Chrysalis Radio have, rather bravely, conducted some research (source: media guardian, free registration required) into people’s usage of podcasts. The headline finding is that 30% of people who download podcasts never listen to them, and another ~30% don’t listen to the whole podcast.
One of the disadvantages of server generated statistics, such as volume of downloads, is that they tend to acquire a validity beyond reality and that are difficult to challenge. It does also demonstrate that it’s worth using traditional research techniques to calibrate server-derived statistics.
These numbers shouldn’t come as a huge surprise given the way podcasts are distributed. Once someone has, maybe on impulse, added a podcast feed to their podcast catcher (iTunes), they don’t have to think about it any more. There’s no penalty at all on them to just have it use up bandwidth and use up a bit of space on their device (iPod). So the server says they’re downloading it, but it’s not getting consumed. And so the listening figures for podcasts appear to climb every higher, disregarding any measurement of churn.
Can you imagine what would happen if Vodafone or Sky continued to report additions to their subscribers without accounting for people unsubscribing? The numbers would eventually become farsical and the game would be up. We’re not yet in the realms of obviously distorted podcast numbers, but the process of measurement is still flawed and therefore the numbers are wrong, just too small to be noticeably wrong yet.
Interestingly, two things might expose the reality of podcast consumption:
- Digital Rights Management – rather than using the authentification phase to control distribution, it could be used to measure actual consumption – when the play key is depressed. It could also be used to measure when the consumption takes place, which would give some accurate information on how many hours content previously broadcasted is timeshifted by (5 hours, 15 hours, 5 days, 5 weeks?).
- Charging for podcasts – if people have to pay for podcasts (either by payment to the content provider, or by being aware of a bandwidth cost for transfer), then they’ll be more likely to unsubscribe to content they’re no longer interested in, thus exposing the churn rate.
Podcasts are an important part of today’s radio programmer’s toolkit, but they’ll be more valuable if they’re part of the established audience measurement techniques. And there lies another challenge, and another post for another day…