Problems with website statistics

Web site statistics are not 100% accurate.
Technology used by many businesses and Internet Service Providers (ISPs), such as Freeserve and AOL, affect the accuracy of web site statistics. Many Internet users may appear to be one user or one user may appear as many viewing your web site. Web site statistics cannot tell you exactly how many visitors are looking at your web site and often record referring web sites inaccurately.

They should be used to measure growth in visitors over a period. You can use the numbers provided to calculate percentage of growth.

You cannot expect to “bolt” a standard web stats programme onto your web site and expect it to be universally accurate. Differences in programming (ASP, PHP etc) used by different web site designers affect the accuracy of site statistics. One size fits all just does not work!

Some examples of inaccurate web site statistics are explained below:

Referring web sites not recorded correctly
Many web site statistics often show referred traffic from a website (e.g. www.carsupermarkets.co.uk) as a direct referral / visitor. While this does not affect total visitor numbers it greatly distorts the overall figures, as the webmaster does not see the correct source of traffic to their web site. Tracking URL’s (such as an online publishers own traffic recording mechanism) and differences in browser technology are just two of the factors that can cause massive discrepancies between the actual number of referrals made by a web site and the data actually recorded.

Ask a publisher to link to a unique page or URL such as yourwebsite.co.uk /uniquepage01.htm that is not accessible from any other source. The number of times the unique page is requested should be an accurate representation of the referring traffic from any single location. In addition to this most online publishers record their own advertiser statistics and should be able to provide visitor numbers and other information.

To demonstrate the problem we have created a file which lists all HTTP Headers sent to the server when a request is made.

- HTTP Headers link 1
- HTTP Headers link 2

If you click on HTTP Headers link 1 you will notice the referer header records:

HTTP_REFERER= http://www.reallygooddomains.com/problems-with-site-statistics.asp

If you click on HTTP Headers link 2 however, you will notice the referer header has been excluded!

Total number of visitors does not add up
Many websites have inbound links that are not to the home page, or often not even to the website itself. A link that bypasses the home page to an internal page of your website is called a deep link. An example of this could the Cars In Stock section of a motoring website. If the stock system is part of the website being linked to then visitor numbers should at least be correct. However, lots of companies use a system called Glass’s Trader Desk. On www.CarSupermarkets.co.uk for example users can click one of two buttons to visit an advertisers website:

1 – Visit Website

This is a link to their home page.

2 – View Stock.

View stock in many cases is a link to the relevant car dealer's stock, but it is NOT a link to their website. It is actually a link to Glass’s Trader Desk and therefore will not show on the car dealer's web site stats as a referrer – because it is not a link to their website.

Inaccuracies in unique users
There are a lot of people using computers at Sue’s work. Sue’s work, like many businesses, doesn't use modems to connect to the Internet. All of the computers are networked together and use a single broadband connection to connect to the Internet. The IT manager at Sue’s work has chosen to use a technology which will assign Sue's computer a new IP address periodically. Sue’s computer could be assigned a new IP address every day, or even while she's in the middle of surfing a web site. Of course when Sue's computer at work is assigned a new IP address, she will then appear as a different person or an additional unique user. Sue's computer may be assigned a new IP address even while she's surfing a web site. This means, in one visit to your site, Sue may appear as two people. This technology is also used by some ISPs, like AOL.

How cache or caching makes statistics inaccurate
Caching is the process used by computers and ISP’s to store information for easy retrieval later. Your browser will have cached / stored a copy of this web page. If you go to another web site and then revisit soon after the computer will present the page more quickly without downloading it again because it already has a copy stored for easy retrieval. Caching enables your computer and the Internet to operate more quickly but this widely used technology makes all statistics inaccurate.

What is a Hit?
A hit is a request for a file and should not be confused with a visitor. People often say that a web site had 10,000 hits yesterday – meaning 10,000 visitors. There is a big difference! One web page consists of several files which each create a hit when downloaded. One user could create 30 hits just looking at one page.

What is a Page View?
A page view is the delivery of a web page. Page views are often used as a measure when buying and selling Internet advertising and will give a good indication of how well used a web site is.

What is a Visit?
A visit is simply someone visiting your site.