|InfoVis.net>Magazine>message n║ 174||Published 2005-10-17|
|TambiÚn disponible en Espa˝ol|
The digital magazine of InfoVis.net
Internet is plagued with examples of carefully designed websites where, nonetheless, the users get lost, don't find what they are looking for, despite its existence or, even worse, they look for something that should exist in the website but doesn't.
On the other hand many web managers don't know in most cases what their users are doing within the web, including whether they find what they look for or if there are concepts worth placing in the website since the users look for them. Without the knowledge of the impact of our marketing campaigns in an e-commerce site, we barely will be able to make it progress in the right direction.
AT the issues numbers┬á 65, 66 and 67 we already commented on some of the solutions to analyse our website logfiles, monitoring its traffic along with the importance of visualisation in order to quickly understand the results. At that time (november 2001) there were very few graphic applications showing the results of logfile analysis. That still holds true nowadays.
The structure of a logfile is extremely simple. Each time someone downloads an element of the website, like for example an HTML page or just an image, the server writes a line in the log file. That line can adopt one of several formats but usually it looks like this:
The important thing is that, despite being so elementary, the statistical study of the aggregation of the many requests that the users' browsers issue to the server allows us to know a lot of derived information from these apparently simple lines.┬á Among them, at least apparently, you can derive the number of pages served per day, month or unit of time you want, sites that link to our web or redirect traffic towards us (referrers), the words most commonly searched for in our web and a long etcetera of findings.
If we review the specifications of many server logfile analysers we'll see that knowing the number of unique visits, visitors, for how long they have been they standing at our web or even on a particular page appears to be extremely easy for those products.┬á
Nothing further from the truth. Most of this information is of reduced reliability due mainly to two reasons, among many others (see the discussion about what you can know and what you can't when analysing a logfile)
Summarising, attending just to the logfile analysis, you can't really know the number of visitors, the number of visits nor the identity of the users. You can't establish reliably the paths they have followed within the website. As a corollary of this you can't know for how long they have been reading a web page either.
Nevertheless, this doesn't mean that the information that can be derived from logfile analysis, although incomplete, isn't valuable.┬á
In the end, you can't sell the idea that logfile analysers find visits and "unique" visitors, paths of the same or even the operating system they use, as some manufacturers appear to want us to believe.
Nevertheless once you are aware of what a logfile really says and what its limitations are, we have powerful tools to understand our website and the use our users make of it, opening up a door to decisions that can improve its profitability.
Links of this issue:
Subscribe to the free newsletter