Why is podcasting so broken?

…and how we’re trying to fix it

By Anne Wootton, CEO, Pop Up Archive (that’s me on the left)

I love thinking about the future of audio. I feel lucky because we get to think about it all the time at Pop Up Archive.

I started Pop Up Archive with Bailey Smith in 2012 to help journalists and producers do a better job wrangling digital audio (wrangling in this context = digitally preserving and accessing through timestamped transcripts, tags, tools for indexing and sharing, and backup at locations like archive.org).

Any podcast fan will tell you that listening to audio on demand leaves a lot to be desired. At Pop Up Archive and as part of our latest project, Audiosear.ch, we’ve been asking people lots of questions about how they listen to audio, what data they wish they had about audio but don’t, and where they find new stuff to listen to (TL;DR answer to that: word of mouth, other podcasts, and Twitter). For example:

“Right now, I use @Pocket to bookmark audio and tag it as “to listen” periodically add those to my @OvercastFM queue. But that is such a broken experience.”
“I often follow links out of Twitter… find things on Twitter, then whatever platform the creator chooses.”
“If someone sends me a file, I’ll sometimes put it in a “Listening” folder on my public folder of Dropbox and listen there. I usually listen on the go on my iPhone — driving, doing dishes, exercising. Rarely sitting at the computer. I WISH there was a great interface to listen to all saved audio — a random piece from Third Coast or Transom.org or Facebook that I could put in an app like a podcast app and listen later… kind of like a DVR for audio or Instapaper for audio.”

Emily White, an avid podcast listener who happened to provide our first example above, summarized these pain points thoroughly on Medium back in March.

So what are people doing to improve podcast tech? By way of introduction, let’s paint a picture of what a truly audio-enabled web and mobile world looks like. In a world where audio listening and discovery works:

Audio is reliable and has better controls

  • You can talk to your podcast app, telling it things like “next” to skip a podcast, “save the last five minutes” to save audio snippets for yourself, or “share” to send your favorites to friends.
  • Streaming works all the time and you never have to worry about waiting for things to download or freeing up space.

Audio is social

  • Your podcast app tells you which podcasts/episodes are most popular.
  • You can play audio that your favorite people, or the people you respect most, are listening to or talking about.
  • You can easily create and share your own playlists.
  • You can share your favorite moments from podcasts in places the people you want to hear them will find them.

Audio is personalized

  • Your podcast app knows what you like and starts playing the moment you put earbuds to your ears.
  • It randomly suggests new content (“play me something new”) and plays audio similar to the stories you like.
  • It lets you follow different topics, people, or alerts, so you know any time they’re talking or being talked about.
  • It knows what you like to listen to at different times of day — say, news in the morning, stories after work, and kid-appropriate content when you’re driving with your family. It lets you filter by these formats, too.

Audio is searchable and skimmable

  • If you’re looking for something and only remember a few words or phrase from an episode, you can type them in and find it instantaneously.
  • You can get 200–300 word summaries of long-form podcasts.
  • You can skim through the transcript of a podcast if you’re not inclined or able to listen to the whole thing.
  • Your podcast app aggregates segments — not just whole shows — from public radio shows and podcasts like Marketplace or Slate Political Gabfest, so you can discover the best parts in curated chunks, jump around, and only take, save, or share what you want.

Audio is supported by completely new, tech-enabled business models

  • Creators of podcasts — whether individual shows or networks like Radiotopia, Slate, Earwolf, CBS, or The Loud Speakers — get detailed audience demographics and metrics for audio.
  • More importantly, they can actually make use of those metrics to control your unique experience of their app. They know what you like, and they know whether or not you’ve ever listened to a certain show before.
  • Rather than just playing ads, your favorite podcast app gives you actionable prompts: links you can click to get promo codes, register for live events, or donate to your favorite shows, spurred in part by a tally they provide of how many hours you’ve listened to.

Brian Boyer (NPR Visuals Team) might put it best: “I want to plop down my phone and hit play and hear good shit forever.”

We’ve heard the complaints about Apple’s inadvertent stranglehold on the podcasting industry and relatively limited native iOS app. It’s limited from both a user experience perspective and a content creator/publisher perspective. I won’t list the reasons. You can read about them here, here, and here.

