contentious.com

RSS feeds are undoubtedly becoming an increasingly popular way for people to keep up with what’s new online. But just how popular are they becoming? That’s an important and tricky question.

Some recent articles and weblog entries have been touting the popularity of RSS feeds. These are great, and I’m happy to see them. However, I think there’s an important part of the puzzle missing from this enthusiasm: How might publishers figure out how many people are really accessing their content via RSS?

I think we need a statistical guideline that could help publishers make a rough estimate of RSS readership: average polling interval. So far, I haven’t seen anything quite like this, and it’s possible this may not be a good idea, but I suspect it might help.

Let me explain…

Counting hits to your site’s RSS feed file won’t tell you how many people read your content by RSS, because RSS feed readers “poll” (or check) your feed file at a set interval to see whether you’ve posted any new content.

For instance, if one member of your online audience has her feed reader set to poll feeds every four hours, then her feed reader would hit your feed file six times a day. Just by looking at raw site statistics, that might appear that you have had six visitors, when in fact there was only one. Furthermore, if you posted no new content that day, it’s questionable whether you could really even consider that a “visit.”

So far, you can’t directly track RSS subscriptions or usage. There are ways that you can track links clicked from within your feed, but so far there’s no fully reliable or easy way to track how many people are accessing and reading the content that you supply via your feed. I recently noted in a posting to the Poynter Institute’s weblog E-Media Tidbits that some marketing companies have debuted what they claim to be trackable RSS services, but I am skeptical of the accuracy or usefulness of the data they’re collecting.

Here’s my idea: If there was a fairly reliable statistic for how often the average feed reader polls RSS feeds, then feed publishers could divide total hits to their feed file by that number to get a rough estimate of how many people are probably reading a particular RSS feed.

Like I said, I haven’t yet found a source for that sort of statistic. (If you have, please comment below.) However, I would love it if some organization found a way to measure this statistic and publish it regularly. Wouldn’t it be great if you could look this up on ClickZ Stats (formerly CyberAtlas)?

That kind of statistical guideline would be useful context for news of the growing popularity of RSS, such as the March 9, 2004 weblog entry titled RSS Tipping Point by Chad Dickerson, Chief Technology Officer of InfoWorld Media Group). Dickerson wrote:

“Over the past several weeks, requests for InfoWorld’s Top News RSS feed have regularly exceeded the requests for our home page. …During the business day, we track hour-to-hour performance …and in any given hour, about 8 of our top 10 most requested files are RSS files. …Feels like a tipping point to me.”

True, BUT…. the average polling interval could provide a clearer indication of the relative popularity of InfoWorld’s RSS feed vs. its home page, right? I agree that this trend in InfoWorld’s raw numbers is significant, but having that extra context could turn that number from “significant” to “really useful.”

Mark Frauenfelder of The Industry Standard was impressed by InfoWorld’s claim, and I don’t doubt it will be widely circulated and touted. But really, until we can find a way to at least roughly estimate how to translate those raw hits into probably visitors, the true value of RSS to online publishers will sound more like hype than reality.

Why bother counting RSS subscribers? Speaking as someone who publishes on the Web, by e-mail, and by RSS feed, I’d love to get a handle on this for one big reason: it’s getting almost impossible to keep an e-mail subscription list from shrinking. Yet, at the same time, the raw numbers for usage of my RSS feed file are growing substantially each month. It would be comforting and useful to know whether, as e-mail readership declines, it’s being replaced by gains in RSS readership.

Thanks to the spam problem, people change e-mail addresses so frequently that their e-mail subscriptions often get lost in the shuffle. Every time I publish the CONTENTIOUS e-mail alert, I get a surprising number of bounces from addresses subscribed within the last six months! Even though I’m now getting more new subscribers each month to my e-mail alerts than at any time in the last four years, my e-mail list continues to shrink slowly each month – due mainly to bounces, not unsubscriptions!

When I surveyed CONTENTIOUS readers in November 2003, just over 12% reported unsubscribing from my e-mail alerts because they switched to my RSS feed. Also, since 71.1% of respondents reported that they receive my e-mail alerts, and 53.5% reported that they subscribe to and read my RSS feed, I think it’s fair to assume that there must be a lot of CONTENTIOUS readers who use both channels. I’ll do this survey again in a few months – and I will expect more and more existing and new CONTENTIOUS readers will report that they prefer my RSS feed to my e-mail alerts.

Thoughts? Would my idea for estimating RSS readership by using a statistical guideline work? Would it help? Does that statistic already exist, or is there some reason why it could not be measured or would not work?

Side Note: On Nov. 24, 2003, Gary Lawrence Murphy noted in his Teledyn weblog entry, The End of RSS, that while RSS solves some bandwidth problems it presents others, due to the polling nature of how this medium currently works. He proposes some options to consider. It’s a bit technical, but worth at least a quick scan in order to be aware of this potential complication.