92 comments

[ 6.2 ms ] story [ 159 ms ] thread
The article implies throughout that audiences are shrinking but the only mechanism for that is Apple Podcasts won't download episodes for a user who isn't listening anyway.
The article implies there never were audiences as estimated, due to faulty mechanism.
It does seem like the "shrinking" is confusing there. Probably should have been "The Podcast Industry Is Smaller Than Producers Thought"

It is possible that overall podcast listening is still up yoy. It is also possible that even if the top podcasts decline (which the article sorta hints at?) that it may just be because the market is now saturated with thousands of competitors that each take a very small slice.

It like those articles that talk about the year millions of US children disappeared- because the tax forms start requiring social security numbers for claimed dependents.
So it's not necessarily shrinking, it's more that the biggest podcast publisher altered how they are delivered based upon current listener usage rather than automatically deliver every episode forever, and it turns out that most people actually didn't listen to every episode forever.

Certainly this discrepancy has to have been known about for some time. I speculate that the ad industry had some influence on this decision to bring listener metrics more in line with reality.

There's no need to speculate, as the article literally says that: "Industry groups agree: Sounds Profitable, a podcast monetization trade group, lobbied Apple to make the change, and said that it would lead to more accurate audience info and more effective advertising."
Great example of how the invisible hand of big tech enacts large scale behavioral change.
(comment deleted)
I'm not sure it's an invisible hand - seems pretty well known amongst people affected by it (podcast producers, advertisers). This is the second time I've heard of this downloading change in the past few weeks, and I've got no connection to the podcast industry.

And what's the large scale behavioural change? People weren't listening to all the podcasts that the app had been downloading, and they still aren't?

The behavioural change will be one the podcaster's side, who now have to worry more than previously about keeping listeners enganged.
It’s amazing how little analytics there are on podcast listening by default as a standard that all apps should provide. Stuff like if an episode was started, marked as listened (but not actually listened to), etc.

I get the reason for this change in general and the idea behind it, although this still gives inflated numbers for a minimum of 5 episodes.

Please no. I don't want to be forced to use a special player to listen to podcasts just so analytics can be collected
But if we put it on The Blockchain, you can mint PodCoins by listening to podcasts! Of course, to ensure you're really listening, you'll also need a secure cryptographic module installed in your auditory cortex. Depending on how well that surgery goes, maybe it finally can put the killer in killer app.
This is the wrong kind of cyberpunk, I was assured there would be USB-key knuckledusters so I could punch 1337haxX into corpo terminals.
No, I fully agree, the idea being that there’s a standard set of analytics and a way to report on them that apps can openly support.

Granted the odd ball out would be Apple and if they’d support the analytics in their podcast app, but much like Apple’s podcast directory is the defacto directory, there’s an opportunity for a better, OPEN, solution.

EDIT ADD: I’d be surprised if every podcast app doesn’t already collect a robust amount of analytics on listening, they just don’t report it where podcast owners could see easily.

These are common events used by the ad industry as well
The reason podcasts are so great as a medium is that they have not yet been “platformified” and many still just distribute with RSS.

Do not enshittify podcasts.

Spotify took a dear podcast of mine 'private' (it has since been freed). Some have their own paywalls. Even some 'smaller' podcasts now feature network leaders. Many thus feature localized ads.

Not sure we're not already on that path.

> Some have their own paywalls

Nothing wrong with paying for a product. A common method is a private patreon RSS (Atom) feed. It's still an open standard but the podcast earns money. It's at least a better option than the path Spotify took.

> Many thus feature localized ads.

I'm not against dynamic ads per-say, but it really reminds me of TV with the Ad's being wildly irrelevant to me and often being significantly loader than the actual podcast

That kind of reporting has clear privacy implications. I'd avoid any client that reported those metrics.
Pretty Sure Netflix has this, though only used internally.
I’m sure they have incredibly granular information about exactly where in a video most people spend their time watching, where they stop watching, where they pause, where they fast forward/rewind, and probably more. As a company which has first-party content, it would be shocking if they weren't collecting it.

Some porn sites (i.e. PornHub) make info about where people spend the most time watching available in the UI in the form of a small graph over the seek bar at the bottom of the video. Even though the adult entertainment and sex work industry has historically been a pioneer of new tech, I don’t think they’re alone in tracking that stuff.

And I’m not thrilled about any of that. I’d be upset if my podcast app started reporting more info. I don’t really care about the specific privacy implications of knowing which podcasts I listen to, but I do care about the loss of privacy in general, especially because some people might be potentially more affected than me.

