Over the weekend I was listening to Desert Oracle and learned of an Arizona-based band called Giant Sand. Ken and his guest had high praise for the band and used some descriptive language that piqued my interest. So I opened up my very popular music streaming app, searched for the band by name, and started listening.
And I was hooked immediately. Their sound is right in my wheelhouse. The perfect kind of weirdo stuff that I and weirdos like me like. While I was mentally making a list of all the people I wanted to yell out for not introducing me to this band when I first arrived in Arizona back in 1997, I was struck by the experience I just had. Specifically, I started contemplating the difference in how music is presented to new listeners in 2020 versus how podcast episodes are presented to the uninitiated.
Here’s what the music streaming service didn’t do: They didn’t present me with the most recent track released by the band. A band that’s been around for some 30 years and has an evolving sound. The service also didn’t present me with a list of the 30-ish albums the band has released over the decades, forcing me to pick one to listen to in its entirety. Because this isn’t the ‘70s and I don’t even own a record player.
Instead, this app mined its own data built from roughly 260 million users and presented to me a list of songs by that band ordered by popularity. Their assumption is solid: there’s a very good chance that the more popular a track is, the more likely that a new listener will not only like it but then keep listening to more content.
No, That’s Not How It Works In Podcasting Today
On the surface, it seems like podcast listening apps do that for us as well. If you hit the home screen, they’ll show you the most popular podcasts, and that’s probably based on internal usage data.
But then... what is the listener to do next? All the app has done is line up the shows themselves by popularity, which is akin to showing the most popular albums. Sure, die-hard fans often have a favorite album. But normal people just have favorite songs. And most of us would struggle to name the album that featured their favorite song.
To which episode should a person brand new to that podcast listen? And which episode after that, ensuring that the listener keeps listening? How should episodes be presented to help new listeners better enjoy content from that podcast on that app?
Put yourself in the shoes of a brand new listener for just a moment and imagine a podcast app working more like the music streaming apps. Someone tells you of a new show and you search for that show by name on your listening app. This imaginary helpful app then presents to you the very best episodes of that show for you to listen to first, allowing you to quickly confirm the recommendation given was right for you.
That’s a very different world than we live in today, podcasters.
One Size Does Not Fit All
No, this is not going to work for all types of podcasts. And yes, I can already hear the pearl-clutching of many of my conventional podcasting friends as they envision a dystopian world where an app has the audacity to present their content differently. The horror!
This concept won’t work for time-sensitive content, where you’re covering “the news” or current events. It’s possible that an episode of The Daily from 2018 or Geek News Central from 2007 is quite popular, but those episodes have a shelf life that expires in single-digit days. Regardless of popularity, those episodes are stale and shouldn’t be presented as a first listening experience.
It also completely breaks for serial podcasts like, well... Serial or with podcast fiction/audio drama content like Valence. Serial podcasts, by definition, need to be consumed in order from the first episode. By way of example: Even if Amazon could somehow determine Chapter Six of a Kindle novel was most enjoyed by readers, they’re still going to start new readers at Chapter One.
Those notable categories aside: This concept is perfect for timeless, evergreen podcast content. Content that is neither immediately newsworthy nor builds toward a conclusion in a linear fashion.
A Better World For Evergreen Podcast Episodes
A huge portion of the podcast landscape is evergreen, with episodes that maintain their relevance months and years after they’ve been published. Shows like Sleep With Me or The Snooze Button are good examples. The former will clearly have “favorite” episodes that help you fall (and stay) asleep by listening. The latter almost certainly has stand-out episodes with great information from guests on how to get a great night’s sleep.
For those shows, and the vast swath of the podosphere that is also timeless content, the forced adherence to linear episode consumption in apps runs counter to what music app developers have figured out.
As I've said previously on this show, the most recent episode of your podcast isn't necessarily the best episode of your podcast. Yet it’s the one that the apps present to new listeners and would-be subscribers first. See the disconnect?
Podcast Downloads And Ketchup Packets
What data can be relied upon to determine the best episodes of a timeless podcast? Popularity is what the music streaming apps use, but getting to that is a problem in podcasting.
Downloads aren’t helpful in making this determination. Sure, you can run an easy report that clearly shows which episode gets the most downloads. But that’s not a good data set, and I need to use the metaphor of you running a fast food restaurant to explain why.
Let's say you, in your new role of fast food restaurant manager, want to understand which of your three condiments -- ketchup, mayonnaise, and mustard -- is most popular. If you count up the number of packets of each leaving your store every day, you’d probably learn that ketchup is by far your most popular, so you better order more ketchup, right?
But there’s a lot of noise in that signal, and it comes from drive-thru and take-out orders. To maintain the efficiency of your quick-serve (the term they prefer) restaurant, your staff has been trained to automatically, and with almost every order, toss 2-3 packets of ketchup in the bag. Conversely, it’s exceedingly rare for mayo or mustard packets to be added without the express wishes of the person who ordered the meal.
Because of that noise in the signal, you can’t rely on the total volume of packers moved to base your decision upon. It’s really only the volume of packets moved through the on-the-counter condiment receptacles that you can count on. But because of efficiency, you use the same packets in both places, so your end of the month analysis of condiments is a fool’s errand. So you just order more ketchup?
Many (most?) podcast episode downloads are like ketchup packets tossed in a takeaway bag. They often (usually?) happen automatically, which is why you see a spike in downloads the day your release an episode. Having episodes automatically appear in podcast listening apps or other distribution channels is one of the magic powers of podcasting.
But that magic adds noise to our data. Because a download does not equate to a listen. And podcast hosts remove additional downloads/streams when they detect them to preserve the integrity of “unique downloads”.
Play That Podcast Episode Again, Sam
Podcast hosting companies could give us this data if they wanted to. (And wanted to invest the development resources.) They could do deeper analyses and identify when something out of the ordinary happens on one of your episodes -- a possible popularity signal. They could also better report on repeat listening, connections between episodes downloaded by the same unique user, and other insightful data that could help us build truly timeless episodes our listeners want to listen to over and over again.
Podcast apps should have a better time at this. Forgetting for a moment the fractured reality of podcast listening apps, they at least have actual-play data. And because podcast apps require an account to keep track of subscriptions and listening history, the apps have great insight into actual listening behavior that could be shared with the creators of the content. At the very least, the apps could use repeat-plays as a strong signal to help present the best timeless episode to a brand new listener
Blame It On IAB 2.0
Actually, blame it on us. We’re the ones who decided we wanted accurate download counts more than anything else. We didn’t see any value in knowing how many times a single user listened to the same episode. We didn’t understand how that behavior at-scale might give us incredible insight. We couldn’t see why we might want to use that data to present our episodes in a better way for people who’ve just discovered our content.
Maybe, as we approach the third decade of podcasting, our attitudes will change. Maybe this concept will percolate through the listening app community. I’d actually bet money on that, as the much-more-data-savvy app makers now listing podcasts in their apps will force that change.
They’ve solved the problem for music. They’ll solve it for podcasting too.
If you enjoy my pontificating, please go to BuyMeACoffee.com/EvoTerra. You can not only buy me a virtual coffee (please and thank you), but you can also become a member. Membership, as you’ve no doubt heard, has its privileges. I just announced some special perks for members over the weekend and have a few more ideas as well.
If that's not in the cards for you (thanks, impending global economic collapse), then pretty please, with sugar on it, tell one other podcaster you know about Podcast Pontifications today. Even if you've done so previously, please mention the show to somebody new you’ve just discovered is podcasting.
I shall be back tomorrow with yet another Podcast Pontifications.
Cheers!
