Last Saturday, there was a gasp of collective horror in the blogosphere as FeedBurner subscriber stats plunged for many sites. Today, it’s happened again. Don’t panic! Your subscribers are probably all still there, with Google Reader to blame for the missing numbers. Below, our comprehensive guide to how FeedBurner compiles subscriber stats explains all, today’s glitch, and why those occasional plunges happen.
Last weekend’s drop of Feedburner subscriber numbers by as much as half was a temporary glitch. Google Reader didn’t report figures, and all was back to normal the next day. Today’s drop appears to be the same issue. Joost de Valk for example, notes how Google FeedFetcher stats (which report combined Google Reader/iGoogle subscribers) are missing. I see the same.
The panic when stats go awry underscore how Feedburner is a vital tool for many bloggers, as it’s one of the only ways for them to know how many readers subscribe to their blogs. Indeed, it serves as a type of preferred currency to assemble some top blog lists. But what underpins that currency?
Readers subscribe to blog feeds using a number of feed readers (some of which visibly report numbers; some of which don’t), and it would be impossible for a blogger to keep track of the number of subscribers for each feed in each system manually. Few, if any, comparable services exist, and according to the Feedburner home page, over 631,000 publishers have burned over 1.1 million feeds so far.
Even after the explanation after the first drop, many blamed the glitch on Feedburner itself, rather than realizing Feedburner simply reports numbers that it’s given. Last week, I talked to Rick Klau of Feedburner to get the scoop on exactly where the numbers come from and why fluctuations (both the panic-inducing ones like Sunday’s and today’s, and more minor ones such as regular weekends) come about.
Burning a feed What does it mean to burn a feed? First, let’s back up and talk about what a feed is. A feed is a delivery method for your content (generally blog posts). Most blogs have a feed available by default, and in fact, many blogs have multiple feeds. WordPress, for instance, provides at least three feeds with a default installation. Why so many feeds? Several feed formats have emerged over the years (notably Atom and RSS), and much like the Betamax/VHS conflict (or, if you’re not old like me, the HD/Blu-Ray debate), no one was sure which would emerge as the leader. Unlike the tape/DVD wars, however, feed readers have decided to accept them all. You really only need to provide one version of your feed because no matter what format you use or what feed reader your visitors use, all will be well.
A visitor to your site can “subscribe” to your blog by adding the feed to a feed reader. Two major types of feed readers exist:
- Web-based services, accessed via a web page (such as Bloglines, Google Reader, and My Yahoo)
- Standalone aggregators, accessed from a desktop application (such as Outlook or iTunes)
Your content is delivered by “subscription” to the feed reader, which serves it up to users. You can choose to make either full feeds (that contain your entire post) or partial feeds (that contain only a portion of your post and require the user to visit your site to read the rest) available.
When you “burn a feed” with Feedburner, you create an account and provide the URL of your blog. Feedburner then gives you a feedburner feed URL, like http://feeds.feedburner.com/YOUR-FEED-NAME-HERE. (See Stay Master Of Your Feed Domain for information on how to create a feed through Feedburner that uses your domain instead.) You should then provide this URL as your feed location for visitors and should redirect all other feeds to this one. (As noted earlier, your blog may come with several feeds by default. With WordPress, you can use a plugin such as Feedburner Feedsmith to ensure all variations of the feed are redirected correctly.)
From that point, Feedburner reports daily on the number of subscribers to any variation of your blog’s feed. This number fluctuates and in particular tends to be lower on weekends. And then, there are times like today, when the numbers are cut in half.
How does Feedburner come up with the numbers? All of the major feed reading services report subscriber numbers. When most web-based services request the latest content from a feed that’s been burned through Feedburner, that request goes through Feedburner. The request includes a report of the total number of users who have subscribed to that feed. (The exact way each feed reader counts total subscribers varies by service.) Feedburner keeps track of the number reported by each web-based service for each version of the feed and totals them up.
