There is a not at all surprising disconnect between what Google analytics reports for this blog and what logfile analysis reports. In light of the FTC's push for an "opt out" button for tracking, I'm guessing the javascript-method of website tracking is going to be less effective.
Operating system:
Interestingly, log analysis also breaks down the OS versions in use. I'm happy to note that the large majority of the Linux users are Suse variants. XP users still outnumber Vista/Win7 users.
Browser:
The other/unknown is likely the log-analysis engine's inability to figure out some agent strings. At a guess it's really under-reporting all the Chrome users out there. Even so, there are significant differences between the two. To me this looks like Firefox users are much more likely to be using NoScript.
And finally, once browsers start scrambling the User Agent String, even that will be not useful for this kind of tracking.
Operating system:
Google Analytics |
Log analysis |
|
Windows |
66.7% |
58.7% |
Linux |
23.8% |
22.9% |
Mac |
9.2% |
3.7% |
Other |
14.6% |
Browser:
Google Analytics |
Log analysis |
|
Firefox |
37.3% |
44.3% |
Internet Explorer |
22.22% |
21.9% |
Chrome |
29.37% |
9.8% |
Opera |
3.17% |
6.7% |
Safari |
4.76% |
1.6% |
Other/unknown |
3.18% |
15.7% |
And finally, once browsers start scrambling the User Agent String, even that will be not useful for this kind of tracking.
The conclusion that your log analyser is out of date is more likely the cause for the majority of the differences than people are opting out or not using javascript. You have an adnormally large "other" category compared to the Google Analytics stats. There is a difference between hits that will be recorded in your logs vs GA but I don't think it's anywhere near what you're reporting based on your log analysis.