Copyright Beaver 2000/2003
So now the user has found your site and stayed round, they had a good time, may be they even keep coming back. But do you know this? You probably don't. Your site could be popular with lots of visitors or have none at all. It is easy to get statistics on the number of accesses your site. We outline below some basic information on monitoring traffic levels from site visitors.
A simple solution is to use some type of tracker or counter system, which can range from free services in return for an advertising banner which generally on a commercial site is not an option, through to specialist server installed software where costing £5000 upwards, requiring specialist knowledge and your own dedicated server to install it on, not just a web host server that most people use for their site. However simple counters that just increment each time a page is requested, including each refresh, are not suitable in providing user statistics data. Here we'll focus on the mid-range type of trackers, which add a piece of code that so that when a page is accessed to transfer the hmtl text and graphics to the users browser from the host server it also registers an access on the trackers server, so collecting the same information as that going to your hosts server. A lot of this data is accessed via asking the browser software for data it collects as you move through each web site mostly relating to the last page you accessed. Any of these statistics systems will log access data and then convert the raw data logs into a variety of nice tables and graphs and use rules to deal with handling refreshes, returning visitors, and other aspects we'll discuss below under server log statistics. The tracking user statistics data is periodically analysed and sent to you.
Trackers tell you how many visitors you have had, if they are new or returning, where they came from including the domain and site name, and if it was a search engine which one and what key words were used, and once the user is on the site the statistics tell you how they moved round between pages on the site such as the entry page (the door way used), sequence of pages viewed, the page they exited on, and time spent on site. Summarised this starts to give useful information - what are they interested in, popular pages which might mean interest in the product or service, revisted pages, and it may indicate pages or navigation structures that do not work causing users to leave, prompting you to think of redesign, or identify major interest areas which can contribute marketing activitys.
Server log statistics are not exact for a number of reasons, for example dynamic IP connections or those coming from a large company just tells you the domain name they are coming from. The access could relate to a desk at a cyber cafe or office with 100's of users over a short period of time, or it could be a single visitor regularly coming back. So you could be double counting. Some tracking systems use more sophisticated methods such as putting a cookie onto the users PC having registered the user in some way, giving a name or email address for example to make the contact unique. A cookie is a small text file which is set to contain personalisation information. However some users switch cookies off, other companies regularly clean out cookies, so you may not be able to use that method and you will not be able to recognise new visitors from those returning back as each visit will cause your system to consider the visitor to be new. Back to double counting. Even if people have cookies enabled they can access your site using different machines and devices e.g. desk tops, PDA's and laptop again leading to possible inaccuracies in user numbers. Therefore the larger the business traffic and its benefit the more critical the tracking systems become and the solutions become more complex, a mixture of cookies and other methods to get users to physically log-on to the site where during registration you have obtained from them their name and maybe other details, for example. At this stage how ever, for the average maturity site, any indicative statistics will be useful.
You can also use some techniques to evaluate responses to particular promotions or marketing campaigns, for example from adverts, newsletters or similar primary media, by using an extended web address, to a specific new page which is an entry page to a visitor, or an extension to the web address which can be extracted and counted from the web statistics, these are often of the form www.anysite.com?iamfromthisadd , note text after the ? is ignored by the web server, so it used to pass additional information.
Despite the problems with the absolute accuracy of user data, statistics do provide useful information and it is relatively easy to get, and feed into decision making activities. This is important as sites become more active when the information can also be used to relate the number of simultaneous users to the web server and back office systems response times, identifying when enhancements may be necessary, and particularly when the data is used predictively with expected traffic growth. Also many site hosts and internet server providers have charges depending on the level of data a site sends over the internet, so information on actual and likely numbers, with the likely number of pages and hence size of data is important not only for costing purposes, but to ensure a service level agreement is reached with them so acceptable down load times can be achieved. Visitor statistics are used in these processes.
Link here to return to Beavers main web site evolution page or here to read about how to keep site visitors coming back, building a relationship with them, and producing business benefit, and how you should think about integrating your web site with your other business, computer and communication systems.