Since Web sites have been counting traffic, they've bitterly complained about their Nielsen or comScore numbers. "We reach twice as many uniques as they say" is a common contention. "Those guys are a joke" is another often-heard diss.
Many grouse that employing data from panels in any way, shape or form in a medium that can track every pixel is backwards. Yet there are major problems with using a site's own data, according to the third-party researchers—you can double count people who delete their cookies or log on from work and at home, for example. As part of a compromise, comScore has begun blending panel data with server data—appeasing some while driving others nuts since year-over-year comparisons are now laden with asterisks. Meanwhile, Nielsen recently admitted to a serious problem with audience undercounting.
A slew of alternative players have emerged, such as Quantcast, Compete and Google, which promise more precise tracking tools that will count user activity on a page-by-page level. Yet they've only seemed to make the traffic quandary worse. Now there are half-a-dozen services employing different methodologies. Even worse, they rarely seem to correlate, confusing many buyers while allowing publishers to pick and choose which numbers they like best. Here's a quick look at the reporting disparities.
Nielsen (June): 12.3
THE HUFFINGTON POST
Nielsen (Mar): 13.0
THE DAILY BEAST
Nielsen (Apr): 21.1
All unique audience numbers in millions; numbers other than Nielsen's are for October 2010. *Does not disclose traffic. Source: Internal figures derived from Omiture (in the case of The Daily Beast) and Google Analytics (in the cases of Break and HuffPo). Internal numbers for Twitter, Break and Hulu are global.
Download a PDF of the page here.