Sponsored Content WEBINAR: ELECTION 2008: E-mail Tips from a Political Insider   WEBINAR: Becoming the Illusionist: The Magic of Using "Brand" to Drive Response without Results Disappearing in the Process   WEBINAR: Score and Win Through Cooperation   TOOLS OF THE TRADE: Holiday's are prime-time for email marketing - maximize your ROI by getting to the inbox   ON DEMAND WEBINAR--MASTER CLASS: High Performance Online and Off-line Strategies for Large Scale Customer Acquisition   FREE WHITE PAPER - Direct Mail and the Internet Find any supplier you need - agencies, CRM, fulfillment, lists, e-commerce, paper, printers, telemarketing, and more. Featured Categories Lists and Data Telemarketing Database Marketing E-commerce Web Marketing Agency & Creative Services Print, Production & Paper Lists and Data Processing :: view all categories Get free access to more than 50,000 list data cards - one of the most comprehensive databases in the industry. >> Search Now DIRECT Buzz Blog Webinars Buyer's Guide Calculators JobZone ListFinder Promo Calendar Research Store E-Newsletters Print Magazine This Month in Direct Magazine Deal With It Direct had a full house for this year's list roundtable. Considering all the additional responsibilities on brokers' plates, that's impressive... See Full July Issue

JupiterResearch set the Web metrics world ringing two months ago with a published report that found Web users deleting Internet cookies in numbers far larger than previously thought.

In the course of a survey conducted this past February and March, Jupiter researchers discovered that as much as 39% of the Web-using audience in the United States say they delete cookies from their computers at least once a month. In fact, 17% of consumers polled told Jupiter they erased cookies once a week; 10% said they cleaned them out daily.

If true, Jupiter's findings would add a large element of uncertainty to many measures of Web behavior, including unique site visitors, sales conversion rates, customer targeting and advertising's effectiveness. The report went on to suggest that the proliferation of anti-spyware might be leading users to get rid of their cookies, either those placed by the Web site or third-party cookies inserted by an Internet marketer or advertising firm.

Other behavior uncovered in the survey suggests that consumers are taking an active part in managing their cookies, determining whether to accept them at all and, if so, from which sites.

For those unfamiliar with the term, an Internet cookie is text that a Web server can store on a user's hard disk. Cookies let a Web site store information on a user's machine and later retrieve it.

Jupiter found that 25% of the users polled said they'd blocked these cookies from being set, while 28% had changed the cookies they would accept or block. And 21% of survey respondents said they had added a site to the “safe” list in their browsers, allowing cookies to be placed.

The report's authors added that users did not seem to understand the purpose of cookies. Nearly two in five said they were “an invasion of privacy and security online.” They suggested that consumer education might be required if cookies were to remain a useful tool for Web metrics.

But a new study published by the Atlas Institute suggests that cookies' fate may not be so dire, and that users in fact may not be practicing the rigorous anti-cookie habits they're preaching.

Atlas Institute director Young-Bean Song says his group chose to field a survey on consumer attitudes to Internet cookies last August. But the company went a step further and deposited a third-party cookie on survey respondents' browsers, then used its ad tracking network to trace the lifetime of that cookie. As a result, Song's team was able to amass actual behavioral data on how the respondents handled their cookies over a set period of time. They could hold what those users said they did up to their actual practice.

DO INTERNET COOKIES HAVE STAYING POWER?

			Cookie Life Span (Days)
How often do you delete Internet cookies?	Percent of Respondents	Self-reported	Actual behavior
Every week	42.6	7	45
Every month	13.7	30	59
Every three months	9.6	90	127
Every six months	7.8	180	117
Less than once a year	6.5	365	164
Never	7.7	N/A	172
Source: Atlas Institute

“Because Atlas serves ads and tracking pixels for about 1,000 advertisers over the Internet, we were able to figure out a cookie's ‘life span’ — the time that we first saw that cookie minus the time we last saw it,” Song says.

In brief, the responses that Atlas got in its poll tracked fairly closely with those given in Jupiter's survey. About 42.6 % of respondents said they deleted their cookies every week; 13.7% told Atlas they did it every month. That means 56% of respondents told Atlas they dumped their cookies at least once a month. That's higher than the 39% who told Jupiter the same thing, and the 44% who said so to a Nielsen/NetRatings poll.

