UPDATE 10 January 2010: Use the technique described here on your own RSS feeds at YourStats, an RSS4Lib tool. Upload your web server’s log file and it will provide the number of readers based on the log file.
A thread on Web4Lib about measuring RSS usage through web logs made me realize how tricky this is. Aggregators (and browsers, such as Firefox, Safari, and IE 7) all request RSS feeds from your server, often several times a day. It is hard to tell how much your feed is being used — the RSS feed for this blog, http://www.rss4lib.com/index.xml, was accessed 19,455 times in October. Which sounds impressive, right?
However, that means that a constellation of individual web browsers, news aggregators, and search engines was checking the feed once a day, once an hour, once a week… Or at some frequency.
I know how many Bloglines subscribers there are (334 as of right now). But I can’t keep track of how many are reading this through Yahoo!, Google, NewsGator online, or this, that or the other aggregator.
Looking at the detailed web server log report (which is generated by my host using Analog), I see that some aggregators add the number of users they are collecting data for — basically, a subscription report. So I can see, in a recent month, the following details:
Bloglines/3.1 (http://www.bloglines.com; 320 subscribers) NewsGatorOnline/2.0 (http://www.newsgator.com; 7 subscribers) AttensaOnline/1.0 (http://www.attensa.com; 1 subscribers) Feedshow/1.0 (http://www.feedshow.com; 1 subscriber)
(This is for several different feeds for several different RSS services — taken from my entire server report, not just for the main feed for RSS4Lib.)
I also see the hits for all of one kind of web browser — in this example, Safari, get lumped together as one browser type, “AppleSyndication/54”. Different versions of Safari have different browser types, so I also see “AppleSyndication/53”, for example.
In short, it’s very hard to gauge readership — to separate reads from aggregator or browser “are you updated” hits. This is doubly true since so many people, myself included, read the full text of a post within the aggregator and rarely click through to the site where “spider” hits and “user” hits can be separated, mostly, by a good web log analyzer application.
P.S. I find this amazing, but this is the 100th post on my blog…. Happy “centennial” to me!
Update — 21 Feb 2007 Google Reader’s crawler, Feedfetcher-Google, now includes a subscriber count when it grabs your RSS feed. In my log file, a sample line looks like Feedfetcher-Google; (+http://www.google.com/feedfetcher.html; 133 subscribers; feed-id=1495776793707971617). Thanks to Taming the Beast for this tip.