I've long been suspicious of some of the stats that certain companies and bloggers push. We all know by now that industry stats tool alexa.com must be taken with a grain of salt, but until today I hadn't realised the extent of how a site's own measured traffic can be exaggerated.
You see recently I've been using Urchin (owned by Google) web stats as part of my MediaTemple web hosting arrangement, along with statcounter.com which I've used for years. I've been noticing that Urchin page view stats are MUCH higher than statcounter.com - in fact 5 times higher! This difference has been bugging me and I wanted to find out why. Today I came across this interesting thread which explains all:
"At the raw, basic level, Urchin counts pretty much the same numbers as any web server log analysis program will do (much prettier). Every page read by a browser is counted as a page view; different IP addresses signify the individual visitors. That's not very accurate because robots and crawlers 'read' pages, too, and show up in the count (visitors and pages) by different web server log analysis programs, including Urchin. Robots and crawlers don't read Javascript."
(emphasis mine)
Products like statcounter.com, by comparison, only count page views from browsers with JavaScript enabled - which gets rid of the robots and crawler hits.
So in a nutshell, Urchin stats can be highly misleading. Although it must be pointed out there is an advanced version of Urchin that uses something called UTM, which is Urchin's version of a Javascript enabled count. From the above thread Urchin with UTM means:
"...visitors are cookied, tagged, etc. Then, both log data and JS data are combined in Urchin's reports. Without UTM, Urchin is just counting server log stats."
I wanted to bring this up because I've always used statcounter.com when I tell people my page views - when I'm looking for advertising, or trying to get into a network, etc. But I have a feeling other people may use Urchin or similar log stats when they're promoting themselves. Which would mean my stats don't compare well to theirs.
I've certainly heard some bloggers quote extraordinary stats in my time - and have been skeptical without quite knowing why. Well now I know and I think others should too: sometimes people promote their sites with stats that are grossly exaggerated. If you want REAL stats, you need to get their statcounter.com stats or the advanced Urchin with UTM ones. There, now I have that off my chest ;-)
Listed below are links to blogs that reference this entry: On dodgy web and blog stats.
TrackBack URL for this entry: http://www.readwriteweb.com/cgi-bin/mt/mt-tb.cgi/2629
Interesting article from Read/Write web on the reliability of web statistics. Read More
Comments
Subscribe to comments for this post OR Subscribe to comments for all Read/WriteWeb posts
Even more interesting: I'm using Measure Map and stat counter simultaneously. I was previously aware of the php/javascript/bot problem, but this is entirely different. For tuesday, statcounter records 80 unique hits (Sad, I know), whereas measure map has a slight 49 unique visitors. Care to find why this is?
Posted by: Glen C. | March 9, 2006 6:24 PMAWStats filters robots away based on User-Agent strings. So it's a fine choice too.
Posted by: Sergey Schetinin | March 10, 2006 2:30 AMI could be wrong, but isn't Google Analytics the UTM version of Urchin?
Posted by: Ted | March 10, 2006 7:57 AMYup, it is. I use it on my site. Javascript tracking only.
Posted by: Walker Hamilton | March 10, 2006 3:02 PMAnother thing I've noticed about Urchin stats is they also count an rss hit as a hit. So if you have a lot of subscribers to your feeds they will show up in the first level of Urchin stats. At least for Apple Matters this is why the stats looks so good at first glance! Only upon filtering out the xml feeds in Urchin can I get a more realistic number. I imagine the same thing is happening for you too.
Posted by: Hadley Stern | March 11, 2006 5:39 AMHadley, yes RSS hits are a huge part of my Urchin stats too.
Glen, I too have noticed MeasureMap counts slightly lower than the other (JS) stats services. I'm not sure why...
Posted by: Richard MacManus | March 11, 2006 8:31 PMI found an article on this subject:
http://counterguide.com/article/6_log_analysis_and_javascript_counting_mutually_exclusive.html
Posted by: Tanskelem | March 22, 2006 9:05 PM