I’ve been having some conversations recently with SFMOMA on how exactly to count the number of people that are subscribed to our podcasts. The quick answer is “look at your stats” but that doesn’t always work. For example, we run stats on our website as a whole. I can see how many page views there have been to our podcast RSS feed, but as many people know, this doesn’t tell you much.
This is because RSS readers will look for new content multiple times a day. Each request is a “view” in the eyes of the stats package, but it tells us nothing about how many unique people actually are subscribed. If we have 5000 page views on our RSS feed, is that 500 people requesting the RSS feed 10 times, or 50 people requesting it 100 times? There’s no way to know.
The answer is to generate unique stats for a filtered version of your log file. Luckly, our stats package, AWStats, allows for this. Nate was able to run a unique stats report for our New Media neighborhood, but only count stats to the URL “/aoc/rss.wac” which is our podcast feed. This allows us to see all stats on this one URL, which includes unique visitors.
Unique visitors are how many unique IP’s have accessed that URL in the given month. So if my podcast client accessed that URL 30 times, it’d still only count me as one visitor. This gives us a good indication of the number of people actively subscribed to our podcast. Those numbers are as follows:
Unique visitors to podcast RSS feed
Sep – 37
Oct – 194
Nov – 303
Dec – 524
Hey, we’re gaining subscribers! But one thing we realised was that not everyone accessing the podcast feed were using a podcast client. Some came from Firefox or IE. Most likely these people just clicked on the RSS link in the browser without knowing what it was for. They make up about 10% of the unique visits.
At the same time there are sites that aggrigate RSS feeds, where one site requests the RSS feed for their many users. Those sites would be counted as one visitor, even though many are accessing that feed. I’m not sure of any that do this for podcasts (I’m sure someone will let me know if there is), but for blogs, there are many that do (like Bloglines), making counting total subscribers for blogs much harder (just because it’s more prevalent). Thus even these numbers aren’t exact, and have the potential to skew more as time goes on and aggrigation sites become more popular, but they’re about as close as we can get for now. We’ll just have to keep on top of it.
We also decided to run another separate stat ouput on MP3 downloads. This time we limited it to MP3s downloaded from podcast clients only, too see how many audio files were actually downloaded from subscribers (as opposed to people downloading them from our website, which you can do as well). Here are the stats there:
MP3 file downloads from podcast clients
Sep – 1816
Oct – 2117
Nov – 1564
Dec – 2696
Also of note is that 86% of these files were downloaded with iTunes, which means it’s certainly the number one podcasting client out there.
Hopefully that helps others who are trying to get a handle on subscriber counts. If anyone has any insight on generating even more accurate numbers, please share!