Don’t get me wrong — I use the Apple podcasting app. My co-founder Bailey uses it too, and she’s the earliest of early adopters when it comes to giving new technology a chance. But there’s a void here begging to be filled. It’s a huge opportunity.

Stitcher and SoundCloud aren’t quite filling it: Stitcher is full of advertising on top of the ads already in the podcasts, and even though SoundCloud opened podcasting features to everyone in April, the platform is still music-centric: random EDM tracks crop up between clips of Fresh Air and Planet Money. Marco Arment’s Overcast and Downcast apps capture a sliver of the iOS-enabled population, and there are many others filling in the gaps. The space is crowded, but still woefully underserved.

None of these apps offer particularly social or personalized experiences — most of them have small user bases and don’t integrate with other platforms beyond some basic Twitter OAuth. None of them have particularly granular metrics unless they’re a walled garden (SoundCloud, Stitcher, PRX) serving up their own audio files. Former Stitcher CEO Noah Shanok has pointed out that digital media ad buyers aren’t even asking for metrics like “listens.” They’re focused on downloads and demographics (podcast ad sales company Midroll has surveyed over 1,000,000 listeners to date), and more granular metrics don’t really make a difference given the way podcast ad units are sold today. Better podcast tech coupled with smarter timestamped metadata could drive a lot more revenue to podcasts.

Source: http://www.cjr.org/first_person/the_economics_of_the_podcast_boom.php.

Buzzfeed launched their first podcasts in March, and PRX is baking listener support into the core of Radiotopia’s business model. Spotify rolled out podcasts in May. Gimlet Media is committed to content for the time being, but talked a lot in the first season of Startup about tech plays and the “Instagram for audio.” The pre-launch Spoken.am, currently a weekly podcast recommendation newsletter, is led by two former members of the SoundCloud product team who promise more on the way.

Most are focusing on content creation as a strategy for the time being, and that makes sense. But wow, there is a lot of functionality lacking in this ecosystem right now.

We’ve been having way too much fun with Audiosear.ch, a podcast intelligence API and search engine that we hope will help solve some of these technology problems. We’re building Audiosear.ch to be the most comprehensive source of data about podcasts and radio. Thanks to the same pipes that have powered Pop Up Archive since we started building with PRX almost three years ago, we can transform sound into information through time-stamped transcripts and descriptive tags. Since we’re a bunch of geeks, we’re having a gleeful time building taxonomies for spoken word audio — processing all the audio that’s available in one place or another and making it searchable from our API.

We’re also taking it a step further, by collecting and centralizing as much information about podcasts out on the web as we can find and structure. We’re synthesizing thousands of podcast data points to generate some magical intelligence for spoken word audio. What does that mean? Recommendations that can be broken down by show, network, category, interest, topic, and more. Trending podcasts based on what the web is talking about…

or what people whose opinions you care about are listening to:

The current Audiosear.ch database is small, but growing. It currently includes thousands of episodes from over 200 top podcasts and radio shows on iTunes and SoundCloud (and we’re taking requests).

It’s not clear if the future of spoken word audio will be dominated by platforms with comprehensive offerings, à la most streaming music services, or if spoken word distribution will be a balance of broad platforms and more narrowly-focused ones like Earwolf, Slate, Vsporto or NPROne, all of which have built up large archives of particular content that audiences are hungry for (did you catch Terry Gross repping the Fresh Air archive when she talked with Marc Maron recently?).

At the end of the day, we want to make it easier to find, discover, and share audio stories we love. Is that something you’d like too? We’d love to hear your feedback, suggestions, critiques, and hey, recommend us a podcast while you’re at it. The tech problems facing podcasting can’t be solved in a vacuum, and we’re inspired by the conversations we increasingly find ourselves having with like-minded audio geeks. That pipe-dream podcast experience gets closer every day.

There’s an API for that

Are you an audio app developer? Podcast listener? Pop Up Archive just released v1 of the Audiosear.ch API, and it’s yours for the taking. Our goal is for it to help people explore, share, and get recommendations for great audio stories. We’re adding to the API all the time, influenced mostly by things we hear from people like you. Hop over to the Audiosear.ch API developer docs and try it out for yourself. Give us a shout to tell us what’s missing. SDKs abound in Ruby, Python, and PHP.