More podcast downloads than actual listeners - how can we know for sure?

Published by neil on June 5th, 2006

On the surface it would initially appear that accurate recording of listeners as opposed to subscribers or downloads is essential in developing awareness of a podcast’s actual audience. But despite some companies suggesting they can achieve just that, an all encompassing, accurate solution is far from reality.

This post was triggered by Neville Hobson and later this Britcaster discussion.

We have the initial problem of tracking listeners using mp3 players - pretty much impossible. There are a couple of ideas here, however. First a wireless mp3 player which networks to a central server every time it plays a track. Ignoring the overwhelming lack of permanent wireless connectivity for mobile devices (excluding phones), this actually would be an attractive option for users who like to keep those cluttering ‘recently listened to…’ panels on their blogs. But for the purposes of listener monitoring, not workable.

Secondly, we could have, on every portable media player, an immediately updated record of audio tracks played on the device, which is then synchronised with a central server as the device’s audio is synchronised or updated.

Both these require acceptance and adoption by a diverse and directly competing range of hardware product manufacturers. Though with the ipod’s significant dominance in the space, just that device’s adoption would be a viable statistical resource.

With consistent statistics suggesting as few as 20 percent of podcast listeners using media players, the need to measure offline listeners becomes a minority consideration.

It is possible to track the opening and access of a PDF, but that is because it can contain external asset URLs. Now if the AAC enhanced format used by iTunes offered the inclusion of chapter images from remote URLs (instead of the embedded images it currently uses), or perhaps offering both options simultaneously so that offline listeners do not lose out on the imagery, then tracking would be a doddle. Just tag an image with a unique identifier and the listener record is captured at the http request. In my personal experience the takeup of ACC podcasts versus mp3 is low - a recent music podcast I began developing with parallel mp3 and enhanced AAC feeds showed just above 10% AAC takeup compared to the mp3.
[Anyone else have their own figures on this?]

With the majority tuning in to podcasts on their computers, there are existing options availble.

Podbridge, as Neville mentions, requires the podcast to be accessed using their own proprietary software - not exactly difficult to track listeners with that model. My question would be a concern over how listeners might have to be persuaded to install yet another bit of software just to listen to a limited number of poscasts - a strength of podcasting is the (almost) one click operation for all your subscriptions, and the comments on Neville’s blog post suggests just that concern. Admittedly, if the content was truly compelling, that may be a no-brainer for taking the additional installation step, but how many podcasts are really that important to an individual? And if they really are that important, and popular, then they could likely successfully switch to a paid subscripton model (because monitoring actual listeners has to be financially motivated anyway - to encourage advertising, I know there are others, but mostly advertising - otherwise it is merely geek interest), in which case subscribers, not listeners, become the primary focus. If, to encourage more widespread takeup, the podcast distributes via something like Podbridge in addition to conventional mp3, then the Podbridge stats become watered down and irrelevant.

I believe the listener tracking issue might be more straightforward to apply without the need for proprietary software.

We still have to discount offline listeners (personal media players) and those who will download and listen from within their iTunes or other podcatcher. But it would be relatively straightforward to force the online listener to use a Flash based player (usually streaming) and have that player register and store a hit to both the beginning and end of an audio file, thus giving you a pretty accurate record of whether someone listened. I guess this could be taken further by recording the listening activity at various key points throughout the podcast which might enable tracking of advert or chapter listenership. I’m not 100% sure if Flash could handle event triggering midway through an audio stream, but it could certainly register if the user decided to use a control button to fast forward or rewind.

Such a solution would be more open - because the vast majority of web browsers can handle Flash and would not require registration - and therefore provide potentially more accurate stats than even Podbridge (though with less detail as Podbridge likely requires registration by the listener - I haven’t checked this bit out as I’m writing this on the move.

In the end we may have to simply do some broad research and gain a general understanding on average as to how many podcasts are listened to against how many are actually downloaded. I suspect that the majority of even regular podcast subscribers end up downloading far more than they get around to listening to. Not the best of news for the increasingly common stats-obsessed new podcaster struggling to get over 100 downloads per episode, but the reality is potentially much less rosy than simple numbers might suggest.

Technorati Tags: podcasting, listeners, stats

Tags: Audacious-Communications, Business
Posted in Business |