Thursday, April 19, 2007

Cookies will get you fat

The recent press release from ComScore about cookie deletion rates fired up very good discussions on the web analytics forum. Here's my 2 cents about it.

Nutrition facts

  • Cookie deletion rates appear to be very similar for 1st and 3rd party cookies
  • Cookie deletion might inflate unique user stats by 2.5 to 12.5 times
  • About 30% of the people surveyed delete their cookies at least once a month
  • The previous study, from JupiterResearch, reported a rate of 39% of people deleting their cookies (2005)
  • The study was conducted for a month, using 400,000 home PC's.

Health analysis

  • When users delete their cookies, they don't discriminate if they are 1st or 3rd party cookies... they delete all of them. The slight difference between the two would be the impact of using privacy tools that automate the blocking/deletion of 3rd party cookies (between 1% and 3% more deletion for 3rd party than 1% party - not really significant)
  • The margin of error when reporting on "unique users" becomes larger as you increase the time span. Looking for unique users within a week is more accurate then looking at them within a month (or worse - a year).
  • The study from ComScore doesn't cover the fact that more and more users are using multiple devices (home & company computer, PDA and other devices) and even multiple browsers (sometimes switching between Firefox and MSIE, both installed on the same computer). So the notion of "unique user" should be retitled to "unique device + browser"
  • The inflation of the numbers doesn't change the trend, it only changes the scale.


  • Avoid providing hard numbers; trends and context are essential.
  • Use the "visit" as the metric of choice, think in term of "opportunity".
  • The only instance where you should report on "unique users" is when you can use a real unique identifier such as a login.

Salted cookie

  • ComScore business model relies on panel users, it's hard to imagine another conclusion than something implying panels are more accurate than other methods.
  • ComScore filled a 85$M IPO on April 3rd, the timing of this report is certainly not a coincidence.
  • ComScore have not published the details about the methodology. What could be the implication of rewarding panel users for their participation? Exactly how/if anything that could influence user behavior was was communicated to the panel users.