comScore: Cookie Deletion Skews Site Stats

Market research firm says cookie deletion can lead Web sites to substantially over-estimate their number of visitors...and publishes a white paper to detail its stance.

In the Internet advertising game, the number of visitors and hits a site can draw is the name of the game, so there's a battle between companies, like Nielsen NetRatings and comScore Media Metrics, who attempt to measure the vistorship and user bases of major Internet sites, and those sites themselves, who sometimes claim these media analysis numbers can't possibly be right, because their own server logs show substantially higher levels of use. The sites are up in arms because the amount of ad money they bring is is tightly bound to these ratings, and the analysis firms aren't big on disclosing their rating methodologies.

Back in April, Internet Advertising Bureau head Randall Rothenberg had had enough, and posted an open letter to the two largest ratings firms (comScore and Nielsen) essentially demanding they disclose their methodologies and prove their numbers are on the money. In response, last month, Nielsen announced it would submit its audience measurement system to the Media Rating Council's accredidation process, essentially opening up to an independent audit to justify its numbers.

Comscore hasn't decided to open up to an audit, but it defends the accuracy of its Web usership numbers, and today published a white paper highlighting how cookie deletion can cause sites to inflate their number of users by as much as 2.5 times. Cookies are little snippets of data Web sites can request Web browser software store on a computer; when users go back to the site—or a site which knows about a particular cookie—the site can ask if you have its cookie, and, if so, ask for it back. The data stored in cookies can have significant privacy implications (as borne out through the DoubleClick fiasco back in the early dot-com boom), but many sites still use the technology to identify returning visitors and analyze users "paths" through a site, measuring what visitors load and what they don't.

With the privacy concern over cookies, many Web browsers and add-ons now include tools which let users decline cookies, automatically reject them from third-party sites, delete them altogether based on criteria or a certain period of time, and otherwise directly manage how cookies are used on their systems—if they so choose. comScore argues that these tools—combined with users' increasing propensity to use multiple browsers, multiple computers, and run security software, often lead sites to over-estimate the number of unique visitors they have. Instead, comScore argues, the sites are often assigning new cookies to returning users and inflating their audience numbers.

Next move? Expect sites unhappy with their audience measurement numbers to fire back, explaining they know all about cookie deletion and have already accounted for it in their own numbers—which are still higher than comScore's.