Measuring Downloads in the Age of Streaming

This week the industry has been revisiting that perennial topic: measuring your audience. Inside Radio has a report from a NAB panel:

“The good news for podcasters and buyers is measurement challenges are 97% solved,” Midroll Media CRO Lex Friedman said on a podcast panel at the National Association of Broadcasters (NAB) Show last week. “What we can report now is more specific than we could before.”

Nick Quah, in his excellent HotPod newsletter, followed up with Mr Friedman about it:

Today in podcasting, the measurement problem is solved; the remaining 3% is getting everyone standardized. It doesn’t happen often, but every once in a while Midroll loses a show to a competitor. When we sell a show at 450,000 downloads, and the next day the same show and same feed is being sold at 700,000 downloads, that’s a problem.

If you’re thinking in terms of downloads, this is good news. If you’re thinking in terms of downloads. Keep that in mind as I look at another report from NAB, this time from RAINNews:

Spotify is now a firm #2 in terms of podcast downloads on the Libsyn platform which includes some 44,000 podcasts (though not all on Spotify).

Ah, here is the rub: Google Play Music, Pandora, Spotify and all other streaming services don’t have the concept of a download. A user either started playing an episode, listened to some of an episode, or listened to all of an episode. Even if the download measurement is “97% fixed”, it sure feels like the industry is moving to streams. Which raises the question: how do you reconcile a download with a stream?

This reminds me of the state of album charts in mid 2010’s: the entire music industry was counting the number of album sales (and of course you never knew if the purchaser ever listened to it!), then streaming came along and the industry tried to figure out how to reconcile streams with purchases. Billboard came up with Album-Equivalent Units where 1500 streams would equal one album sale. The IAB Podcast Measurement Guidelines are currently the closest thing we have to an agreement on what a download is, but they are vague on how to measure a ‘listen’ on a streaming service. So let’s say a user listens to 60 seconds or more of a podcast to count as a ‘stream’.

How many podcast episode downloads equal a stream (i.e. an actual listen)? I’ve heard estimates in the region of 20-40% of downloads are actually listened to. So five downloads equals 2 streams?

All of this talk of downloads and streams is just a proxy for being able to guess how many people heard an advertisement. What ad agencies really want is to be able to tell their clients how many times an ad was heard, not how many times an episode was downloaded. The stream is a far more accurate metric but it is certainly going to be lower than the download number - which, I get it, is a blow to the ego and potentially a problem when trying to sell to advertisers who have been used to a certain download number.

The one thing to remember though, is that no matter how you measure your audience, it doesn’t actually change reality. The same number of people are listening and enjoying your episodes whether you count them once or five times.