Does... Netflix do podcasts now?
No, just commenting on how streaming of any digital content, the source company is tracking hundreds of data points on each person.
That is because a podcast is just an rss feed of audio files… to get that sort of information (from all your listeners) you have to add to rss and break compatibility.
What about spotify? Their model should be strong if the others don't count streams.
I think we have the capability in RSS since the age of blogs and RSS aggregators.

Basically the aggregator can grab the main RSS feed and tell the server "I'm not just one person, I'm actually serving this to {subscriber count} people" and the server can use that information if it wants to.

There's a beauty in the fact that Podcasts are decentralized and built on RSS. Let's not ruin that.
It's "surprising" how bad the important metrics are even when it's technically so easy to get better ones. "years in the making"
My synopsis:

1. The Apple Podcast app has been updated to longer automatically download episodes of a show if the listener has stopped listening to the show.

2. This has resulted in the "download count" metric crashing for several high profile podcasts the publish frequently -- The Daily from NYTimes is mentioned.

3. This lowers advertising rates and is basically bad for the publishers of those podcasts -- the "download count" for some shows has fallen by 40%.

---

Now to me this seems like an obviously good move because podcast episodes can be long and can eat up a lot of space on a device -- especially if they are never played, meaning that they never have the "delete after being played" opportunity.

Also, advertisers obviously care about how many people hear their advertisements, not how many people download the podcast episode, so the fact that advertisers even have to consider the "download count" metric at all means they are operating with inferior information.

Even when I do play a podcast that I have downloaded, I often skip the advertisements because it is so easy to hit the fast-forward-30-seconds button a few times.

But I have been assuming that podcast apps collect information about whether or not a podcast episode is ever played, and also on what parts of the episode the user rewinds or skips, and so forth.

For instance, it would be valuable information to know that of the listeners who actually download and start playing and listening to a certain 2-hour-long podcast episode, 99.4% quit listening within the first ten minutes.

I don't use the apple podcast app but I can verify I subscribe to 3-4 podcasts I rarely listen to. One is a news discussion podcast. Every episode is 1hour they post 7 days a week. Another is Escape Pod (scifi stories). I love but I rarely listen. There's probably over 100 back episodes on my app. There's also a few music podcasts that update 1-3 times a weeks each. Once in a while I'll play a few but never enough to catch up with what I've downloaded.

I suppose I should turn them off but I'm always thinking I might listen

> One is a news discussion podcast. Every episode is 1hour they post 7 days a week.

There isn't enough happening to fit 7 hours of news every week.

But it's a news discussion podcast, and you can discuss nearly any topic for a nearly unlimited time.
> Even when I do play a podcast that I have downloaded, I often skip the advertisements because it is so easy to hit the fast-forward-30-seconds button a few times.

The best thing is that ad segments are _always_ multiples of 30 seconds, so it's really easy to skip them :D

Overcast even has a "skip intro" option for podcasts that have a regular ad of the same length in the beginning every time.

It is their own fault as well, I for one have not skipped a single ad from Conan O'Brien needs a friend. That's how good the ad-reads are.
>2. This has resulted in the "download count" metric crashing for several high profile podcasts the publish frequently -- The Daily from NYTimes is mentioned.

>3. This lowers advertising rates and is basically bad for the publishers of those podcasts -- the "download count" for some shows has fallen by 40%.

As I understand it, when YouTube about a decade ago started using more accurate metrics, it turned out that the viewership for some very prominent corporate channels was way overinflated. This is why The Onion, CollegeHumor, and Funny or Die all suddenly vanished. (And probably also why QCode has become way, way less active on producing new podcasts, or delivering on its goal of turning podcasts that hit it big into film/TV projects.)

> CollegeHumor, and Funny or Die

I was under the impression that these two died due to Facebook. Facebook massively inflated video views (a video playing for 3 seconds without sound as it scrolled past in your feed counted as a view), these companies believed the bigger numbers on Facebook, so they heavily pivoted into Facebook videos over YouTube, with no monetization strategy, completely cratering in the process.

This is corroborated by someone who worked at CollegeHumor.

https://www.ccn.com/facebook-lied-about-video-metrics/

I was thinking of Facebook, not YouTube. Thank you for the correction and the cite.
> But I have been assuming that podcast apps collect information about whether or not a podcast episode is ever played, and also on what parts of the episode the user rewinds or skips, and so forth.

Why would you assume this?

(Most) podcasts are RSS feeds. You can use any app to listen to them. You can even write your own app to listen to them.

Why would you intentionally pick an RSS reader that told some rando everything you listened to?

