Nieman Foundation at Harvard
Are you willing to pay for Prepare to be asked before year’s end
ABOUT                    SUBSCRIBE
Sept. 15, 2016, 12:15 p.m.
Business Models

A growing layer of technology will help determine where podcasting goes next

Shows are moving well beyond a simple MP3 file and an RSS feed. But will new data, targeting, discoverability, and social tools push podcasting in the direction of commercial radio? Part 4 of a five-part series on the business of on-demand audio.

Get ready for age of “listening telemetrics.” That’s Sean Carr’s shorthand for the revolution beginning to sweep through the podcasting world.

Carr is CEO of San Francisco’s Art19, which just signed up The New York Times as a customer, after an announcement just a month ago that it had signed podcast networks Wondery, Midroll, and DGital Media.

Art19’s technology can do lots of things, but Carr sums up its value in a single sentence. “We can protect the CPMs [cost-per-thousand ad] rates that podcasters are getting.” As we’ve noted, podcasts with big audiences drive high ad rates, often 2× to 4× greater than most digital advertising.

So consider the work of Art19, Panoply’s Megaphone, PRX’s Dovetail, AdsWizz, Acast, RawVoice/Blubrry, and Libsyn, alongside those of longtime radio tech player Triton, as an effort to hold on to those great rates — and multiply revenue as ad inventory grows and grows.

It’s just the tippy top of the podcast world — a few thousand of the more than 300,000 podcasts found in U.S. iTunes store — that now embraces analytics, customer databases, and the potential of personalization.

Those in the industry estimate that no more than 10 percent of podcasts may ever make money. Building on the loosey-goosey foundation of a largely non-standard, artisanal business, we’ve seen this year a quest towards standards. Two Interactive Advertising Bureau groups, with more than three dozen participants between them, have just birthed one document with basic definitions for the industry. Ahead: Standards on measuring actual podcast listening. Standards on measuring ad listening. Standards on concepts as fundamental as what’s a “download” and what isn’t.

This is all part of an industry maturation that will bring it closer to the information precision other areas of digital media are perfecting. For podcast publishers, technology will be a force multiplier.

As with all tech decisions, the first important choice is: To build or to buy? For most, the answer is a mix. A few, like longtime public radio tech platform PRX and Panoply, are building toward being 360º operations, aiming for a complete platform offering — and then to sell some of that tech for use by others. Others do some building and some buying.

Gimlet Media president and cofounder Matt Lieber takes the most contrarian approach at this point. Gimlet — home to the breakthrough StartUp, and now producing six programs — is all-in on content.

We’ve been able to get to a scale where we’re distributing about 5 million downloads per month of immersive audio shows, shows that people spend half an hour with. We’ve been able to do all that without having a single technology engineer or product person on our staff.

That means that our business is pretty capital-efficient. We’re now 55 people, and they’re either business people, sales and operations, or they’re editorial people or sound engineers.

The flipside would be if you don’t own your tech, you don’t have direct distribution, so you’re exposed. I think that’s something we certainly like to think about and talk about a lot.

Gimlet, to be clear, uses state-of-the-art tech, supplied by Panoply, the Graham Holdings sister company of Slate. (Graham Holdings was the lead investor in Gimlet’s last venture fundraising.)

Panoply’s Megaphone hosts audio, reports data, and dynamically inserts fresh ads, and is itself built upon the work of Audiometric, an Australian company that Panoply bought a year ago.

Megaphone is a platform, and that word fits the tech upgrading of the podcast business. Just as digital publishers from Vox Media to Gannett have aimed to (and sometimes struggled with) the building of singular publishing tech, podcasting’s leaders are now doing the same. As their businesses become more complex, they too will look to systems that do more, that are built for the lengthening list of todos from publishing to audience measurement to ad serving to social promotion.

Increasingly, those platforms are usually being built new from scratch, an idea that The Washington Post is using in the wider publishing world.

I recently talked to one of Art19’s new clients, who wished to speak confidentiality on its choice of the company:

We saw two main differences with Art19. For one thing, this is all that they do. So their CMS isn’t an add-on that comes with other services like ad sales or program development — it’s their entire focus. As a result, and as a result of them being in the game longer, their service is more developed and with a more robust array of APIs. In our case, those APIs are crucial for integration with our platform, which is why we ultimately decided on working with them.

These technological demands are far removed from podcasting roots. “In podcasting, we always lived in a world of platforms, because we’ve always been distributed via this very old-fashioned but in some ways great infrastructure called RSS,” explains Matt Lieber.

