Since I started blogging I’ve always tried to keep an eye on my traffic statistics, referrers and other activity on the site. It’s nice to know who’s reading what, and see what is effective in creating interest and attracting readers.
I moved the site over to WordPress earlier this year, and put in some more determined effort into blogging. Posting regularly to the blog has been great for traffic and search engine ranking, but I’ve never completely settled on a single stats package to use.
Over the past month or so I’ve had three different free statistics packages running on the site, and the comparison has been interesting:
WordPress Shortstats plug-in

Spotted this one on Neville Hobson’s blog – it provides some simple stats that show up in the WordPress dashboard area. No installation required – just upload the plug-in and activate.
Google Analytics

An extremely powerful stats package, based on the Urchin software that was acquired by Google. Required a short piece of code that was easily added to the WordPress template files.
Statcounter

Another powerful package, although the free version limits the amount of historical data that is stored for trend analysis and the like. Again required a small piece of code that is added to the WordPress template files.
I have tended to check the Shortstats data most times I log into the WordPress admin area. I’ve beeing checking the other two services less often, although Statcounter does have a regular email report which does act as a reminder. What would be really handy would be a regular RSS feed (Performancing does this, but when I tried this service the results seemed to bear no resemblence to those on the other services – they just looked very wrong).
However the real trap I fell into was looking at the number of unique visitors I’ve been getting. This metric is notoriously hard to track – it’ll never be truly possible to track numbers of individual people who visit websites, because of people sharing computers, dynamic IP addresses, cookies, cookie blocking and a host of other factors.
But unique visitors is still a useful measure to track over time to spot trends and get a general idea of traffic. There are two main ways of identifying unique users: IP addresses and/or cookies (good metrics summary here).
It’s a fact that unique visitors based on IP addresses will always be more than those based on cookies, but even I was shocked when I compared the stats packages I’m using. I’m guessing Shortstats uses IP addresses, although I couldn’t find anything that actually confirmed that, and Google Analytics/Statcounter definitely use cookies (confirmation found here and here).
The chart below shows how the three services compare when measuring unique users:

Google Analytics consistently picks up the lowest number of unique users, with Statcounter averaging 18% more unique users than Google. The trends seem relatively similar on these two services.
The real shock was that using IP addresses alone, Shortstats tracks an amazing 514% more unique users than Google.
In the future I will have to remind myself to ignore the unique user statistics in Shortstat. I won’t delete the plug-in as it does give a quick reminder on recent resources used and recent referrers. But for more serious analysis and more accurate unique visitor figures I’ll still be using Statcounter and Google Analytics – which of these two is better remains to be seen.