Website usage statistics (or metrics) have moved through many incarnations since the early days of the web, a constant evolution of services focused on improving the accuracy and variety of metrics data tracked all while accommodating the explosive growth in web traffic. Broadly speaking, the current state of the art divides metrics into 3 main types that use different methodologies and are targeted at users with different needs both in terms of accuracy as well as performance:

    Web server log statistics
    Third party Tag based analytics
    Ad Impressions

Web server log statistics

These are compiled from the records kept by the web server whenever a request is made for it to deliver a file. Typically this log contains the time, IP address, Browser Agent, the file being requested and the referring web page plus various status and size information. An analysis tool is employed on a regular basis (most commonly daily) to count and categorise the requests made to produce aggregate reports detailing the site usage grouped by time period, type of usage, browsers used and so on.

The strength of this type of usage analysis is that it gives you a very good idea of what is happening on the server so you can analyse broken links and other errors, but with the proliferation of caches and proxies it has the downside of missing traffic which did not involve the server.

Third party tag based analytics

These operate by adding a piece of code (the tag) to your website pages which then notifies a third party server whenever a page is viewed. Because it executes every time, even cached and proxied pages are counted and the use of tracking cookies allows a much more accurate report on user behaviour, including an accurate reflection of entry and exit pages and popular user paths.

With the addition of optional codes to the tag this approach can also provide enhanced tracking of conversion pages such as form submissions or payment confirmations. By adding identifiers to a page's address (URL) the origins of incoming traffic can also be identified. Originally sold as commercial premium services, various popular free services like Google Analytics now exist.

Ad Impression

Impressions and click-through statistics are collected for websites used to serve ads and are generally more accurate in providing numbers on which to forecast and sell website based advertising.

Other website usage statistics collection and analysis methodologies exist like those which source data from large internet service provider logs, or load ratings agents onto users' web browsers and track their behavior much like TV ratings. These are often used specifically for analysing user behaviours or for accurate comparisons between disparate websites.

Which is better?

Rather than one metric system being innately superior it is best to use multiple systems for the purposes they are most suited for. Web server log based analysis is ideal for error checking and system administration while third party tag based analytics work well for user interface optimisation, information architecture and analysing usage patterns, and you should use ad server statistics when forecasting and selling and reporting on website based advertisements.

What is crucial is ensuring that your website has the built-in technical flexibility to accommodate the various types of usage analysis so you can confidently use the approach most suitable for your need without having to worry about the limitations of your website, if any.

Brief glossary

1.    Browser Agent: The type and version of the user's web browser eg: Internet Explorer 8.0
2.    IP Address: The numbered internet address of the users computer. eg: 10.10.67.193
3.    Pages: An actual website page
4.    Hits: Other web server requests like images and code
5.    Entry page: The first page viewed by a user during a session
6.    Exit page: The last page viewed by a user during a session