Your comparison with radio — which chases both TSL and Cume — doesn’t quite work for me.
In radio, if you charge $5 CPM (“cost per thousand” to play an ad to a person), in order to earn double you need to EITHER double the amount of listeners, OR double the amount of time they listen. So TSL is really helpful: but cume is really helpful, too. Narrow your playlist, and people will listen for less time — but you’re playing more of the hits, so MORE people will listen overall. Widen your playlist, and people will listen for longer, BUT you’re playing the hits less often, so FEWER people will listen overall. Both TSL and cume contribute to the money.
But in podcasting, a “download” is the only figure you need to charge for advertising: because everyone will only ever download a podcast once. (If we assume every download is also played, which is a wild assumption). So if you’re charging $20 CPM, in order to earn double you need to JUST double the amount of downloads.
This effect is compounded by the very real problems of measuring unique reach. You can’t technically do it, except with research like yours — the best a podcast host can do is to measure unique DEVICE reach, not unique HUMAN reach. If I listen to The Daily on my Google speaker at home one day, and then listen to it on my mobile phone while walking the dog the next day, I’m TWO “people” according to any available technology. Unless I use the same app on both, and sign in on both, and the podcast publisher somehow gets that data: but podcasting doesn’t work like that.
Sponsorship is of course a thing for podcasting; but typically, that’s not sold retroactively; whereas CPM-based advertising often can be.
Absolutely, podcasting should be measured on more than downloads. Download measurement leads to podcasts rerunning old episodes during holiday periods, rather than treating listeners with respect and intelligence and just not publishing anything. But while podcasting remains open, RAD is not a standard and the RSS feed is the primary method of distribution, the download is really the only data point we have.
