Unique users and hosted web analytics

Every now and then I'm met with a question about the uniqueness of unique users. Often the question is triggered by someone who has tested the website with their analytics solution and found that the number of unique users is higher than expected.

This imprecision in the unique users count is common in most hosted web analytics solutions. SiteAnalyze is, like most other hosted analytics solutions, cookie based. So when a user reaches the website we are tracking, we look for a cookie, to see if that user has visited the website before. If we don't find a cookie the user will be registered as a new unique user, and we'll place a cookie on the computer.

It is this cookie, that is the cause of the imprecision in unique users, and that is because cookies signify one browser for one user on one computer. Here is a simple example of one user resulting in two unique users sessions in the analytics. Imagine John Smith surfing away on his laptop in Internet Explorer and reaching your website, he will then be counted as one unique user. Later on the same day John is surfing the internet, again from his laptop, but this time he uses the Firefox browser, he will then be counted as a new unique user, even though he already visited your website. This is because the browsers don't share cookies.

He will be counted as yet another unique user, when he enters your website from his desktop computer at work, and when Jane Smith logs on to John's computer with her own user account, and enters your web site, she will also be counted as a new unique user, because cookies aren't shared between user on a computer, even though they use the same browser.

One more thing that can add to the imprecision of the unique user count, is when your users delete their cookies. Most users do this once in a while, and that will result in them being counted as new unique users the next time they enter your website.

So to sum it up, the unique users count (or unique visitors in some tools) is not an exact science. The more web savvy your users are, the more different browsers and computers they are likely to use. Which again will result in less exact unique user counts.