That's what people do with YouTube; why wouldn't they do it with podcasts?
Because historically people listen to podcasts mostly offline not in websites or apps.
Sure, but the publishers could change this and require apps.
It's been tried. People don't want to use proprietary apps for podcasts. E.G. Spotify podcasts is a failure.
I think I'm going to throw my life savings into starting a new proprietary podcast app.
Honestly sounds like a solid business opportunity. Can I invest?
Certainly. Please DM me and I'll tell you all about it.
Because:

1. It comes preinstalled and is the default podcasts app

2. It works pretty well

3. People don’t care

4. The “told some rando” is basically the default with mass-market technology

5. That’s kind of how web ads work

This is good news for me. It would weed out those VC-backed podcast platforms that are selling these high numbers to advertisers, and those will be left are who have always been there since the bubble happened.
It strikes me that the mechanics of podcasts are so consistent and simple that it would be straightforward for podcast apps to implement truly privacy-preserving analytics. As an example, Apple could make available to podcast publishers numbers like “how many people actually listened to this episode”, instead of the current status quo of “how many people downloaded it” using server logs, which is obviously inaccurate as this episode makes clear. There’s no need for Apple to disclose or even collect anything identifying to make these numbers available. It’s just a big distributed SUM() operation.

Of course, once that was available, advertisers would next want attribution for who had listened to which ad segments, etc. and it would all be downhill from there.

They already do:

https://podcasters.apple.com/support/5392-listener-analytics

That’s what makes this article slightly puzzling. The change described in the OP article would not affect Apple’s analytics ( I think). It would only effect download counts for their party analytics.

Wow, thanks for the info! It’s crazy that the article didn’t mention this and that no one interviewed seems to have been aware of it. Feels like a huge oversight, since with this info the change should have been basically unimpactful.
That's called telemetry.

We hate that here on HN and always disable it.

Do you want Microsoft to snitch on your listening habits? I bet not.

The podcast industry is not shrinking.

It was never that big in the first place, apple simply made it clear.

It kind of has, many NPR shows have been killed, they were in trouble for a bit even.
Perhaps podcasts have broader appeal, but I always thought that talk radio was a good proxy for podcasts. I’d guess that only 1 in 20 people had a talk radio show they routinely listened to growing up pre-podcasts, and a smaller subset of that was engaged enough to not miss an episode. It felt popular, but still niche within the set of reasonably popular interests.

It would surprise me a lot if the switch to the internet as a medium changed the fundamentals much here. It’s probably easier to discover new podcasts than it was to find new talk radio shows, but does that really move the needle enough to justify the VC funding in this space?

I think so. The big shifts are:

-production costs have dropped so much so there are more options from more content

-broadcast costs have dropped so you can have 4 hour podcasts for basically the same cost as 1 hour

-audience is global instead of local, so way more listeners

-asynchronous nature makes it easier to fit into your schedule. I can’t listen to my favorite talk radio show because I’m working. But I can listen to my favorite podcast whenever.

So this should mean that there’s more people, and more money. So investment makes sense.

I think narrow appeal is the best part.

I listen to niche podcasts from all over the world in a variety of topics spaces that would never have mass appeal required for radio or television but a global asynchronous audience is enough.

> Perhaps podcasts have broader appeal, but I always thought that talk radio was a good proxy for podcasts.

I prefer audiobooks. At least audiobooks tend to have a beginning, middle and an end (and an editor!) so the content is a little more well planned vs just rando jabronis bloviating which sometimes happens in podcasts.

I can summarize most plots in a paragraph. The goal of most entertainment is to kill time and meander.
I just can't stand a podcast where it's just 2-3 (usually) men shooting the shit in front of microphones for multiple hours. There is no subject, no editing and no point. They might end up with a few 20-30 second funny clips on TikTok.

There are podcasts that are actually edited, those are the good ones. They stay on subject, boring / unfunny bits are edited out etc.

My favourite example in this is the No Such Thing As A Fish -podcast from the QI crew.

I didn't know at first it was carefully edited, but then they started releasing the outtake bits as bonus episodes and the curtain was lifted =)

There is a podcast I really loved called the Future of Coding, where they dived into cool CS papers, which now has turned into 2.5 hours of two people talking about everything that tangentially related to the topic at hand, with no editing. I really do not know who their target audience is.

I don't have 2.5 hours to dedicate to a single podcast episode, on a topic that might be quite complex, while having to suffer through jokes and poor pacing.

Some people tend to form parasocial relationships to other people.

Then they listen to those rambling podcasts to "keep up to date" about what's happening to whoever they chose to follow.

I had a colleague at work who listened to a YouTube "podcast" that was 3-4 hours per episode, they just had it on a second screed yammering away while they worked. I don't get it.

