I think we can all agree that the podcast listening experience on smart speakers is pretty terrible. And by “we” I mean podcasters. Chances are, you’ve used your Amazon/Apple/Google voice assistant device to find and listen to at least a portion of one of your episodes. But you probably don’t use your smart speaker as your primary podcast-listening device. And it’s quite likely that you’ve relegated the entire smart speaker platform as one of the things you might as well do, but don’t have much use for.
I get it. You and I, we’re working podcasters. We know the best podcast experience is listening via our earbuds with our preferred listening app in our hands. We have the habit of listening via that experience, and it’s lightyears better than the voice-only, limited-choice, and rather unintelligent podcast discovery and listening experience provided by smart speakers today.
And we must not be alone in that assessment, as every data point I’ve seen shows that less than 1% of people listen to podcasts with their smart speaker. So we’re vindicated
But that number is going to go up. It doesn’t matter how we cranky podcasters feel about the current experience. And it’s only the most foolish among us who think that Amazon/Apple/Google aren’t planning a radical overhaul of the podcast listening experience on their devices.
Of course they are. All three of those companies have bet heavily on voice assistants to fill a variety of their customers’ needs for information, entertainment, and more. Sure sounds like a perfect fit for podcasting.
But we know it’s not. Let's say I’m going to use my ai-powered voice assistant to listen to a podcast I just heard about it. I don’t recall the entire name, but I know it was called “[Something] Confetti” Or maybe “Confetti [Somthing]”.
If I ask the device to play the podcast called confetti, it will. It will fulfill that objective and play for me the most recent episode of a podcast with confetti in the title. But the audio that’s playing is almost certainly not the Confetti [somethng] or [Something] Confetti podcast I was looking for.
As it turns out, there are a lot of podcasts that match on “confetti”. Without a screen to display results to me, these devices are limited in their response. Telling me “I found 36 podcasts with the keyword “confetti” in their titles. Would you like me to read out the titles of each podcast? Just say ‘That’s it!’ when you hear the name of the show you were looking for.”
How’s would anyone think that experience a good one? Boo.
But that experience can be (and will be) significantly overhauled in the future, making for not only a usable podcast listening experience on smart speakers, but possibly making voice assistance one of the best ways to find and listen to podcasts.
Feed Your Assistant Better Data; Get A Better Assistant
Think of how much information these voice assistants have access to. Not by surreptitiously recording your every utterance for use in ad targeting. How droll. And also how unnecessary.
A quick check of your Google Account Data & Personalization tab hints at what’s possible. Imagine how much better of an assistant it would be if it could parse your “play the podcast called confetti” request with other things it already knows about you to if not find the right show then at least cut out the obvious misses.
I know you don’t want anyone looking at your browser history or search history, but Google is the entity that keeps that data! If your Google-powered assistant noticed in your recent search a similar typed-in query, it could check both lists for the best matches. And if you’d previously visited a URL for a podcast with confetti in the title, that’s a pretty strong signal.
That same Google-powered voice assistant could look for similar clues in your YouTube videos - either watched or published - for context. Your physical timeline history could help give context. Or a quick scan of your inbox to see if you’d received (or sent) an email about the podcast you’re looking for.
I don’t see privacy as an issue here. Again, this is data that this tech giant already keeps about you. It’s not being shared with anyone but you!
Imagine how much better your Amazon-powered smart speaker would be at returning relevant podcast results if it was pulling from the behavior profile you’ve built up with your purchases on Amazon.com. Or the types of books you order. Or the books you abandon mid-read on your Kindle. Or the shows you watch on Amazon Prime Video.
Imagine if your Apple-powered smart speaker not only understood that you were looking for a podcast, but knew what podcasts you currently listen to, and what Apple services you use - and how you use them - on your iPhone.
Now imagine a future where your smart speaker is actually smart. That starts by using the data - data is already is connected to - to narrow the 36 possible results down to three. Much more manageable.
But we can go beyond that. Instead of reading aloud the 3 keyword-stuffed podcast titles which lack the context to jog your memory, imagine if your intelligent voice assistant came back with:
“Do you mean the confetti podcast about confetti the paper product, the one from the SciFi convention in [City], or the one about the beverage tea?”
Now that would be an intelligent assistant. And an incredible podcast discovery experience!
But now wait a minute. Think about the audio that’s going to play. Which episode will it be? The most recent episode? What makes that the best episode to play for a new listener? It’s never the right answer for serialized podcasts, so I prefer to imagine a future where these helpful devices become even more helpful:
“There are two seasons of this podcast. Season One is complete at 33 episodes. Season Two appears to be ongoing and currently consists of 17 episodes. Where would you like to begin?”
I promise I’m not high right now. What I just described is not a pipe dream. Setting aside non-trivial issues like development effort, processing time, and response time from request to results returned, Amazon/Apple/Google could do everything I mentioned tomorrow, fundamentally changing the experience of podcast listening on smart speakers.
Feed Assistants Better Data On Your Podcast, Get Better Love From Assistants
It took me all of 20 minutes to come up with those made-up better voice assistant responses when presenting podcast data. So that’s just the start. But regardless what “better” looks like, you’ll only be able to take advantage of it if your podcast data is in better shape. Yes, it’s cleanup time.
Start with your show-level data. Yes, that means the title of your existing podcast. If it’s too long, too clever, or too keyword-stuffed, it shouldn’t be. Make sure the name of your host(s) are in the author field (yes, that’s a dumb name). Make sure descriptive elements like your subtitle and your show description are well-written, actually descriptive of your show, and are clearly and uniquely yours and yours alone. You don’t want these voice assistants having guess if your show is the right show. You want their confidence very, very high.
And then begin the clean-up process of your episode-level data. Yes. All episodes. I don’t care if you have 361 episodes. Start at the most recent and work backward. Make sure you’re using episode numbers properly and consistently. Even though we don’t have good RSS tags for it, categorize each of your episodes so the assistants can see how your episodes are connected. Correct your hastily-entered “show notes” to transform them into useful episode details the assistants can use.
The way you and your podcast win in the future where smart speakers are made much smarter is by being the better-described show.
I’ve played the “optimizing” game professionally many times during my life. Webpages, videos, apps, podcasts… In all of those, it’s the better-described content that almost always outranks those that are poorly described.
So make a project out of it. Start cleaning up your past sins and prepare for the future where smart speaker listening becomes how a meaningful number of listeners discover and enjoy podcast content.
We’re now a week out from Evo's Long Winter's Nap. If you have an idea you'd like to share as a podcast Pontifications episode while I’m on break, get in touch with me as soon as possible: firstname.lastname@example.org.
Please visit BuyBeACoffee.com/EvoTerra if you would like to support the program, and please tell a friend about Podcast Pontifications. The only way the show grows is when working podcasters like yourself share it with other working podcasters. So please share.
I shall be back on Monday with yet another Podcast Pontifications.