The Trouble With Cookies Persists

How can something with such tasty connotations be such a source of distaste?

Cookie deletion has long been a blight on digital campaigns, skewing conversion tracking and affecting the validity of reach and frequency metrics. It’s a boundless, borderless issue that affects our industry the world-over (for a breakdown of cookies and how they work, see this recent story on ClickZ Asia).

A recent industry analysis from comScore puts it in sharp perspective: as outlined in its new report, the research group found that current cookie measurement systems overstate the number of users by over 2.5 times. The study looked at a first-party cookie from Yahoo and a third-party ad server persistent cookie from DoubleClick as served to the Australian Internet population to determine the degree to which Internet users were deleting their cookies and causing publishers to deposit new ones.

ComScore concluded that measurement systems using first-party cookies are likely to overstate the number of unique users by upwards of 2.7 times. Ad-measurement systems that measure third-party ad server cookies, meanwhile, could overstate users by up to 5.7 times. And these results aren’t exclusive to Australia: a previous study conducted in the U.S. found nearly identical results.

Publishers, agencies, and ad servers have long been working to overcome this campaign hurdle, and they seem to be making some progress. ComScore continues to promote an approach that it believes can compensate for overstated unique user counts: its Unified Digital Measurement system, which was introduced in 2009, uses the famed comScore user panel data but also combines it with site server data to deliver a more complete report.

Last summer, audience measurement firm Quantcast was in the news for receiving accreditation from the Media Rating Council (MRC) on its Quantified Publisher Program – and beating out both comScore and Nielsen in the race to get it. The company’s program measures the audience data of those publishers with sites that feature a Quantcast tracking tag.

Most recently, MediaMind Technologies (formerly Eyeblaster) received MRC accreditation for its Audience Reach Measurement system, which uses a statistical algorithm to calculate actual audience reach and frequency. It’s the first third-party ad server to be accredited by the MRC in four key categories; in addition to audience measurement, it’s also accredited under ad serving, rich media measurement, and broadband video. “In this data-driven space, consistency and accuracy are more important than ever,” says Gal Trifon, president and CEO at MediaMind. “We are extremely pleased to be accredited by the MRC and offer true metrics to a global audience and meet the high standards of the digital advertising industry.”

Accreditation aside, MediaMind may benefit from the way in which it differs from audience measurement firms like comScore and Nielsen. MediaMind’s system does not rely on the panel data that’s such a point of contention for many agencies and publishers. Nielsen, for example, claims 230,000 panelists while comScore’s panel consists of 2 million users worldwide, but still both companies have been criticized for omitting huge chunks of the online audience, including users who share computers or have multiple machines, those who use public computers as in Internet cafes, those who use multiple browsers, and consumers who rely heavily on the mobile Web. ComScore asserts that its Unified Digital Measurement system accounts for such factors, but for many marketers, panel-based measurement of any kind requires additional vigilance.

As the industry continues to cope with cookie deletion, we’ll have our eyes trained on new developments in Internet audience measurement. There’s no telling what’s yet to come (consider device fingerprinting as one possible direction), but in the meantime, it’s reassuring to know that everyone shares the common goal of practical, accurate audience measurement.