Wednesday, July 18, 2012
This post probably won't interest general readers, but if you're a blogger, and you use Sitemeter's visitor tracking software, you may find this useful.
I've written before about issues with Sitemeter. Briefly, it has several shortcomings. It doesn't measure most visits via a blog's RSS feed, and seems to have difficulty providing an accurate tally of user and page view counts at the best of times. I find I need to check Sitemeter's numbers against those provided by Google Analytics, Google Feedburner and Blogger Stats to get a true all-round picture of blog traffic. (Blogger Stats in particular is extremely accurate with regard to page views, which is the basic metric it tracks. It logs access from any source, via the Web, RSS feed, e-mail and other links, and even bots crawling your site. You need to tell it not to log your own visits, of course.) For the past year or two, based on extensive comparative calculations, I've found that I have to use a multiplier factor of 1.41 when assessing Sitemeter's figures. In other words, if Sitemeter tells me I've had a thousand visitors, or two thousand page views, I need to multiply that by 1.41 to get a reasonably accurate picture of how many there really were.
To my surprise, in late May Sitemeter began to record what looked like a precipitous drop in my readership numbers. The decline was very marked, and by the end of June it had me concerned. Had I offended someone, or burned out a portion of the Internet with too many Doofi Of The Day?
I spent a week or so researching what was going on, with particular reference to the other measurement facilities mentioned above. Turns out my blog traffic hasn't changed much. On the other hand, Sitemeter appears to have done something to their measurement system. Over the past 30 days, checking their figures against the other metrics, I've discovered I now have to use a multiplier of 1.61 to get their numbers anywhere near what the other sites are reporting. In other words, they're now under-reading my traffic by an additional 15%, over and above their existing under-recording error. My Sitemeter visit counter currently reads just over 3 million, but according to other authoritative sources it should be at least 4.5 million, and probably closer to 4.8 million. That's a very significant error margin.
I don't know whether other bloggers have noticed something similar, but the numbers don't lie. I plugged all of them, from all sources, into a spreadsheet covering the past couple of months, and the trend is very clear. If you use Sitemeter to measure your blog's readership, you should double-check its figures with other, more reliable sources. The only reason I still use it is that Technorati and a couple of other sites use its measurements to assess relative blog ranking; and since it probably under-reads almost every blog to a similar extent, I presume the errors balance out.