But Atlas found a very different picture in respondents' actions. Those who said they deleted their cookies every week actually kept them an average of 45 days; those who claimed to erase them monthly kept them about 59 days. And users who reported deleting cookies every three months in fact held onto them for around 127 days.

Notably, the trend reverses among those who say they keep cookies longer than three months. The 7.8% who said they clean out cookies twice a year actually get rid of them within an average 117 days; and the 6.5% who said they retain cookies longer than a year actually kept them for about 164 days.

Atlas also was able to examine data for the half year before the survey fielded to see if the proportion of ad impressions that went to short-term cookies — those that only lasted for a day or less before being deleted — had changed perceptibly in that time. This would indicate whether anti-spyware programs were making inroads on cookies, including those delivered by Atlas and Avenue A, another subsidiary of Atlas' parent company aQuantive.

“We wanted to know what percentage of our conversions and clicks and all the other stuff we measure was going to those short-lived cookies,” Song says. “Has that ratio of short to long changed over the last six months?”

The study found that the ratio had only increased by about half a percent in the six months leading up to the survey. “Even with all the news and peer pressure about privacy, and with users adopting anti-spyware and switching to Firefox browsers [which offer automated cookie deletion], overall Atlas has not seen a dramatic change in the number of impressions going to those shorter-lived cookies,” Song says.

Another factor in the Atlas study that may moderate any increased cookie deletion is the finding that 70% to 90% of sales conversions happen within 24 hours of a click or impression. “If someone is deleting their cookies weekly, you're certainly going to lose that data if they convert eight days after seeing a banner or clicking through a search ad,” he says. “But the bulk of the activity is well short of that — you're going to capture most of that activity. Cookie deletion is not going to have this gigantic impact on conversion tracking, whether for search or for banner advertising.”

When it comes to figuring the reach and frequency of ad impressions or calculating the number of unique visitors to a site, accelerated cookie deletion could tend to skew those counts, since visitors who delete cookies but then click through the ad again or return to the site will account for more than their share of cookies. Song recommends an approach that allows for the fact that some portion of the audience is creating multiple cookies.

“Rather than basing their reach and frequency analyses on 100% of cookies, analysts should use the 80% they know are stable and persistent,” he says. Figuring the average frequency for that group will allow observers to calculate a “best guess” for the incremental group that's erased their cookies.

Song points out that the Jupiter report suggesting an increased deletion rate for cookies has served to polarize a contrast that exists between two ways of measuring behavior on the Web: the technology-based method that uses cookies, espoused by Atlas and Avenue A among many others; and the panel-based approach that employs a representative sample of Web users and tracks their behavior as typical. This second type of Web analysis is the one chosen by Nielsen/NetRatings and comScore as more reliable. To Song's mind, that split is coloring the debate over cookie reliability as a tool for Web metrics.

Nor does Atlas' report answer some of the questions raised by JupiterResearch — particularly in the results that showed one-fourth of the Jupiter respondents using software to block cookies from being set. The Atlas study tracked a cookie's life span. but you can't track a cookie that never gets placed.

The bottom line, Song says, is that the whole question of cookie persistence needs to be buttressed by more real-life empirical evidence. “We know that cookie deletion is an issue, that consumers are concerned about protecting their privacy, and that we need to do something about it,” he says. “But simply asking people how often they delete their cookies is not the way to size up the problem. You're just not going to get an answer based in reality.”

Instead, Song and his team propose a wider analysis of Web behavior. He's currently working with the Advertising Research Foundation to establish a panel that would include some of the major Internet publishers to examine cookie deletion. While details have not been finalized, Song says the effort probably would rely on a particular site, such as a free e-mail provider with a registered user base of millions. The study would then check those users over a period of months to see how often they had a different cookie in their browsers when they returned to check their e-mail.

“By looking at millions of people over different time periods, we'll be able to get a pretty strong picture of how often people are deleting their cookies,” he says. “It will require time to set and to collect that data, but we hope to have results before the end of the year.”