Nieman Foundation at Harvard
Editor’s note: Hot Pod is a weekly newsletter on the podcasting industry written by Nick Quah; we happily share it with Nieman Lab readers each Tuesday.
Welcome to Hot Pod, a newsletter about podcasts. This is issue 123, published June 13, 2017.
We’ve got a lot to talk about.
Here’s the headline: Apple is finally opening up in-episode analytics for podcasts. The data will be anonymized, consistent with Apple’s general stance on privacy, and the new analytics layer is scheduled to arrive with the iOS 11 update this fall. This means that podcast publishers will, at long last, receive data that tells them just how much of their episodes are actually being listening to — within the Apple Podcast app, at least, which is still largely understood to serve the majority of listening. (Estimates, however sampled, tend to range between 60 and 80 percent). Previously, podcast consumption was chiefly conceptualized based on downloads, a black box metric that’s criticized as lacking the level of granularity that are table stakes for advertisers buying on digital platforms in 2017. With this announcement, that measurement issue — long articulated as the defining problem of the medium — can finally be meaningfully interrogated, with many believing that the hurdle impeding advertisers from committing more dollars to the space can be thrown out the window.
But some are also arguing this change will bring a mixed bag of consequences, and in some ways, the new data puts the space at risk of snuffing out various dynamics that make it special. Which is to say, while there’s a hope that this will finally lead to podcasting realizing its full economic potential, the shadow of Web 2.0 looms large.
The WWDC session also contained a few other useful announcements, including a design overhaul for Podcasts app and new extensions to feed specifications that would give publishers more control over how they can present episodes within RSS feeds. Among other things, publishers will now have the ability to bundle episodes by season and signal which episodes are actual content versus extras like trailers. Noted Apple writer Jason Snell has a good rundown on this over at his blog, and you can check out the spec document here. And as I mentioned last week, this is probably what the redesign looks like, courtesy of this Reddit thread. (Once again, your mileage may vary with sourcing Reddit.)
But let’s get back to the analytics stuff. Since Friday’s announcement — which you can watch in full at this link, but only on the Safari browser, because Apple — there’s been a ton of writing appraising the matter, and in case you’d like a quick primer, I recommend this write-up by Recode’s Peter Kafka, which also contains screenshots of the upcoming analytics dashboard. (I’m going spelunking in some rabbit holes here, so a primer this is not.)
Here, we’ll attend to wonkier questions: What does this new analytics universe portend? How will the podcast business change? If so, who wins and who loses?
I wasn’t born a prophet, so I don’t know how exactly this will play out, but I do have some notes and assessments on a bunch of the key issues. This write-up is by no means comprehensive, and I’ll be exploring more questions in future issues as we deal with the consequences of announcements. For now, let’s jump in, and we’ll move through a bunch of topics.
Just double-checking: Is this really a big deal?
Yep, I’m pretty certain it’s massive, but it’s worth weighing the counter-argument. Even if Apple serves a majority of all listeners, the argument goes, it doesn’t account for the whole listening universe, and as such there might be muted effects to how this ends up moving the way business is being done. I’m not sure I’d buy much stock in that view: first, not only does most listening quantitatively happen on Apple, the company is qualitatively synonymous with the space. Second, there still doesn’t appear to be a strong alternative to Apple with a big enough consolidated market share that could meaningfully challenge (or avoid) the way Apple defines audience measurement. Which means that, in June 2017, it’s still feasible to think that whenever Apple says jump, most folks are still pretty much going to make like Durant.
How will the new analytics layer change the way we currently understand podcast audiences in the aggregate?
A couple of parts to this:
(1) Many believe that an ecosystem-wide audience resizing is in the cards. Because the vast majority of podcast audience appraisal is conducted based on downloads — and because we don’t actually know what happens to an episode after it’s downloaded — the way podcast audiences are represented, understood, and sold is almost certainly going to change. Just about everyone I spoke to frames this in terms of some form of downsizing, which makes intuitive sense, because there will always be some percentage of episodes being downloaded that are left unlistened (and ads left unserved). But the positive spin I’m given is that this change nevertheless comes with a higher level of accountability, and the gains in trust from advertisers will likely lead to much greater gains over the long term.
As Matt Turck, Panoply’s chief revenue officer, puts it, “I’m assuming we will see listener numbers fall short of download numbers; however, the benefit to making analytics far less mysterious should vastly outweigh the concern.”