“We make shows of MP3s or audio files, and our listeners can subscribe to us using any podcast client, which is essentially an RSS reader. We have our files, and listeners can pull them down from wherever they want. The beauty of that is it’s an open system that’s distributed. The challenge of it is that is the level of data and reporting that we can do.”

That shift touches almost all aspects of the business. Let’s touch on a few key areas in ferment in the technology and metrics of podcasting.

Can you hear me listen?

People love podcasts. Right? Well, we certainly believe they do. But how much do people love podcasts?

We kind of know how many people download podcasts, but “downloads” only tell us how many of us moved a podcast off somebody’s server onto on own devices. They don’t indicate if we hit play or for how long we listened.

So how much actual listening is there? Edison Research tells us that the average listener downloads five shows a week, but has no good way of counting how many minutes were actually listened to. Nobody else does either.

With “time on site,” or engagement, now coming to define success in much of digital media, for the most part, podcasters can only offer the hard measure of downloads to would-be advertisers.

Why don’t they know? Chalk it up to two reasons. First, podcasting’s been such a small ad business that the data hasn’t been that essential. Secondly, the company that would be in the best position to gather that data, Apple, hasn’t shown in any interest in it.

Apple — described well by Matt Lieber as a “supportive but hands-off” platform and by Panoply chief content officer Andy Bowers as “a remarkably level playing field”– sits plunk in the middle of our listening. Fully 60 percent of downloads start either through iTunes or the iOS Podcasts app. Apple says it doesn’t collect data on actual listening, and so there’s none to be shared. Though it has participated in podcast industry talks, the company has shown no major inclination to change its role.

Of course, podcasters who also offer listening players directly to consumers can generate valuable first-party data. “I know if they listen for five seconds and stop. I know if they’re moving or stationary. I know all kinds of stuff, because all of that is all captured in the data is given out by that app,” says Audible’s Eric Nuzum.

Then there are the Overcasts, Downcasts, Instacasts, Acasts, Pocket Casts, and Stitchers — podcast apps all seeking market share, each able to learn more detailed customer behavior. And then there’s Google Play, which along with some of the apps gets credit from podcasters for sharing some audience data. Each of the players, though, only can see a narrow part of the audience.

There’s a school of thought that ignorance of listening behavior may be bliss. Within that that debate, we hear one that echoes debated about the free-spirited blogosphere of more than a decade ago. Then, some worried that the little guys with small audiences would get lost in the shuffle and that a mode of free expression will be commercially overrun.

Marco Arment, creator of Overcast and earlier CTO of Tumblr and creator of Instapaper, makes that argument forcefully, headlining one piece, “Apple’s actual role in podcasting: be careful what you wish for.”

On the web, getting more data was easy: web pages are software, letting their publishers use JavaScript to run their own code right in your “player app” (web browser) to creepily record and analyze every move you made, selling you more effectively to advertisers and letting them algorithmically tailor their content to maximize those pennies at any cost to quality and ethics.

Podcasts are just MP3s. Podcast players are just MP3 players, not platforms to execute arbitrary code from publishers. Publishers can see which IP addresses are downloading the MP3s, which can give them a rough idea of audience size, their approximate locations, and which apps they use. That’s about it.

They can’t know exactly who you are, whether you searched for a new refrigerator yesterday, whether you listened to the ads in their podcasts, or even whether you listened to it at all after downloading it.

Big publishers think this is barbaric. I think it’s beautiful.

Big publishers think this is holding back the medium. I think it protects the medium.

People in the industry I speak with understand the sentiment, but still largely push for more data. They’ll tell you they understand the value of not repeating commercial radio’s fall into cacophony — and they think they can avoid it.

While almost all the leading companies in the industry are pushing for better knowledge of listening habits, some suggest that we underestimate the data that is currently available. They point to the crazy quilt of metrics associated with other media and emphasize what they do know.

“I think the people that look at podcast metrics as not yet ready for prime time might reframe their thinking…Look at the metrics you get from radio, for example,” says Tom Webster, Edison Research’s vice-president of strategy and marketing. “Metrics you get from podcasting are superior to the metrics you get from radio.” Radio’s quarter-hour ratings (Average Quarter-Hour, or AQH), assembled by Nielsen, have been that industry’s long-time standard, though its methodology has drawn its share of critics over the years.

They point to the combination of other data points that fills in picture of listening. Dynamic ad insertion increasingly provides data on the listenability of ads. Surveys, commissioned by NPR, Podcast One, and a number of other podcasters, and performed by Edison and others, provide demographics insights into their audiences.

All that said, most expect it will take greater industry standards — first on downloads and then for actual listening — to really bring in and grow brand advertising. Real time spent is the objective here. Podcasters know they’re stealing minutes from old-fashioned listening, but today they can’t assert how many.

