• by Eric T. Peterson , June 26, 2007

We now know who's been bedeviling the number crunchers

It's no secret that the unique visitor counts reported by measurement companies are flawed at best. One explanation and perhaps the most important: Consumers delete their cookies.

Widely documented in March 2005 by JupiterResearch in its report Measuring Unique Visitors: Addressing the Dramatic Decline in the Accuracy of Cookie-Based Measurement, cookie deletion has weighed heavily on the minds of many advertising buyers and site owners.

I authored that study and took a lot of flak when we published that report. Some accused me of yelling "fire" in a crowded movie theater. But, as it turns out, the building was burning and I was right to speak up.

And now we've caught the arsonist with the matches in his hand, reeking of gasoline. Many people speculated that anti-spyware applications recommending the removal of tracking cookies were the problem. Anti-spyware is not without blame, but it turns out the real culprit is more insidious, more dangerous and more prevalent. The real culprit is ... you.

Shame on you.

On April 16 comScore reported that 31 percent of its panel members had deleted a commonly found persistent cookie at least once. If such cookies were used for measurement, the result would be double-counting of that panel member upon the next visit.

But measurement companies face a threat worse than the occasional cookie deleters: serial cookie deleters - the real cookie monsters. Around 7 percent of panel members repeatedly deleted the cookies being tracked. Measurement companies classify people in this group as new unique visitors eachtime they return to a site, because without cookies, these people aren't recognized as returning visitors.

Why would 7 percent of Internet users manually delete their browser cookies? I'll offer three equally plausible answers:

The first, suggested by both JupiterResearch and comScore, is that most consumers don't understand what cookies are for, but are nonetheless suspicious about them. ComScore examined this in a follow-up survey of 500 panel members and found that while 13 percent of respondents said they improved the user experience, 15 percent said they detracted from it. Nearly 48 percent said they did both.

The second is related to a lack of understanding about cookies. Anecdotally, many tech support operations recommend deleting cookies and clearing the browser cache when software and systems are performing poorly. Personally, I was surprised when my 60-year-old mother professed to regularly delete her browser cookies because, she says, Dell tech support told her it would help system performance.

The third reason is believed to be based on a desire to erase evidence of visits to the dark corners of the Internet - sites offering pornography, gambling and the like.

The first and second reasons imply naïveté, the third some level of sophistication. Likely all three answers are correct depending on which of the 31 percent of Internet users you ask. Unfortunately the results are the same, and you end up counting cookie deleters again (and again).

What can you do about it? Nothing yet.

Eventually, some knight in shining armor will ride in with segmented data showing monthly cookie deletion rates for specific vertical and sub-vertical markets. That simple data point - the estimated rate of cookie-based visitor inflation - would basically provide the information that sites need to scale down unique visitor counts appropriately.

One in three of us could persist in our strange behavior, advertising managers could better estimate visitor counts, and all but the geekiest of Web analysts could move on to solving harder problems.

ComScore? Nielsen//NetRatings? Quantcast? Are any of you ready to don your armor and ride in to save us from ourselves?