Website Statistics
Website Statistics
There are lies, damned lies, and statistics.
This was a comment by a famous politician, a rather cynical view of how information was distorted to further political ends. However, statistics are a very useful tool provided that they are interpreted intelligently and honestly.
Hosting Service Statistics
Most Hosting Services provide a statistics package as part of the deal, some even provide two or more analysis programs. My experience of these is that, while more than adequate for a personal website to keep track of how many visitors and hits you get, they are less than adequate for a serious analysis of a commercial website. First, let's clarify a few important definitions:
Hits, Page Impressions, and Visitors
People talk a lot about how many hits they get without really understanding what these are. Hits are often regarded as synonymous with pages accessed or even visitors. They are not! Let's look at a simple example:
You have a web page with 10 graphics. Every time someone visits this page you will get 11 hits - one for the page and one for each of the graphics downloaded. So, if you have 10 visitors in a day, and each visits that page once, you will have 110 hits for that page that day. More graphics, more hits!
Page impressions are the number of times the page has been viewed, and this is a more significant measure of the importance of that page than hits. Hits do have their uses, however, particularly when the hit is to an image which links with something else - another page or website - and can be used as a measure of the success of that link, provided you have statistics for the number of hits for each image.
Let's now take Visitors. Your online webstats tell you that you have had 10 visitors in a day. Does this mean that you have had 10 unique visitors? Or 2 unique visitors who each visited your site 5 times that day? The distinction between unique visitors and repeat visitors is very important: the growth in the number of unique visitors over a period tells you how popular your site is getting and the effectiveness of your promotions, and the number of repeat visitors indicates that these people found your site interesting and keep coming back. A decrease in the number of unique visitors shows that your Search Engine visibility is going down and people are not finding your site as easily, and a decrease of repeat visitors is a warning that your site content is getting boring or, more likely, that you have not been adding fresh content for some time - which can also affect the number of unique visitors.
Statistics you need
In addition to page impressions and visitors (unique and repeat) there are some other statistics which will help you to guage and improve the performance of your website.
- Referrals are one of the most important. These are the other sites which have directed visitors to you, such as Search Engines. If you have exchanged links with other sites, then visitors who click on links to your site will also show up in these statistics. This is an important way of analysing the source of traffic to determine what brought them to your site. A related statistic is:
- Search Terms. These will indicate what people were looking for, and may suggest the inclusion of more content relating to the search term.
- Paths. These are the audit trails of paths taken through the site, one page leading to another, which is very useful in guaging interest in different parts of the site. For instance, I look for paths from my Home page to my Portfolio page to my Order page, which is the path that leads to business. I often find that people have come directly to my Goodies page, having bookmarked it earlier, and leaving the site from that page too. An analysis of entry and exit pages can reveal both interest and boredom, particularly when trying to measure the success of a promotional campaign which is page specific.
- Countries. If you are selling internationally, an analysis of visitor geographical locations will indicate whether or not you have reached a target market, and may reveal an interest from a totally unexpected location which may open up a new market. Or, the stats may indicate that you should reinforce sales in a particular area by adapting content to suit the largest market.
- History files are also very useful, showing you trends for all the above statistics. This is particularly useful when new content is added, as it will provide confirmation of the interest in those pages, over a period of time, as a visible increase in traffic to those pages.
- Search Engines and Robots. The statistics on visits from the SEs and their robots are useful as an indication that they have an interest in your pages, especially if they visit regularly, and indicate that your site is being indexed and will be found through searches. The hits from these can also be used to modify the overall totals, so that human visitors can be separated from the electronic ones.
Software you will need
If you plan on examining your statistics closely you will need two things: a statistical analysis package and a host who give you access to "raw logs", something to bear in mind when choosing a host. You should also check with your host how often the logs are "rotated" or cleared down. Ideally you will have a host who allows the logs to grow indefinitely, leaving the decision to clear them upto you. The next best choice is a host who clears the logs on a monthly basis, on the same day of the month, giving you a chance to download them before they are cleared. Avoid hosts who do not offer one of these systems, as you will lose logs which could prove vital in setting your promotional strategies. You will also need an FTP (File Transfer Protocol) package to download your logs, unless, as is often the case, the host provides the download facilities. Search on Google and Yahoo for "Log Analysis software" and "FTP software" - you will find loads of affordable, and even free, packages.
If you have to download you logs on a monthly basis, you can still perform an analysis over several months by "stitching" the logs together - just open the first file in Notepad or WordPad, then the second, select and copy the contents of the second and paste them to the end of the first. Then repeat with the other files, and when finished save as a new file showing the period covered (e.g. 3Qtr2003.txt). Some log analysis programs allow you to specifiy multiple logs for analysis, saving you this chore.
Representation of Statistics
The easiest to follow representation is graphical, and a picture of the statistic over a period of time will immediately show both popularity and trends. This is preferable by far to tables of numbers. When evaluating statistical packages the format of presentation should be taken into account, in addition to the variety of items analysed.
Hits and Visitors are not enough.
Top