Other measurements bedevil the trade as well. For instance, what’s popular? Podtrac — a 10-year-old podcast measurement and ad company — launched its top 10 rankings by download and stream this spring, to understandable controversy over its methodology. Gimlet Media is among those not participating, but for those that do, including NPR and WNYC, they can use the rankings to point to their positions (number one and number two, respectively).


How do you find all this great stuff? Let’s do the math. 300,000 podcasts. Twelve square inches of smartphone. That’s the beguiling conundrum of the podcast trade. In digital media parlance, call it discoverability.

“Most people still find out about a podcast because someone they know told them check something out,” says Kerri Hoffman, CEO of PRX and a 14-year veteran of podcasting times. “How do we improve that? How do we try to make on-demand listening as simple as turning on the radio?”

Think of how you learned of a new podcast. It’s likely you heard from a friend or from a tout offered by a podcaster to whom you listen. That’s old-fashioned word-of-mouth virality, almost untouched by our social media lives.

Panoply’s Jacob Weisberg believes that current frictions are the enemy of adoption. “One of the things we confronted is it’s still much too hard to get podcasts. People who know how to do it do it quite easily. People who don’t know how to do it find it a little bit intimidating. There’s so much infrastructure that needs to be built and upgraded.”

As Panoply works through those issues, the new RadioPublic is joining the effort. An outgrowth and cousin of PRX, the RadioPublic app will debut this fall. “We think we can do it in a way that really does boil down to just hit play and it’s there for you,” says Jake Shapiro, now CEO of RadioPublic and also a 14-year builder of podcasts at PRX. “RadioPublic is entirely a mobile listening strategy,” Hoffman says.

At the same time, the big public radio players are making new moves to meet the next step of this revolution. The NPR One app — launched two years ago — focused on a curated stream of public radio stories. Now it devotes a third of its smartphone screen to “Featured Shows” — its podcasts. Similarly, WNYC’s Discover app, which focused on streaming, now plans to make podcasts more easily findable. For both, that line between “streaming” and “podcast” thins.

Edison’s Webster says its data affirms the rationale for thinking of streaming and podcasting as fraternal twins. “The dominant paradigm for listening to a podcast is to click on it and listen to it immediately,” he says. That’s a huge shift from the first wave of podcasting. Back then, podcasts were nearly always asynchronous: You downloaded an episode to your hard drive, moved it over to an iPod, then listened to it later.

Says Webster: “There’s a lot more on-demand consumption. Further, the user profile of podcast listeners has become more and more mass. The mainstream consumer doesn’t have this desire to hoard podcasts to listen to later. They’re just clicking on them and listening to them.” In fact, he says, Edison’s data says that about three-quarters of those who download a podcast listen to it, or at least most of it. In fact, he says 87 percent of those who download a podcast start listening within 48 hours.

In other words, as the industry tries to parse the differences between live streamed listening and downloaded on-demand podcasts, that difference may be an abstract one to most consumers.

Says NPR COO Loren Mayor: “They’re not distinguishing in the same way that we might as the content creators between, ‘Oh, this is a new piece, and this is a podcast piece.’ It’s all just good storytelling for them.”

The nuances in all these discovery conversations can have profound engineering implications. That’s the work that’s being done behind the scenes to meet the challenge of current — and newer — podcast listeners.

What is it that all these apps want to do? That’s simple: They want to make on-demand listening as simple as turning on a radio. We stretch back to old media to provide the simple metaphors that takes technologists years to make work. Similarly, in this new golden age of higher quality TV (and complicated set-top boxes), we crave the ability to just turn on our big screens and have the show appear.

On-demand audio, like on-demand video, will remain a work–in-progress…until it’s done. That won’t be 2017, but we can envision it happening by 2020. Whoever figures it out, whether it’s one of the current apps or an Apple/Google/Facebook/Amazon, we don’t yet know.

The ad serving revolution

These two words drive tech’s greatest impact on podcasting in 2016: dynamic insertion. As I detailed in Part 2 of this series, the practice offers three key benefits to the trade:

  • Allows on-the-fly insertion of new (host-read and other) ads into all shows, new and old. That’s a huge plus for shows that find as many listening to older programs as this week’s.
  • Enables the serving of different ads to different listeners, first on the basis of geotargeting by city, a first application of the kind of digital demographic and geographic targeting that’s becoming standard.
  • Collects audience data on actual ad listening, and more data that can be useful both in programming and ad targeting.

Art19’s Carr sums up the two big questions that the new ad tech intends to answer: “How long did they listen? Exactly what did they hear?”

