Basic website metrics
Thu, Aug 18 2005
In the last eBusiness News I talked about the importance of using metrics and analytics to gain an understanding of how well your site is performing, and this week I'm going to give a gentle introduction to some simple metrics that you can obtain fairly easily from your webserver logfiles. These basic metrics are some of the fundamental building blocks that are used to derive all sorts of interesting information, so make sure you understand these before working up to more complex methods of site performance analysis. As I explain in my book How To Build A Website And Stay Sane, it's critical to understand what the term "hit" really means. Most people have a vague idea that it's a measure of how many people have visited a site and that bigger numbers are better. Hit counts can be totally misleading though unless you know what you're really measuring. A "hit" is a request from a web browser to the web server asking for an item on the server. That item could be an HTML file, or an image, or anything else on the server. Retrieving a web page typically involves one hit for the HTML file itself, then one hit for every image on the page and probably extra hits for other things you aren't even aware of including CSS files and alternative images to display when rolling your mouse over a navigation button. Every single one of those items counts as a "hit", so when one person visits one page it could easily count as 15 or 20 hits. Some pages might consist of only a plain HTML file and therefore count as a single hit, while other pages might consist of 100 items and therefore count as 100 hits for just one person to view one page. What does that all mean? Basically that you can't just look at "hits" and expect the number to be meaningful unless you understand the context it's in. When someone says "Our site had over a million hits last month" it's not impressive: it's actually pretty meaningless, and probably just goes to show they don't really understand what they're saying. That million hits could have been less than a thousand unique visitors looking at a dozen graphically-loaded pages each. Still a good number of visitors, but not quite the crowd of one million people they probably imagine! So, a "hit" is just a request for an object. That's all. Once you understand that you can start talking about hits in a meaningful way. For example, you can talk about hits to a certain image and know that you're talking about how many times that particular image has been requested. Or you can talk about hits to an HTML file, and know you're talking about how many people have viewed that page - which leads to a very useful base metric called "page views". A page view is one person requesting one page. The page view might consist of one hit or 100, but it doesn't matter: the important thing is that one person viewed one page. Using page views to compare traffic to different sites or sections of sites is far more useful than talking about hits, because it means you're comparing apples with apples. A page view is a page view, no matter how many hits it involves. To really get a feel for what I'm talking about make sure you have a look at the traffic report for your site. Your hosting provider should provide you with a secure location where you can view a report updated every week or so. Find the section of the report that covers page views and have a look at the top item on the list: that's the most popular page on your site. Other basic metrics you need to understand include "visitors" and "visits". A "visitor" (sometimes referred to as "unique sites" in traffic reports) is a distinct computer used to look at your site. The figure is obtained by recording the network address of each computer that connects to your site, so if two different computers connect to your site it will be reported as two visitors. Obviously this is a crucial figure when it comes to the performance of your site, because the visitor report shows you how many different people are actually looking at it. The more visitors the better! There are factors which can affect the accuracy of this report but it's still a good baseline metric to track. A "visit" is when one visitor comes to your site and views a number of pages. One person may look at one page and then go elsewhere: another person may click through dozens of pages on your site. In both cases it's called a "visit". Visits can't be measured directly by the server because it only tracks individual page views, so a visit report is really only a best-guess by your traffic report software. The figure is derived by grouping together page views by unique visitors and then treating that group as a "visit", so if for example you have one visitor come to your site and look at several pages on one day and then come back again a few days later that would be seen to be two "visits" by one "visitor". "Page views", "visitors", and "visits" are the most fundamental figures you need to know when it comes to your site. Working up from that basic information it's possible to obtain an enormous amount of information about how well your site is performing, which in turn will show you what you need to do to make it more successful. For example, you can divide the "page views" figure by the "visits" figure to derive a figure for "average page views per visit". A small number of page views per visit means people aren't staying on your site for very long, while a large number means they spend a lot of time going through it in detail. Information like that is pure gold when it comes to understanding what your site is really doing, and in coming weeks I'll explain some of the other interesting information you can derive from your site traffic report.