James Cridland

Is it time to stop auto downloads by default?

I think it might be time to stop auto-downloads as the default behaviour of podcast apps. I’d like to explain why they exist in the first place, and why they should go away.

Why we have auto-downloads

When podcasting started in 2004, most mobile phones had little or no data connection.

GPRS could achieve, in the best conditions, a 9.6 kbps connection to the internet. Wifi was almost entirely kept out of mobile phones until the launch of iPhone in 2007 - the cellphone networks called the shots, and demanded that wifi was disabled or removed from phones that they sold and subsidised.

Podcasts, as the word suggests, were for the iPod and similar players, none of which had any internet connectivity. The way they worked was to use a computer connected to the internet to download the audio files: and then, those audio files were transferred to the device.

Podcasting survived on the concept of downloads - for no other reason that the concept of streaming that audio was beyond the capabilities of most of the devices out there.

And, as podcasts made their way to mobile phone, they took this “download” paradigm with them. Apple Podcasts appeared on iPhone in 2012 with a number of throwbacks: a faux reel-to-real tape deck for audio playback, a radio-type tuning dial for genres, and the option of auto-downloads. On the original iPhone app, while you could “stream” the audio direct if you wanted to, you could auto-download it too.

Auto-downloads made sense in 2012. Phones included 3G, a much faster mobile data technology which could easily handle streaming audio and video: but these data plans were both meagre and expensive, and few people were rich enough to be able to afford to use data in such a fashion - if you were lucky enough to be in area with 3G coverage. Many weren’t; and in the UK, at least, all iPhones came with pre-configured access to BT Openworld’s wifi hotspots to partially alleviate pressure on the phone system.

Things have changed

In 2024, though, things have changed. Mobile data is almost ubiquitous - and for many, either unlimited or virtually unlimited. My Australian data plan is 180GB a month for AUD $59 (about US $40): but if I exceed that generous allowance, the phone company doesn’t cut off my data: it slows it down to “just” 2Mbps. In essence, it’s unlimited data.

For most users, YouTube just streams video. Spotify streams music. Google Maps streams the map data to you as you get directions. Netflix streams. Amazon Prime Video streams. Almost everything relies on streaming data.

Except podcasts.

Podcasts are open and unencumbered by digital rights management. It makes downloading them simple, since you don’t need to worry about expiry dates or subscriber authentication.

Perhaps because of that, Apple Podcasts - responsible for around 40% of all downloads in this market - will still automatically download episodes to your phone, as long as you listen to the podcast once in a while. There’s an algorithm that turns off auto-downloads if you stop listening.

Overcast, another popular Android app, also does auto-downloads. It appears to auto-download until your phone almost fills.

The problem with auto-downloads

They’re expensive

In 2021, a short-lived bug meant that Apple stopped auto-downloading. It was quickly fixed, but most podcast hosts saw a dip of around 25% of downloads from Apple Podcasts at the time. (When you hit “play”, the show was a user-initiated download instead, and played instantly anyway). So we can reasonably assume that 25% of all auto-downloads aren’t ever played.

If my experience is anything to go by, it’s likely that the number is much higher for daily podcasts, like The Daily or Up First.

But let’s go with 25% of all downloads not being listened-to. Apple Podcasts has roughly 40% of the market, so you could look at Wondery’s 20 million monthly US downloads in the Triton Digital Podcast Ranker, and calculate that 2mn are never listened-to.

How much do auto-downloads cost the podcast industry?

Podtrac measures 8.1bn podcast downloads a month. Assuming that they’re all 30 minutes long - that’s 93 petabytes of useless data transfer, at an AWS Cloudfront cost of $1.9mn, every month. And Podtrac’s tracking code is in 7.6% of all episodes. The cost of auto-downloads is significant.

They’re bad for advertisers

Auto-downloads aren’t a hidden secret for advertisers; everyone knows that they exist.

But, because 25% of shows are never listened to on Apple Podcasts, that means that on Apple Podcasts, ads are 25% less effective than they should be.

Alternatively - when you stop auto-downloads, that means that the ads will be 25% more effective.

This doesn’t mean downloads should disappear

I’m suggesting here that we stop auto-downloads by default.

You can download videos on YouTube (if you’ve a Premium account), so you can watch in a plane. But YouTube doesn’t download by default.

You can download music on Apple Music, on YouTube Music, and on Spotify. But these music services don’t download by default.

You can download podcasts on Spotify, on Google Podcasts, and on YouTube Music. But these podcast apps don’t download by default.

The cost of data is still higher than it needs to be in some countries and for some income brackets. Downloads while on wifi so I can listen later when I have no connection are a good, good thing. I’m not suggesting we get rid of them.

What I’m suggesting here is that we get rid of the default auto-download behaviour. So you have to deliberately choose to download a podcast, rather than the app doing for you anyway.

If we do that, we can save millions of dollars, we can make advertising work better, we can get better statistics on how many listeners we have, and most people won’t notice any difference.

So - is it time to kill the default auto-download?

Where have I gone wrong in this argument? Please mention me on Mastodon, or email me, and let me know.