August 13, 2007

Your Blog Statistics May Vary, Widely

Try as I might, I can't settle down with a single reliable service to track Web statistics. While I have been very happy with SiteMeter's premium offering for more than a year now, I have found the service often misses a great number of visitors, resulting in significant undercount - made more glaring by the addition of statistical tracking capabilities to other services, including Feedburner and MyBlogLog, giving me multiple sources from which to compare my own statistics.

Last week was among the most-heavily trafficked periods for louisgray.com since the blog's initiation more than eighteen months ago, led largely by a pair of referring links from popular Macintosh destination site, MacSurfer.com, who posted references to two of my Apple-focused stories, "iPhones Aplenty in Silicon Valley Geek Mecca" and "Attention Newbies: Not All Apple News Changes the World". But while it was clear site traffic was up much higher than normal, the varying statistics trackers had significant variations in how they accounted for the spike.

For the purposes of comparison, I'll highlight traffic from Tuesday to Thursday of last week (August 7-9):

ServiceUnique Site Visits Total Page Views
 Tues.Wed.Thurs.Tues.Wed.Thurs.
SiteMeter244255320314325375
Feedburner552542617684625716
MyBlogLog235261310317329386

Looking at the data side by side, it seems Feedburner, for whatever reason, counts upwards of two times the visitors as do MyBlogLog and SiteMeter. While each of the services accurately reported the most popular pages, and most popular referring sites, the total counts vary. This could be because some visitors have voluntarily blocked "*.sitemeter.com" from their ad blocker software, or it could be that one service handles duplicate visits or self-visits (a.k.a. mine) differently than others.

Additionally, running the downloadable software application "Summary" against the same log files confuses the issue even further. While SiteMeter and MyBlogLog report between 230 and 320 or so visitors from Tuesday to Thursday, Summary says those days were more in the range of 2,000 individual visitors, serving nearly 3,000 pages daily. Like with Feedburner, I believe those stats to be high. Unlike the other services, however, Summary reports "hijacked" page graphics, including the ANtics comics, as individual views, and may report visits from search spidering software, like Google or Yahoo! on par with "real" people.

All I really know is that blog traffic is consistently in the hundreds of visitors a day, that it's more likely toward the 200-300 person range than 700-2,000, and that there may be no one right answer to "How many visitors did I get today and how many pages did they read?" What statistics package do you prefer, and am I overlooking anything that could make this process more simple?

3 comments:

  1. Louis:
    I've experienced the same inconsistency of statistical tracking as you. I've tried and am still using a number packages including StatCounter, Google Analytics and Urchin (which is provided by my host). They all report different counts.

    There seems to be a technology gap here. There must be an absolute answer. I'd like to just use one package...but which one is most accurate? It's anybodies guess.

    ReplyDelete
  2. I have just been looking at Google Analytics and the server logs. There are even inconsistencies between those two but I figure the server itself has got to be closest to the source.

    ReplyDelete
  3. At one point, I had Google Analytics running here as well, but during one of my template changes, I lost it. I haven't really felt the need for "yet another" statistical analysis tool. I'd believe the server logs (through Summary) would be the best example, but as SiteMeter and MyBlogLog are close enough, that's a good metric. I just wish I could believe Feedburner, as their numbers are consistently high!

    ReplyDelete