Everyone else has blogged about it, I will also: Great New York Times article on site relationships with comScore. It is amazing that, even with todays amazing server logs that track every file download, it is not possible to really use that data to create apples to apples comparisons of user surfing habits. comScore fills three crucial gaps in the data gathered by sites to paint a different picture of online use:
- Discard international traffic
- Aggregate work/home traffic
- Overcome cookie deletion to identify multiple instances of the same user
Is comScore imperfect? Sure. What is really bitter is that all of these issues also plague ad servers such as Atlas or DART, so the way campaigns can be delivered may differ from the picture that comScore says is possible on a site. And it is amazing that you should expect discrepancies between comScore and sites – after all, they use different methodologies, but you would not expect such big differences between advertiser ad servers and sites. The methodologies should be synced up there.
There is clearly still a lot of foundational work to do here. Having said that, these discrepancies stem from the fact that more data is available in this medium than ever before. If an agency announces that it wants to sit on the sidelines until these metrics get reconciled, than other arbitrageurs will step in to fill the void. On a performance basis, I still believe that most online advertising offers great value!