A new study from comScore has questioned the validity of using cookie-based data to measure website audiences, claiming that unique visitor counts for websites may be overstating the actual visitor totals by as much as 150%.

The measurement group, which surveyed 400,000 home computers, found that three in ten US internet users regularly delete cookies from their computers.

The study found that 31% of US internet users deleted their cookies during the month, which means that each time these users return to a website they are counted as a new, unique visitor.

The study looked at cookies for an unnamed but ‘prominent’ US web property and a third party ad server, each of which reaches audiences of well over 100m per month.

It found:

  • An average of 2.5 first-party cookies were observed on each PC.
  • 31% of users deleted first-party cookies at least once during the month – the study found an average of 4.7 different cookies for the site.
  • Just 7% deleted cookies more than four times, but this group accounted for 35% of all cookies in the study - an average of 12.5 per user for the site in question.

The figures for third-party cookies – those left by an ad server in this case – show a similar pattern:

  • An average of 2.6 cookies were observed for each PC for the third party ad server being studied.
  • 27% deleted these cookies at least once, and their PCs contained an average of 5.5 cookies for the ad server.
  • 7% deleted cookies more than four times, and these users accounted for 38% of cookies. An average of 14.2 per user.

Most cookie deletion seems to be the work of security/anti-spyware software, with the survey suggesting that selective cookie deletion is not common – just 4% of users said they deleted third party cookies only.

Related reading:

Cookie deletion discrepancies take the biscuit

What will replace the page impression?