Podcast stats continue to remain as unreliable as web stats. As I've said before (many, many times) since both web and podcatchers use the stateless http protocol, accurate stats are hard to produce. Advertisers have come to realize that web stats are unreliable, and they're going to have to realize that podcast stats are as well.
Case in point: Feedburner showing that there is a 2:1 download:podcatcher ratio. This means that twice as many podcast listeners prefer to direct download shows rather than use any type of podcatcher. Libsyn (arguably the largest podcast host out there) just released some stats from the first quarter of 2006 that show that web downloads only account for 22% of downloads, or about a 1:3 download:podcatcher ratio. Feedburner handles 44,000 podcast feeds (many of which are Libsyn feeds) and Libsyn claims 3,800 users (presumably that means 3,800 feeds).
Further, Feedburner stats show only 57% of listeners use iTunes where Libsyn's stats show that 72% of listeners use it.
I know in a corporate world decisions are made on hard numbers and facts. However, there is no such thing in podcasting. This thing will settle out, but for now when you're looking at a ROI based on subscriber numbers of a show or shows, be aware that even though the person who gave you those stats can likely back them up, that doesn't mean they're even approaching accurate.
