Bill Kearney

Tracking RSS readership

If you want to track whether or not the links being referred to your website are coming from your RSS file then change the URLs you put in it.

For example, the links to my archive pages here are just plain old URLs. They're formatted thusly:

http://www.ideaspace.net/users/wkearney/archives/000064.html

I could just as easily tack on a bogus parameter like this:

http://www.ideaspace.net/users/wkearney/archives/000064.html?rss

My weblog analysis softeware could simply look for that parameter. This would let me track whether or not the use of a link came from an RSS reader or just from another web page. Depending on how your web pages are served you'll have to do different things. Especially if your RSS feed comes from a dynamic script not just a static file. Of course you'll also have to be using programs that let you change this. MovableType and any others using templates let you, Radio does not. You can go hack the output code but tread carefully.

So when folks like Jason Kottke raise concerns about it I simply tell them to take matters into their own hands. Change the URLs in your outgoing RSS file to help you do your own tracking.

One reason referer URLs where hijacked UserLandave Winer. Their Radio product started doing this and many others have naively followed it's bad example. Granted, they seem to do that in droves, but I digress... The rationale behind using the referring agent URL was one of vanity tainted by a false sense of privacy.

Let's look at how Radio got it's start, it was an offshoot of the Frontier product and it's Manilla web interface (a fine concept in it's own right). A Manila site can show you referral logs but it can't show you user agent or IP addresses. This allows them to show off stats on usage without compromising user IP address anonymity. This seems like a fine thing. One problem, however, where's an RSS reader coming from? And, to the vanity point, how does a maker of both sides of the game show off his numbers? Right, abuse the standards.

In order to show how many people using the RSS reader without reworking the referral logs it was decided to abuse the referring URL. That way the hit counts would reflect RSS reader (Radio) usage. This has been further abused by parameterizing it with user info on a community server. That deserves a whole other rant.

The proper place for an RSS reader to make it's presence known is the user agent URL. The referral URL would be better suited as the source of the item, also know as it's permanent link. That, however, is a whole other disaster.

And sites that want to track usage should update themselves to use the user-agent header. That's what it's there to do!

Meanwhile the logs fill up with useless entries. If you want better tracking then DO IT. Change your RSS URLs and filter them in your weblog analyzer.