Standalone aggregators do things a little differently. They don’t have overall subscriber numbers. Instead, they use a number of factors to determine how many people are requesting a given feed in a day. For instance, these services can determine if 10 people have requested a feed or one person has requested a feed 10 times. Feedburner has chosen to tally up the number of requests reported by standalone aggregators once every 24 hours (which is why Feedburner stats are updated once a day).
Why do the numbers fluctuate? Standalone aggregators are the primary reason you generally see lower numbers on weekends. Lots of people don’t turn their computers on over the weekend (although neither Rick nor I knew who these people could possibly be), so desktop clients such as Outlook on those dark computers aren’t requesting feeds.
Other reasons exist for variations. For instance, a service may not report numbers one day for some reason internal to that service. Today is great example of this. Google Reader, for whatever reason, didn’t report subscriber numbers for Thursday, which meant Feedburner couldn’t include them in the nightly tally. On first glance, it appeared that something was wrong with Feedburner, but Feedburner was reporting exactly the same as always — summing up the numbers reported from all services. If you see a big drop, you can look at break out by service (click “Subscribers” under Feed Stats”) to see if any are missing.
Finding subscriber numbers for blogs other than your own Can you find out how many subscribers blogs other than your own have? You can only see Feedburner numbers if that blog has chosen to publish them (typically using the Feedburner chicklet). Many blogs have. Otherwise, you can only see subscriber numbers for some of the individual feed readers. Recently, Google Reader started reporting subscriber numbers. iGoogle started reporting gadget use earlier this year, and Bloglines has published subscriber numbers for some time.
How accurate are the numbers? Does the subscribed number really reflect the number of people reading your blog? Probably not. Rather, that number indicates how many people have ever subscribed to your feed. On the surface, that may sound like the same thing, but it’s actually very different. For instance, if your feed is included in a default bundle, anyone who adds that bundle is counted in your subscriber numbers even if they never read your feed. And people who subscribe to a feed rarely unsubscribe from it even if they stop reading. Older blogs may tend to have overinflated counts because when users switch feed readers (for instance, from Bloglines to Google Reader), they don’t unsubscribe from the feeds in the first reader, so the original subscriptions still count (as do the new subscriptions in the new feed reader).
How can you tell how many people are actually reading your posts via subscription? The best number to go by is “reach.” Reach is the number of users who viewed or clicked on your feed. These are the people who engaged with your content on a given day. Rick likened the subscriber number to the number of people who have ever bookmarked your site and the reach number to the number who visited today.
You can get more granular information by enabling “item use”. Once you enable this, Feedburner embeds a 1×1 pixel into the feed so it can track individual posts. Some readers (such as iTunes and Tivo) don’t render HTML, so this count isn’t 100% accurate, but it provides fairly good information about how often each post was viewed. You can separately enable “clicks,” which let you know how many people clicked on the links in your posts.
Note that no personally identifiable information is captured in the view of click data, so while you can look at your server logs or use an analytics program to see information about visitors to your site such as their IP addresses, you can’t get this information from feed stats. You can only see what feed readers your subscribers used to read your posts.
If your want stats on visitors to your site, you can enable Site Stats, which provides analytics-type information about your site visitors. This is entirely different from your subscriber stats and should closely align with analytics programs such as Google Analytics.
Tips for Managing Your Feed
- Make only one version of your feed available. There’s no reason to provide a feed for every feed format.
- Ensure that if you have multiple versions of your feed, you redirect them all to the Feedburner version so you get an accurate view of your subscribers.
- Use Feedburner’s MyBrand to publish a feed on your domain so that if you want to switch from Feedburner later you don’t lose all of your subscribers.
Rick recommends that bloggers look at trends over time rather than fluctuations on individual days, since those fluctuations tend to be due to how feed readers are reporting more than reflecting actual subscription changes. And even more than that, he recommends looking at reach more than subscribers, since the reach number more accurately reflects the number of actual readers. I told him that was unlikely to happen. After all, the subscriber number is almost always bigger.