Having Faith in Web Analytics

Do you put your faith in your web analytics tool? Which stats program do you trust to give you an accurate reflection of traffic to your website?

Maybe you have AWstats or Webalizer on your servers, both of which draw conclusions about visitors from the raw web logs generated by your server. Maybe you’re running a snippet of script that reports back to a third-party system like GetClicky. You may even remember the old hit counters webmasters in the mid-1990s used to proudly display on their homepage. Of course, hits mean different things to different people, usually a single person visiting a single page on your site will generate several “hits” because each graphic, script, and button is counted, whereas they actually represent only a single unique visitor.

Some webmasters claim that many of the current web analytics and stat tools over-estimate or under-report traffic significantly depending. Some apparently inflate your unique visitor numbers sometimes by as much as three times, others tell you that you’ve got far less traffic than you actually have. Some look only at visiting IP addresses and cannot always distinguish between bots and real visitors.

Those who set no score by server-based stats programs suggest that alternatives such as the likes of Google Analytics, GetClicky, and StatCounter, are better. Supposedly they reflect your site’s traffic much more accurately by ignoring bots and search engine spiders and by hooking into actual pages visited by people surfing the net. None of these tools provide a definitive answer because there is no definitive answer.

Unfortunately Google Analytics also has a reputation for under-reporting visitor rates. But, Dan Grossman made several interesting points about the javascript-based analytics tools, which includes Clicky and Google.

To use these tools you have to add a snippet of javascript code to your site’s template so that it runs each time a page is displayed. But, says Grossman, this means that: those tools will miss all visits by people who are using a browser that does not run javascript (either deliberately disabled or else not part of the browser). Moreover, while a visitor may read your page, the javascript may not load by the time they hit the back button or a link on the page, and so that view won’t count either. In contrast, Webalizer and Awstats will show that as a pageview regardless because it represents a call to the server.

None of these tools are going to tell you exactly how many people visit your website, and really you shouldn’t care. What you should care about is the engagement with your readers and/or your bottom-line sales, depending on whether you’re providing information or selling something.

However, keeping an eye on your raw logs, via AWStats or something similar, as well as running GetClicky or Google Analytics can let you see some quite useful information beyond raw hit rates, including referring sites and country distribution, browser type, screen resolution, all the usual things. Moreover, it can let you see trends in visitor rates, which might reflect changes you make to your site content or layout…and whether you just got a Digg, for instance.

It’s a good idea to have some idea of visitor numbers and how these change over time and how they reflect on the aims of your website whether commercial or not.
But, just remember, when that web stats tool is claiming so many million visitors, whether to your site, the BBC, or Digg, you have to ask yourself who is it counting?