Lies, damned lies, and statistics

Or “Buh-bye Google Analytics”.

As most of the commentary on my previous post about some statistics from this site (as reported by Google Analytics), the figures seemed… wrong.

Even though the Google Analytics code is included on every page I serve, the two obvious causes were:

  1. People were blocking it (this seems more common than I realised)
  2. It’s broken in some undefinable way

Seeing as I have full access to my server logs, I decided to go directly to the proverbial horses mouth and see how the numbers work out from there.

The results are interesting to say the least.

Note: These statistics are purely for the same period – January 2008 up to the date of the this post. This means there may be some discrepancy between the GA figures in my other post and this one.

MPB (Most Popular Browser)

Google Analytics Logs
# Browser Percentage # Browser Percentage
1. Konqueror 42.55% 1. Firefox 61.2%
2. Safari 41.13% 2. MS Internet Explorer 15.3%
3. Opera 15.60% 3. Konquerer 6.7%
4. Firefox 0.71% 4. Unknown 4.2%
      5. Safari 3.6%

Bit of a difference, isn’t there?

Operating Systems

Google Analytics Logs
# Browser Percentage # Browser Percentage
1. Linux 45.39% 1. Linux 46.6%
2. Macintosh 36.88% 2. Windows 40.6%
3. Windows 11.35% 3. Macintosh 6.9%
4. (not set) 2.84% 4. Unknown 5.6%
5. iPod 2.13% 5. Sun Solaris <1%
6. iPhone 1.42% 6. FreeBSD <1%

Hmm. Some confusion there.

…and the moral of the story is:

Look before you…

No, wait, too many cooks…

That’s not right. It’s never work with animals, children, or a third party statistics package that relies on JavaScript. Especially when you have access to your own logs. So the real moral is don’t be lazy. Here endeth the lesson.


About this entry