Traffix
August 25, 2010
Since there are several traffic meters running here at DM on Scientopia, I was curious as to how the GoogleAnalytics, Sitemeter and WordPress in house statistics compare.
The WordPress views and the SiteMeter pageviews seem to match up quite nicely. As usual, SiteMeter visits run in good correlation with, but lower than, the Sitemeter pageviews. Surprisingly, Google Analytics visits are a bit out of register with other numbers.
Takehome message would appear to be that you can’t readily compare GA numbers to the other ones, right?
DrugMonkey 
August 25, 2010 at 7:07 pm
Yeah, I have read that several places (that GA reports quite differently than most other services). Not sure why.
LikeLike
August 25, 2010 at 10:56 pm
And yet Google Analytics seems to be the gold standard in some places. Curious.
LikeLike
August 26, 2010 at 3:09 am
Dude, in what universe does “a bit” mean “three- to four-fold”? Maybe these Google numbers are fucken low because those are the numbers on which they base their payments to bloggers for Adsense ads?
LikeLike
August 26, 2010 at 4:09 am
Do any of these include activity from RSS feeds, where you don’t ‘directly’ visit the site? Just curious.
LikeLike
August 26, 2010 at 4:27 am
Walt – of Walt at Random – compared a bunch of these as well as one running on the server. GA was always significantly lower – like a third or quarter. Part of the problem he found is that GA is intolerant of bad html and it also doesn’t work if the viewer has cookies disabled. See his discussion here: http://walt.lishost.org/2010/02/google-analytics-v-urchin-5-a-metrics-quandary/ and here: http://walt.lishost.org/2010/02/a-metrics-update/
LikeLike
August 26, 2010 at 8:12 am
As with all statistical analyses, you should choose the numbers that best suit your hypothesis and remove the others.
LikeLike
August 26, 2010 at 9:27 am
Precisely, PiT, precisely.
actually, so far the GA/Sitemeter difference is proportionally much, much larger than we ever experienced at the Sb blog. It will be fascinating to see how this difference tracks over time since we’re only a month in here at Scientopia
LikeLike
August 26, 2010 at 11:03 pm
so then I guess I’m not the only one who primarily reads this blog when procrastinating at work (as opposed to free time on the weekends)?
LikeLike
August 27, 2010 at 10:00 am
Think of each one as a slightly different experimental setup – perhaps Google Analytics uses krebs buffer with higher glucose content, or SiteMeter incubates its cells at room temp while WP stats incubates at 37*C…
The point is, you couldn’t compare absolute data from different experiments, but you can still make plenty of valid conclusions about each experiment on its own, and hopefully, the overall conclusions will support each other and together provide a more complete picture.
Here’s a bit more info on these “experimental” differences:
LikeLike
August 27, 2010 at 11:44 am
aaaha! Got it. My mistake.
I was indeed pulling GA’s numbers from the graph so I guess that is visits, not pageviews. That explains the difference. When I made the comment about the Sb stats, GA vs SiteMeter I was talking about monthly pageview numbers, not pulling from the graph.
LikeLike
August 27, 2010 at 1:38 pm
Some other things that might affect results:
Javascript based analytics packages will record 10-15% less than packages based on server logs, and I’d expect Google Analytics to record slightly more than Site Meter, as the code for that is in the pages head section, while Site Meter’s is towards the end of the page, so GA will record visitors who leave before the page has finished loading, whereas SM won’t.
Gotta love “internet science”…
LikeLike