[ About Us | Popular | Marcom | AdNet | IChannel | Glossary ]

Feb 5, 2008

NEWS: Cookie Deletion Inflates User Metrics

Here is the summary from a recent comScore study.

Assuming proper Web traffic hygiene such as filtering “bots” and internal traffic, it’s possible to explain most differences between panel-based data and server-based data with the following simple equation:

Difference = (Inflation Due to Cookie Deletion + Inflation Due to Cookie Blocking) * Inflation due to unfiltered International Traffic

Due to the multiplicative nature of these effects, the overall inflation can reach 10x or more. For example, a site that has a 2.5x cookie deletion inflation and 4.5x inflation due to international traffic will have a staggering overall inflation of 11.25x in its counting of unique visitors.

Deleted cookies disconnect users from their actions.

Request your copy of the Cookie Deletion White Paper at comScore.

Click Fraud Keeps Rising, Up 15 percent in 2007, Techcrunch, Feb 22, 2008

Forensics has some data out on click fraud (clicks on Internet ads that are not real) in the fourth quarter of 2007 and for the full year. The industry-wide average click fraud rate for the entire year went up 15 percent, ending the year with 16.6 percent of all clicks on Web ads being fraudulent. The click fraud rate for search engine ad networks alone, including Google AdSense and Yahoo Publisher Network, grew even more. That was up 47 percent in the fourth quarter, ending the year with a 28.3 percent click fraud rate. According to this data, nearly one out of every three clicks on a Google or Yahoo ad is fraudulent.

Accidental clicks fade into the background, Google AS

Earlier this year we stepped back to examine the value users, advertisers, and publishers derive from clicks on content ads. As you integrate ads with your site's content and navigation, we want to ensure a positive user experience. We identified a few areas for improvement and began implementing changes, starting with our new ad formats in April.

Continuing these improvements, we've just changed our text ads slightly to help reduce accidental clicks. In the past, users could click on both the background and full text of an ad, but now they can click only on the title and URL of a text ad. By allowing users to click only on the ad title and URL, we aim to decrease accidental clicks, better aligning visitor behavior with their intent. Overall, the decrease in accidental clicks will keep users on your website, interacting with your content, until they intend to click on an ad.

In addition, this new clickable format better aligns with the text ad formats shown on Google.com. We hope this format change contributes to a better, more consistent user experience.Finally, this change won't just improve user experience on your site; it benefits advertisers as well. We currently monitor clicks on Google ads for accidental clicks, and the format change complements our monitoring system by further ensuring advertisers only pay for meaningful clicks. By reducing accidental clicks, we hope to increase advertiser campaign value and satisfaction, encouraging additional spend and facilitating higher monetization for all publishers.

Mobile UU Problems

Hello all. I would like to get your feedback on the latest strategies for tracking WAP UU. It is generally accepted that tracking WAP unique users is inherrantly a more difficult process then tracking UU on WEB, with the leading cause being that the IP address exposed is that of the WAP carrier gateway, vs. on web where it represents the individual PC/user. As such, IP's are not useful in tracking WAP UU.
As far as I can tell, and what we have implemented at the company i work for, there are 2 basic strategies available for tracking WAP UU, and one theory I do not have much experience with: 1) Tracking users by cookie - PROS: Arguably the best current method of tracking WAP UU, based on accuracy + device compatibility . CONS: users can clear cookies, devices can auto-erase cookies, potentially leading to counting the same user more than once. 2) Tracking users by sessionID - assigning a unique session to each user per visit - PROS: sessioning removes the handset from the equation, and is a server side measurement. CONS: You cannot count a unique session as an individual unique user, as generally speaking, sessions are closed when the mobile browser closes, and re-issued when the browser starts up and visits the site again. Experimental strategy: 3) Tracking users with the openwave gateway parameter "x-up-subno" passed in the HTTP header. PROS: Given the parameter is passed, it is by far the most accurate method for tracking WAP Unique users. CONS: Unknown availability of this parameter. Carrier needs to be using the openwave gateway in order to then pass the parameter.
Questions for the group: 1) Generally, we see about 75% of the user agents we detect as able to be cookied. That leaves 25% of our users to track using either sessionID behavior or the x-up-subno openwave parameter. What approaches have you taken to glean WAP UU from sessionID data. Can you come close to approximating X users to X unique wap sessions? 2) Does anyone currently use the x-up-subno parameter as a way of identifying WAP UU? If so, do you have any anecdotal information about relative success or failure in using this method? If you point your mobile browser to: http://t.wurfl.com/info.php - you may be able to see the x-up-subno ID for your particular device. I see it nearly 100% of the time with Cingular/ATT devices. 3)Are there currently any tools in the market that may provide either a) an out of the box "solution" for measuring WAP UU? b) some sort of analytics tools to run across a WAP site to validate currently used methods for tracking WAP UU? 4)Any other points of interest we might all be able to use to help get a more accurate picture of WAP UU? Many thanks, Lee 

Half of Google Adwords Traffic is from Worthless Link Farms

west-wind.com — A couple of months ago I started in earnest tracking the advertising hits that Google generates for me. And what I found is not a happy picture. About 30-40% of the traffic generated - ie. the traffic that I pay for - comes from link farms, that is Web pages that have nothing more than a bunch of links that redirect to Google Adsense links...

Exclusive: Google’s Click Fraud Rate is Less than 2%

Monday, December 11th, 2006;-- Andy Beal


  1. The international effect is well known among newspaper and magazine publishers. We call this the flyer paper effect. Throw enough paper into the air, something sticks.

  2. Always suspected that the metrics from many sites are largely bullshit. Good reporting.

  3. "Counting of impressions has proved to be difficult due to the Internet infrastructure. This is caused mainly by caching which is intended for cutting bandwidth costs
    (Lockhorn 2001) but creates discrepancies between publishers’ impression counts and third-party ad servers’ counts used by the advertisers. As a result, advertisers cannot be certain whether the counting discrepancy is due to the ad serving technology or due to the publishers’ acting opportunistically and under-delivering. The average discrepancy rate is reported to be between 20% and 30% (Picard 2002) and has been widely cited as the reason for traditional advertisers not embracing the CPM model on the Internet (Lockhorn 2001)."
    Lockhorn J., July 11, 2001. Cache Busting: Busted?
    Picard, E., October 28, 2002. The Three Big Ad Headaches

  4. The news comes after years of frustration from advertisers and publishers with comScore and Nielsen Online’s numbers. The Internet Advertising Bureau found marked discrepencies between ComScore and Neilsen Online’s reported numbers and member websites’ own internal logs. In one case, Nielsen Online reported mlb.com drawing in 6.2 million monthly users, while mlb.com’s internal metrics showed over 19 million. Nielsen Online and ComScore have agreed to audits in 2007 by the Media Ratings Council, audits that have yet to be completed.


Comments accepted immediately, but moderated.

Support Our Sponsors: