appropriate gender and age groups.
We tested the query "Viagra" and were presented with these results:
The gender information was heavily in favor of men, as to be expected The "Predicted Distribution" numbers most heavily favored the age group 35-49 (34.97%), followed by 50+ (23.92%) and bringing up the rear of the category was <18 (4.17%). The combined results of the male demographic and the age distribution falls in line with what is to be expected of a product such as Viagra. Therefore, it seems that the Demographics Prediction has some measure of validity. What is peculiar, however, is that the "General Distribution" figures are not clearly explained. The percentages remain the same regardless of the search terms or URL entered.
A separate URL test revealed an interesting side note. Cosmo.com, Cosmopolitan Magazine's site, revealed a 55% orientation towards men vs. 45% towards women. Either the Demographic Prediction is flawed, or men are still desperately trying to understand women. My vote is towards the latter.