Tracking Podcast Downloads Is Worthless To Everyone [S3E41] - a podcast by Evo Terra

I’m aware that most podcast advertising inventory today is priced and purchased based on the number of downloads an episode is projected to receive. And it is with that full awareness that I assure you no advertiser wants to buy podcast advertising inventory that way. Instead, advertisers want to pay when their ads are listened to by the right audience in the right geography at the right time.

Downloads, by themselves, illuminate none of that. Downloads are nothing more than an internal count of computers talking to computers. 

Spotify is changing the game, working with select advertisers to optimize campaign performance around true listening activity. And they’re not alone. I encourage you to read the great dialog between Amplifi Media and Sounds Profitable, where Steven Goldstein and Bryan Barletta walk through the sweeping changes we’re already seeing in podcast advertising. 

Then there’s the fact that downloads of podcast media files can be faked in a way that podcast media hosting companies following IAB 2.0 guidelines cannot detect. Anthony Gourraud details exactly how he faked downloads that were tracked and measured by IAB compliant hosts and tracking services. 

As advertisers gain a better picture of the disparity between actual listener behavior provided by Spotify and file download activity provided by their media host, the inferiority of downloads will become more obvious, eroding the trust placed in that metric. Evnatually, high-trust data (measuring listens) will be sold at a premium. Low-trust data (measuring downloads) will be sold on the cheap, forcing publishers to apply pressure directly on other apps to provide similar data to reclaim the value of their mutli-platform ad inventory. 

Tom Webster of Edison Research published an article that illustrates what  I'm calling “The Serial Problem”. Popularity rankers aren’t smart enough (though they could be) to separate automatic downloads from promotional content by legacy subscribers from actual consumer behavior of the actual podcast episodes. 

The obvious improvement here is to track the actual popularity among consumers against the totality of episodes of the show itself. Which we can do today, getting rid of these out-of-date rankings that are deeply, deeply flawed.

Want to double the number of downloads your podcast gets in a month? Produce twice as many episodes. Even though you’ve not gained a single new listener, you’ll see ~100% growth.

A much better way to measure show growth is by measuring the unique audience of a podcast over a period of time, noting how often each unique person accesses one or more episodes of a podcast. 

Once you have a good handle on active users over various time frames, you’re able to examine and segment on activity. We might find interesting, data-driven learnings on Time To Listen by calculating the amount of time that elapsed from the file request (taking into...

Further episodes of Podcast Pontifications

Further podcasts by Evo Terra

Website of Evo Terra