I got too much ADHD already on my own, I cannot imagine how it would be to have other people talking between themselves on a second screen while I'm trying to concentrate on work.
Howard Stern alone had 20 million listeners in 2006, which was more than 1 in 20 Americans. The biggest 4 US PM drive time talk programs anime are about 40 million, just going from Wikipedia.

Long form audio content isn’t a niche market all up. I do think it is a market that lends itself to a high degree of fragmentation because of low production and distribution cost, which makes it hard to monetize.

I never understood the boom given the economics of the space.

Pretty much all the podcasts I regularly listen to are part of a larger brand/media ecosystem so they clearly aren't intended to make money as a standalone. Mostly a mix of guys with day jobs blowing off steam talking to people they find interesting, or media people with full time jobs on radio/tv or in print/etc. Bloomberg for example puts out a lot of interesting ones.

Looks like Gimlet and Joe Rogan got out at a good time. Just a few years later and we'd inevitably see the bust as we do now. Gimlet is apparently already gone and was merged into Spotify's other podcasting brand.
What!!! That stupid feature that gobbles up your phone storage was fudging everyone’s engagement metrics (that they misinterpreted)? That is just bonkers.
The stupidity doesn't end there, OPUS/Speex has existed since 2013 and is supported everywhere. Why do they use mp3's for a medium that is mostly speech?

Mind you I'm not saying, toss away mp3's, but they could offer the low/mid/high bandwidth option, same as youtube.

And most of the content would work just fine in mono.

Even npr’s app would cache several hundred mb of content when you started and eat up bandwidth to the point I had to stop using it. Maybe that’s been fixed but yeesh.

I regularly stream a pirate radio station and AAC 32kbps is pretty damn good.

"Half the money I spend on advertising is wasted; the trouble is I don't know which half."

Then Apple stopped automatically downloading podcast episodes for people who aren't actively listening, and Spotify presumably counts actual plays rather than downloads, and now we know which half is wasted.

When you are what you measure, measuring more accurately isn't always a fun exercise.

Advertisers typically assign unique discount codes to each podcast so they immediately see conversion counts.
I think that’s only part of the equation. There also must be many people who hear about a product or service on some media and then sign up later without remembering the code (or that the podcast was the reason they are doing it). You could have a “how did you hear about us” question, but those are annoying.
Should track plays like music
It should have never become "an industry".
Why? (Honest question.) In my eyes podcasts is just another form of media, like books, movies or video games - all of these are industries?
Podcasts were great when they shared Web 2.0 ethos "Broadcast yourself". But today's conveyor belt of Spotify and other grifty podcast enterprises turned iPod casts into a lame adjunct to MSM/talk radio.
Forgot to mention that google podcasts will no longer be operational as of this year.

Honestly, at this rate, I'm surprised google hasn't killed gmail.

https://www.theverge.com/23891397/google-podcasts-youtube-sp...

Google consolidated the podcast features into YouTube Music. Many of the "killed by Google" entries are actually Google shuffling around features between different apps. Annoying, but the functionality is still there.

From the outside looking in, it feels like Google lets their org structure dictate their product strategy to a comical degree.

They did the same with Google Play Music, but the YouTube Music app was so much of a worse experience for me that I just switched to Apple Music.

So I do think moves like this still fit into some category of “killed by Google” lists.

Not to mention, you needed to download and use a special tool to download your Google Play Music songs that you'd paid for before the shutdown, which didn't actually work.
Functionality is also the interface, it demonstrates the functionality, a podcast app is an rss feed reader tuned towards playing longform audio. I sincerely doubt youtube music integrated with rss feeds would be any good for podcasting, I've tried it and it always tries to play the video (which, its youtube music, why would it play video?). Not to mention that podcasts already have ads in them, so now its going to have youtube ads and also ads in the podcast. Not to mention the ad numbers have tripled, there's 2 minutes of ads every 5 minutes.

Also we already have an example of podcasting functionality being integrated into a music app: spotify. And IMO it is a complete failure, the interface is constantly fighting between podcasts and music. Two methods of consuming media that are different, making the entire experience subpar.

Surely any sophisticated ad buyer would understand how "audience" is calculated, and know to account for people who download but don't listen, right? Like isn't part of the reason to have per-podcast promo codes is to figure this out?
Sophisticated? Ad buyers? Not likely.
This shrinking might be localised. Where I am from, every other celebrity or famous person from any other similar medium (TikTok, YouTube, Instagram etc) have started their own podcast channels on YouTube. And many of them are able to get many well known people on their 'podcast' YouTube channels.

If that is what podcast is about, then it's booming in my region(Pakistan).

You're right, it's a pretty common thing in many fields now. For example streamers on Twitch have realized that it's pretty easy to just get a group together with other guests for easy content.