Server log files were created to monitor server performance, not to measure visitor behavior. Server side tracking software is blindly counting activity on the server and trying to determine what this activity means. Server based systems that count “clicks” are normally free with a hosting contract but when a system only Tracks what the Server sees there are serious Over counting/Undercounting problems:

• Counts Robots/Spiders
• Doesn’t count cached pages
• Unique Visitors based off of IP Addresses (which change often)
• Requires – Servers, Software, Licenses, Maintenance, Storage for Logs
• Essentially Raw data not management information.
• Originally designed for IT to monitor Server/Network issues
• Knowing the number of clicks, just does not help you make any marketing decisions

Server Side Over Counting Problems (Too much activity being reported)

Server Based software records hits that are not representative of how many real visitors/users are coming to your site.

Search engines like Google have spiders running all the time on almost every site looking for how relevant your site is to their search engine ranking algorithm. These may account for hundreds of hits to your site over the course of a month. These are not people using your site. Rather they are machines wandering through your site.

Every time a page is requested from a server (a page is made up of multiple elements – text files, graphics, scripts etc), the server logs these as multiple requests. If you have 20 files that are on a page, the Server Side Tracking will count this as 20 hits each time the page is requested. Whilst counted multiple times this is really only one user session, hence an over counting problem.

Server Based Software will report every spider, robot, and machine crawling your site, causing a 30-70% increase in visits/hits to your site unless you take the time to filter the data. Some examples of over counting to look out for that need to be filtered include:

• Internal Staff viewing site.
• Internal Staff’s default Internet page is usually set to the company’s site.
• Server Based Systems (Log Files) log an IP address. One unique IP address should equal one unique user. If an IP address changes multiple times during one user session, the server-based application will count each change as one new user therefore grossly over counting. Some proxy servers can change IP address with a click of a mouse.

Server-Side Under Counting Problems (Too little activity being reported)

• Proxy Servers are used by many large organizations (Boeing, GE, Ford) and ISP providers (AOL). Large organizations use proxy servers to make it easier for employees and customers to download websites and information by placing numerous servers closer to the end user versus having to pull information or that page again and again from a server on the other side of the country.
• So when a customer connects to the Internet and pulls down your site, the site is actually being viewed from a proxy server (not your server). It will download your site once to their Proxy servers. Each Proxy then serves your site to their customers / subscribers.
• Server based tracking only counts the first download. It does NOT count activity on Proxy servers thus grossly undercounting the actual number of visitors.
• In addition, a huge amount of activity goes unaccounted for as a result of pages being cached on Proxy servers.

Please review your web tracking data system and contact me at john@johndiethelm.com if you have any questions.

For more information, see www.webclicktracker.com or click here for the Website User Behavior  & Analysis Tool sign up form.