I’m revisiting the issue of podcast numbers out there, because this is an important factor in our service and its implications.

Feedburner reports that it currently serves 230,000 audio and video podcasts. That’s a sizeable number already, but what percentage of total podcasts do they host?

A couple weeks ago, I wrote that DailySplice’s directory has about 1000 podcasts (a fairly small directory so far, but growing steadily.) Feedburner hosts about 33% of those, indicating that there could be over 650,000 podcasts out there.

We’ve since acquired a list of 33,000 podcasts from various hosts. By hosts, I mean the server that provides the RSS file, not the media files themselves (although that’s an interesting issue as well.) The largest RSS hosts are:

• Blogtalkradio, which hosts about 16%;
• Feedburner, which hosts about 10%;
• Libsyn, which hosts about 7%

These numbers are quite a departure to what I’ve seen in the DailySplice directory; I didn’t expect Feedburner’s contribution to be as low as 10%. This list contains only 1% of Feedburner’s total hosted feeds. The numbers for blogtalkradio seem to be in the same ballpark: only about 6% of blogtalkradio’s  87,000 shows are in this list. If the percentage of listed podcasts are the same for other hosts, this could indicate that there are between 550,000 and 3.3 million podcast feeds out there. The latter number seems outlandish, so I intend to do some more verification when more data is available. I’ll also take a look at how many of these are actively producing new content, which will be a more convincing measurement of podcast growth.

One of my burning questions is how many podcasts are out there, and how is that number changing on a month-to-month basis? Over the next month or so, I want to put a system in place to maintain an up-to-date estimate of this number.

We are in the process of building a set of services to give newcomers easier access to the massive pool of podcast content (keep reading for actual numbers). To make a scalable service, we need a reasonable estimate of the size of the data set. Running an algorithm for 100 podcasts is one thing, but running that algorithm for 100,000 or more takes some careful technical planning.

I surveyed some of the larger podcast hosts and repositories, and as of May 8, 2008:

• feedburner hosts 216,512 podcasts feeds.
• podcast.com claims to have over 60,000 podcasts indexed.
• podcastpickle.com has over 9000 podcasts (319662 episodes)
• dailysplice.com has over 1100 podcasts (70000 episodes)

Unfortunately, I couldn’t find the numbers for libsyn, one of the most prominent podcast hosts on the net. About 30% of DailySplice’s indexed feeds come from feedburner, indicating that the number of podcast feeds out there is approaching 700,000.

The other question I have is what proportion of those are regularly updated. Of DailySplice’s feeds, over 1000 ( 90%) have released an episode within the last 3 months. However, we have a relatively new directory, so many of the podcasts in there are popular and/or recently created. I’ll revisit this question once we’ve indexed more legacy content.

If anyone has more accurate estimates, I’d love to hear them! Stay tuned for updates to this topic.