Nielsen Corroborates Jupiter's Cookie Deletion Report

7 comments
Source Title:
Don't like our cookie data? How about some from Nielsen/NetRatings
Story Text:

Jupiter's Eric Peterson posts notes on his conversation with Nielsen/Netratings about their much talked about and subsequently defended report on cookie deletion among web users.

The data would appear to corroborate Jupiter's findings, and some...

  • When N/N fielded a survey question asking about cookie deletion in the last 30 days, 43.7% of the 9,492 respondents indicated that they had deleted cookies in the last 30 days (remember, our number was only 39%)
  • When comparing visitor retention rates between N/N's web analytic application (SiteCensus, formerly known as RedSherrif) and their NetView measurement panel, the cookie based retention metric reported half-as-many returning visitors as the measurement panel on five-out-of-six sites and was significantly lower in all cases
  • When N/N leveraged their measurement panel (a direct measurement from participant desktops, able to report back when cookies are being set) to examine seven different portal and publishing web sites, N/N saw one-month unique cookie deletion rates that ranged between roughly 7% to nearly 50% for both at-home and at-work audiences
  • Finally, when N/N looked at Google.com in July 2004, a site being used by over 40% of their panel and visited an average of eleven times per month per person, nearly 25% of their measurement panel was observed to receive a new google.com cookie during the study. What's more, nearly 15% of these cookie deleters were receiving multiple cookies, resulting in a 55% over-estimate of unique cookie counts. I tested the Google.com cookie to make sure it was persistent (similar to most-third party tracking cookies) and discovered that the cookie is set to expire on January 17, 2038)

Jupiter got quite a rough ride over their report so we'll forgive Eric his slightly jubilent/desperate title heh...

Comments

nice find

...what all this means is not that we should dismiss cookies altogether though. Cookies are still nice tools; both for ecom, aff biz, tracking, advertising, and personalization.

What those studies do not say, is that the time span is important - cookies to be recognized within the same day are the most reliable, cookies set within a week are somewhat less reliable, and cookies set one month ago are less reliable still.

So, apart from research firms measuring bodyparts, the practical implications comes down to this:

The shorter the period the cookies are observed, the more reliable the cookie data is.

(and you may quote me for that - Nielsen and Jupiter will say the same eventually)

even lower usage rates observed here

The shorter the period the cookies are observed, the more reliable the cookie data is.

Consider yourself quoted :)

Some of the original criticism depended on arguments such as "users don't know how to delete cookies". Which may be true true in many cases. But it could be argued that the same class of users is much more likely to let anti-spyware programs do whatever is suggested, which includes deleting cookies.

The stats reported by various web stats pages had me confused for a long time. These report rates of over 90% while my observations are at under 20%. What are others seeing?

Finally I will report that the official position of w3c in the context of a discussion last year about removing the ability of using embedded usernames and passwords in url's is that cookies are for session information. And they were adamant about this.

differences

Differences might have to do with the specific audience type.

I would like to see some reports that relate cookie deletion to number of years with online experience and/or number of hours spent on the internet per week/month. Of course, technical insight (eg. because of job/education) could also affect deletion rate.

 

>Differences might have to do with the specific audience type.

I think so, too. But I'll add that a spot-check with the boys gives anecdotal confirmation (in rough numbers) to Jupiter's report when dealing w/ JohnQ.

~And~

My own personal observations of SEO-types indicates that you shouldn't even think about setting a cookie or the newly outed flash scumbug for marketing purposes. They'll keep the ones that suit their purposes, but nuke the rest in a heartbeat.

Nagging, nagging...back of head....

I may be getting this completely the wrong way but...is this a further effort to bolster PPC?

First we had "it's great, you can track it, ROI ROI ROI!"

then "but don't forget the branding value"

now we have, "you can't track it effectively but trust us its great"

Just me?

 

>now we have, "you can't track it effectively but trust us its great"
>PPC

I think there's an element of that in there, NF, mainly because we're talking HUGE sums of money involved in who-gets-the-cookie and where there's money.....

But having never been a beneficiary of cookies (even in my aff stuff) and being paid almost solely by either click or CPM I don't get too concerned about the issue.

third party cookie behaviour

With respect to my earlier observation I have attributed some of the difference to the highly technical audience of the tracked sites. They only go there for specific pieces of server software. The clue that got me to ask the following (long winded) question is that the observed rate of cookie acceptance seems to be just about the inverse of the market share of IE6.

Since there are a number of old hands participating in this thread I wonder if I might ask
for observations on the definition of third party cookies. In particular, for sites a.com and b.com, if a.com serves up a page containing a script originating from b.com, and said script sets a cookie with the domain set to .a.com, is that a third party cookie as defined by Microsoft? Looking at MSDN and KB did not yield much help other than to point me at compact policies.

If this is the case, then the two choices I have to test are:

1) serve script from same site
2) serve script from subdomain, for example lib.a.com which is an alias back to b.com

#2 is the preferred option for maintenance reasons

Am I barking up the right tree?

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.