(2) That said, there remains the possibility that the new in-episode analytics layer might reveal inconvenient truths about audience behavior. I’ve been told there are a few non-Apple tools and platforms (like Spotify and some third-party listening apps) with in-episode analytics already in the market, and while they only supporting a minority share of listening, the consumption data they’ve been collecting suggests there’s nothing especially revolutionary hiding in those new numbers.
Aaron Lammer, of Longform and Stoner, is one among the skeptical. “I would push back against the idea that there is some great insight lurking in these analytics,” he said when we chatted over Twitter. “As people who’ve set up elaborate app-based analytics hooks where you can track everything will tell you-there isn’t that much interesting… I’d rather look [at] it as standardization rather than revolutionary shift.”
That point on standardization, I think, is really important to file away in your head.
(3) Bryan Moffett, the COO of National Public Media, made a good observation on how the proliferation of dynamic ad insertion technology might mean the transition to an in-episode analytics world would still contain tricky imprecision.
To quote him in full:
A dynamic ad server will serve up many different versions of a single episode. They could vary in length by a few minutes or even more. For example, if one user gets an episode of TED Radio Hour with four dynamic :30 sponsorships and a :30 promotion block in its hour of content, but another user for some reason gets the same episode with just two :30 sponsors, the length difference is over a minute and the content is not aligned minute by minute for each episode.
Apple’s analytics rolls up all listening to a given episode and averages, so there is bound to be some imprecision. It’s not a lot, and it’s certainly a better world than the one we live in now.
It’s never easy shifting gears.
How will the podcast business be affected?
Time will tell, obviously. But here’s the range of the thinking out there:
(1) As I mentioned, there is a sense from some bigger publishers that this new analytics layer will finally allow them to kick open conversations that may meaningfully unlock long coveted brand advertising dollars. Contrary to direct response advertisers, whose intended outcomes (and measurement methodologies) additionally revolve around conversions off promo codes, brand advertisers are generally thought to require a higher level of trust in the impressions being reported back to them. Podcasting’s black-box download-oriented measurement universe has long been described as the primary hurdle preventing brand advertisers from allocating more dollars to the medium, and it is believed that Apple’s in-episode analytics are a significant first step forward in opening up conversations between brand advertisers and podcast publishers across the system (conversations that have to do with perception as much as actualities).
(2) But how does this development affect the direct response side of the podcast advertising business? There’s a general belief among the folks I’ve talked to that direct response advertisers, or performance-based advertisers, will likely be stable, though there appears to be suspicion that the new analytics layer presents yet another horizon of opportunities for those advertisers and their respective agencies to haggle more over prices. I’m also being told that there are expectations of some oncoming turbulence/fluctuations in price points, as those advertisers go through the process of figuring out how to integrate this new data layer into their current practices.
(3) There are two versions of the apocalyptic view on the business end. The first takes the shape of some worries about ad-skipping, and what the new analytics layer is going to reveal about the extent of this behavior. (For more background on this, read this Wall Street Journal from last summer). The end-times scenario is said to be one where it’s discovered that podcast ads are skipped over at such a volume and intensity as to kill their value. On this front, the responses seem to generally track along the built-in split between brand advertising and performance-based advertising; there is a sense that, even if there is a problem, it would mostly affect the former, while the latter would remain somewhat stable, because conversions are still taken to be more important than impressions. Again, the positive spin I’m served ties back to a sense of greater accountability that the new analytics layer brings into publisher-advertiser interactions: we’ll know who is actually providing value to advertisers, and we’ll know who isn’t doing so as much. As Midroll chief revenue officer Lex Friedman said, “Podcasters who are confident that people are listening to their ads should be very happy about this.”
The second apocalyptic argument presents a scenario where podcast CPMs plummet, ultimately leading to the collapse of the market. This view generally draws on a parallel between podcasts and what happened to blogs once the format started experiencing waves of ad tech development. Personally, I can’t quite see the specifics of how this move by Apple could bring those dynamics to podcasting just yet. My understanding of the plummeting blog CPMs pegs the phenomenon to the continuous structural devaluing of blog advertising real estate brought on emerging ad technologies that gave advertisers (and ad tech companies) unchecked leverage. And while I think the broader risk of podcasts possibly going down the road of blogs is absolutely real, I don’t have a sense that this new analytics layer alone automatically leads to a devaluing of podcast advertising real estate. If anything, Acast’s recent rollout of a programmatic podcast advertising product is more likely to incur those types of effects, should the tool ever get traction — this development from Apple strikes me as a step forward that’s small enough to stop short from these effects.
Who wins, who loses?
(1) Obviously, publishers who have made a practice of inflating download numbers will get checked — though the counterargument that all metrics, without active third-party verification, can be gamed over time is certainly a prudent one.