Among the top several hundred podcasts, this dynamic insertion revolution will be a fast one. Perhaps five percent of podcasts used dynamic insertion in 2015, says Carr. By 2017, as many as 90 percent will. That change both opens lots of doors — more ad inventory added overnight — and forces all kinds of challenges in ad production and sales.

Into the future

How quickly can podcasting use — and consciousness — grow? Let’s remember the big numbers: Of the 288 million people online in the U.S., 80 percent are still only the potential market for podcasting. Only 57 million listen monthly now.

One place they may be further exposed: Facebook.

It’s been a curiosity: Why haven’t podcasting and Facebook yet married? If non-digital social virality is the leading referrer of podcasts, why hasn’t the greatest social referrer of all time emerged here? While Facebook most recently, and Google over a longer time, have directed us to much news and entertainment, they aren’t significant players in podcast listening or the podcast business.

In part, that’s because of that unique file structure of podcasting. It’s a format apart. In addition, podcasts can’t — yet — be easily snippetized. They are a longform medium.

Also, we’re all probably more careful about trusting a recommendation that demands 20 or 40 minutes of our attention, not just a few seconds or minutes of reading or skimming.

But the gap between the social and podcast worlds may be closing. Celebrity podcasters lead one way, eager to exploit the followings they’ve assembled on social platforms. When Shaquille O’Neal wants to promote his PodcastOne show, using his 12.6 million Twitter followers is a natural.

Emerging podcast network Wondery just bought ads on Twitter and Facebook for its highly rated true crime podcast Sword and Scale launch. Does it work better or faster than the word of mouth that has built early podcasting listener-to-listener recommendations? “It’s too early to tell,” says CEO Hernan Lopez.

WNYC tests out another social approach. “Our data news team developed what we’ve been calling an audiogram generator,” says WNYC head Laura Walker. “Shows can sort of serve up select quotes from their podcasts as like a kind of teaser.”

Whether it’s audiograms or some other forms, look for more sampling of podcast audio on Facebook. Sampling is an imperative in all digital trades.

I believe this question of audio segmentation will find greater life. Just as The New York Times has figured out how to include the right 45 seconds or so on its mobile home page, embedded appropriately in a story, audio will find a way to become part of presentation. It’s unclear how many moving pieces in podcast tech that may require. But it’s clear that the right audio segment, 15 seconds to several minutes, can both boost podcast consumption and bring greater context to current text/video storytelling.

In fact, the Times just added well regarded Andy Mills, a Radiolab producer to its audio team of half a dozen. Said the Times in its release: “His focus will be on creating mini-series around some of our best enterprise reporting and storytelling.”

For Audible, “The next version of the app will have a lot more social sharing tools in it,” says Nuzum. “Those who are more fluent with Facebook could take clips of things that you hear in Audible or share.”

Lastly, let’s consider personalization. Currently, the state of that art in podcasts is geographic. If Ira Glass or Marc Maron is coming to your town soon, your podcast may include notice of an event, while others listeners would get other promotions.

NPR One’s evolving streaming/podcast app personalizes what we hear and when, depending on the choices we make. At WNYC, the quest is similar. “In our Discover app, the biggest change will be an increased relevance of results,” says Walker. “Users will be able to customize according to show, not just topics. For instance, if someone is a huge Here’s the Thing fan, but Studio 360 less so — and they put in arts as a favored topic — they’ll only get arts segments from Here’s the Thing.”

Such personalization, in the welter of infinite choice, should be at first welcomed by listeners — and then, like all things digital in our lives, become an expectation.

Tomorrow in Part Five: A peek into the near future of podcasting.

The research for this series was commissioned by Public Media Futures Forums, a project funded by the Wyncote Foundation to support in-depth analyses of strategic challenges facing public media. The analysis was done independently and without any foundation review.

Photo of microphone by lincolnblues used under a Creative Commons license.

POSTED     Sept. 15, 2016, 12:15 p.m.
SEE MORE ON Business Models
PART OF A SERIES     The newsonomics of podcasting
Show tags
Join the 60,000 who get the freshest future-of-journalism news in our daily email.
Are you willing to pay for Prepare to be asked before year’s end
The cable news network plans to launch a new subscription product — details TBD — by the end of 2024. Will Mark Thompson repeat his New York Times success, or is CNN too different a brand to get people spending?
Errol Morris on whether you should be afraid of generative AI in documentaries
“Our task is to get back to the real world, to the extent that it is recoverable.”
In the world’s tech capital, Gazetteer SF is staying off platforms to produce good local journalism
“Thank goodness that the mandate will never be to look what’s getting the most Twitter likes.”