Atlas Corrects Cookie-Deletion DMI

The new version is here:

Here's what Young Bean Song, director of analytics and Atlas Institute, sent me in an e-mail:

Upon a deeper review of the data, it became apparent that the average lifespan was not the best way to describe and compare user behavior. Our analysis of the cookie lifespan distributions have revealed a different picture of consumer behavior than what the previously published average lifespans suggest. We found that most of the people who said they delete their cookies weekly do in fact have relatively short cookie lives. Of the people who said they delete their cookies weekly, we found 60% had cookie lifespans of two weeks or less. So what does this new information mean for the DMI that we just released?

The behavioral data from our study is actually in relatively close agreement with the survey-based studies. As a result, we have restated the portion of the DMI that focuses on the average lifespans and have included a more complete view of respondents' behavior. In doing so, you'll see a shift in our view of the survey-based studies. However, it's very important to point out, the main thesis of the DMI remains true - that the sky is not falling on cookies. The insights on conversions tracking, reach and frequency and profile targeting from the original DMI are sound and backed by rock solid research.

Among the survey-based studies Song refers to is the earlier one by Jupiter that showed nearly 40 percent of Internet users delete their cookies at least monthly.

My Online Spin column (registration required) on Mediapost of April 19th quoted from the earlier Atlas report quite a bit. While it attributed everything back to Atlas, it would be prudent for me to write a follow-up this coming Tuesday.

Song insists in his e-mail regarding the revised DMI that "[t]here has not been a radical shift in consumer behavior towards deleting cookies." I plan on reviewing this data with him as soon as possible, which will likely be tomorrow. For right now, the jury is still out as far as I'm concerned.

The conclusions I put forth in my column on 4/19 still hold true:

The fact remains, however, that several popular anti-spyware software packages available for download and/or purchase online have been identifying ad server cookies as potential threats as a default, and this could be prompting widespread cookie deletion as more people download and run the software. As I mentioned in a previous column, two such popular packages - Spybot: Search & Destroy and Ad-Aware - have been downloaded hundreds of millions of times from While the average cookie lifespan may currently be long enough to convince us that the sky isn't crashing down on us, cookie deletion is almost certainly becoming more widespread. It's something we need to watch carefully.

So, while I write my follow-up piece for Tuesday, check out the revised report from Atlas.