Changes to Cookie Counts in CookiepediaBy: Richard Beaumont | Monday, August 11, 2014 | Tagged: Cookie Stats, Cookiepedia | Leave Comment
In the last few days we have been making some significant changes to the way we recognise and record certain types of cookies in the knowledge base that powers our site all about cookies – Cookiepedia.
The result has been a big re-calculation of the prevalence of different types of cookies on the websites we have scanned. In particular we now believe that the percentage of cookies involved in visitor tracking and targeting – nearly 60% of all cookies – is much more representative of the true picture.
Cookie Name Patterns
The two most important elements of any cookie are its name and value. The traditional view of cookies is that when they are set by a particular program, the name is always the same and the unique information is stored in the value.
Most cookie programs work like this – so when you come across two websites that contain the same cookie name, you can make an educated guess that it is the same program setting the cookie – and it has the same purpose. So when you come across the cookie name __utma – you can pretty much guarantee that this has been set by the Google Analytics program.
The trouble is that some cookie programs also store information in the name, and therefore the name of the cookie changes each time it is set – even though it is essentially the same cookie performing the same purpose.
The names of such cookies however do tend to conform to specific patterns, so that once we have recognised this – we have been able to do something about it. For some time now we have had an algorithm in place that can identify the purpose of unique cookie names, when they conform to a known pattern.
However, until recently we have still been counting each example of these cookies as unique – and we felt that this skewed the statistics as some websites have many different copies of pattern cookies, that were really just instances of the same cookie doing the same job.
In fact, the stats became dominated by one particular pattern of cookie, set by Facebook, known at the EagleEye cookie. This has a name that begins with _e_ followed by string of letters and numbers which is unique each time it is set. Multiple copies of this cookie can be set in a single user interaction – but it is destroyed almost immediately it is created. Users almost never see it in their browser records, and it is believed not to place any data in the browser, but to confirm that the platform is working correctly.
By counting each instance of this as one cookie with lots of copies, instead of lots of unique cookies, we believe our headline statistics are now much more accurate. It has also enabled us to work on some new functionality the we expect to release in the coming weeks, which will further improve the usefulness of the information published on Cookiepedia.