(2) An argument can be made that this system-wide shift to a new analytics standard would usher in a weeding-out period. Podcasts delivering strong ad value will get additional data to strengthen their appeal for more advertising dollars, and podcasts not doing so will be flushed out of the ad market. It would mean that high-performing podcasts would be in a better position to extract more value, while not-so-high-performing podcasts would have a harder time accessing advertising dollars.
(3) It should be considered that whatever audience readjustments happen will probably disproportionately and negatively impact smaller podcasters’ ability to derive advertising revenue. Which is to say, just as how every publisher experiences the turbulence of discovering that its meaningful listening audience size is probably going to be smaller than its downloads, smaller podcasts will be whipped around harder, and in some (if not most) cases, that could lead to those shows falling beneath a certain threshold for advertising consideration. That’s bad for podcasts with already relatively small but meaningfully engaged audiences. In these cases, there are presumably two available moves: first, lean deeper into a niche that maintains a specific appeal for relevant advertisers, and second, pursue other non-advertising revenue streams.
I suppose, generally speaking, it’s worth keeping in mind that advertisers need to be served value too, and also, advertising isn’t necessarily the only business model available to publishers.
Content considerations. Metrics and measurements have long informed the way programs are created, and we should probably expect to see the dynamic express itself further with the new analytics layer. A couple of threads to consider:
(1) Knowing just how much of episodes are being listened to presents a much better feedback loop to improve not just editorial products, but also advertising products. And there is also the likely effect that we’ll see the blossoming of new formats, genres, and show structures that come from playing toward what the new metrics tells us.
(2) On the flip side, there should also be room for the more general worry that we’re sliding into a world where metrics outweigh creative decisions. I think there’s always room for that concern, regardless of whatever metrics are available — there will, to some extent, always be operators looking to play to the numbers rather than actually use the numbers to make better work.
(3) I’m pretty drawn to the question, raised here on Twitter by The Atlantic’s Alexis Madrigal, of whether increased data granularity within a medium would lead to the detriment of experimentation within that medium. Instinctively, I feel as if there is some truth to this, but I also suspect experimentation has less to do with the available metric universe and more to do with the ways in which compensation is structured off those metrics. (A quick tangent: I also find myself wondering how “experimental” material is defined; personally, I tend to grade experimental-ness relative to however the medium currently behaves, and think experimental programming will exist in any format regardless of where it is in its life cycle. I think the more interesting question here is about the conditions under which “experimentation” can exist within high-budget and high-scale productions.)
I’m not even close to being done, but I’ll leave it here for now. Obviously, this enormous and complex development contains many, many layers, and I’ll continue to dig around and write about them in future issues. (I mean, that’s why Hot Pod exists, right?)
Here are some of the questions I’ll be thinking about:
As for my own normative view on all of this, I’m still figuring it out. I do think that the podcast industry is indeed still comparatively tiny, as Recode’s Peter Kafka points out, with podcast ad spending projected to only be about $250 million this year. While it’s growing at a solid and steady rate, it’s still peanuts compared to where radio (about $14.1 billion) is today, and there’s more to be gained and lost from changing how business is being done today. And like Kafka, I do think change was going to happen no matter what.
Also, as I mentioned on Twitter, I find myself skeptical about the nostalgia and privileging of the status quo. But that’s a story for another day.
Roman Mars, Esquire. New Hampshire Public Radio’s Civics 101 has some new competition in the form of a somewhat surprising side project from the 99% Invisible chief: “What Trump Can Teach Us About Con Law” is an explainer podcast that features Mars being taught the basics of constitutional law by UC Davis professor Elizabeth Joh based on ongoing developments in the current iteration of the White House. I’m told that the podcast is officially produced under the Radiotopia banner, which brings the number of Radiotopians with two podcasts up to two (the other is Hrishikesh Hirway, who makes both Song Exploder and the West Wing Weekly for the indie podcast collective). Mars’ new podcast comes mere days before the launch of another new Radiotopia podcast, Ear Hustle. That’s scheduled to roll out later this week.
Career spotlight. Spend enough time in the New York podcast scene — or any major city with a podcast scene, really — and you’re bound to bump into someone who came up through WNYC, which was once the city’s only major institution dealing with narrative radio. In this week’s Career Spotlight, we’re bumping into Leital Molad, who currently leads podcast development for the Pierre Omidyar-backed First Look Media.
Molad adds that she’s on the lookout for more female voices, and that interested parties should get in touch. You can find Leital on Twitter at @leitalm.
Bites
