Google Reader subscriber stats #1: a roundup

Google Reader recently began to show the number of subscribers for each feed in the result list when you search for a new feed. (Ionut Alex Chitu on Friday, Oct. 12) On Sunday, TechCrunch published a preliminary list from Gabe Rivera, weighted towards tech and general news sites. (A reasonable starting point considering the source.) Robert Scoble posted numbers from the new TechMeme Leaderboard plus 55 of his favorite feeds. Together, the posts hit the top of TechMeme late Sunday night and gathered reactions all day Monday. Here's a roundup.

On the Official Google Reader Blog, Mihai Parparita provided some details:

Google subscriber counts: These numbers include subscribers across all Google services, including Reader, iGoogle, and Orkut.

FeedBurner numbers: If you use FeedBurner to manage and track your feed, you will see a subscriber count there that is attributed to "Google Feedfetcher." This number is a sum of all the feeds that you have redirecting to your FeedBurner feed URL.

Alas, he also delivered the bad news that the weekend data collection wasn't accurate:

Reader's feed search was recently showing stale and incomplete data, but as of today (October 15) the numbers should be the same everywhere.

The update resolved a discrepancy between Google Reader and Feedfetcher that Tim Bray observed on Sunday.

To find some data for yourself, click on "Add subscription" and type a keyword, e.g. a blog name or the domain name without the ".com". (Entering the full domain name will subscribe you without showing the feed count.) To search using the domain name, click "Browse" and then use the "Search and browse" option at the bottom. Pro: fewer false-positive matches. Con: likely to miss some actual feeds, e.g. hosted at Feedburner or other service.

As shown in the (edited) screenshot above, a blog or site may have several feeds. It's tricky to get an accurate count: add numbers for actual feeds, skip duplicates, skip keyword matches on other sites, and make a judgement call for comments and other feeds tied to specific tags or sections of a site. (Someone may subscribe to a tag rather than the whole site, but the same person may subscribe to multiple tags.) I suspect that most people who subscribe to comment feeds are already subscribed to a blog's main feed -- though perhaps that's less true for blogs that have a separate comment feed for each post.

At first, the several URLs for a single feed looks like a bug -- e.g. a URL with or without the trailing slash ("/"). However, Andy Beard points out that it's a great hidden feature to see how someone subscribed to your feed:

1. Using A Subscription Button

2. Autodiscovery

3. Javascript Bookmark

Alas, some of the duplication is in fact a bug, e.g. the exact same URL but different title or description. See Nusuni Dot Com for a good example.

Another bug (per Technosailor): Google continues to fetch the old feed URL despite a 301 "permanent redirect" code in the HTTP header. (A 302 "temporary redirect" is one way to have the feed URL on your domain but have the actual feed served by FeedBurner or another service. TBD: research how this affects the reported stats.)

Try gathering data for several sites: it takes time. Now that Google is publishing the data, they should publish their own "top feeds" list. Louis Gray pointed out that he requested this feature over 7 months ago. (Might be interesting to see whether Google Reader has made any progress on his other 9 suggestions.) I'm not holding my breath, e.g. FeedBurner never released such a list based on their data.

With Google Reader's use of AJAX, the data isn't easy to grab with a script either. Perhaps the solution will come in the form of an API, e.g. FeedBurner provides one. (Though that only works for blogs who use FeedBurner and haven't opted out of making their data public.)

Meanwhile, please stay tuned for further analysis of the data that's available today.