Podcast news roundup: 5 November 2026

Two measuring sticks got swapped this week, and both change what a podcast number means. RAJAR, the UK's radio audience survey, stopped asking only whether you listen to podcasts and started asking whether you watch them too. Amazon, in a different market entirely, decided an advert that never reached a screen should not count as an impression. In between those two, a British show became the biggest podcast on YouTube, by subscribers, with nearly a quarter of its views arriving in languages its host does not speak. The edit desk takes the question a host asked me this week: how do you know when an interview is over, and what do you say?

Did The Diary of a CEO win YouTube by speaking twenty languages?

Podnews reported on 2 November that The Diary of a CEO has overtaken The Joe Rogan Experience as the most-subscribed podcast on YouTube. The figures are from FlightStory, the show's producer, so weigh them as a claim from the winner: 21.3 million subscribers against 21.2 million at 13:40 GMT on 1 November, up from about 14.2 million at the end of last year. Channel views from 1 January to 30 October 2026 were just over 1.03 billion, against 399 million for the same stretch of 2025.

The number I care about is the one Podnews pulled out: 261 million of those views were on audio tracks other than English, per YouTube Studio data. On the release's own count that is 23.3%, which it rounds to nearly a quarter. Spanish, Russian, German, Portuguese and French lead. The Zelenskyy episode, released on 22 October, was recorded in Ukrainian with two human interpreters, then published in more than 20 languages with AI voices, and passed 7 million views in about ten days.

From the edit desk, the multi-language audio track is the most under-used feature on YouTube for a show of any size. It is one upload, one master video, with extra audio tracks attached, and the viewer picks the language. You do not need a Ukrainian interpreter to start. You need a clean dialogue stem with the music and effects on separate tracks, which is how we build every master anyway, because without that stem nothing can be dubbed cleanly later. If your editor is mixing voice, music and sound effects down to one stereo file, ask for the stems now, before you have a reason to want them.

What does 35% of the UK actually count?

RAJAR's Q3 2026 results came out on 29 October, and Adam Bowie's write-up the same day has the detail that matters. RAJAR changed the podcast question this quarter, with separate measures for listening and watching. Weekly podcast listening is 30.7% of adults, up from 26.9% last quarter. Add watching and the combined weekly podcast consumption figure is 35%. Podnews, on 29 October, ran the 35% as the headline and noted that the old 26.9% counted listening only.

So be careful with the comparison. 35% against 26.9% is not a rise of eight points in a quarter; it is a new question. The like-for-like move is 26.9% to 30.7%, which is still a healthy step, and the gap between 30.7% and 35% is roughly the share of UK adults who watch a podcast in a week without, apparently, also listening to one. That gap is the video audience in one number, and it is the first time a UK survey of this weight has put it on the page. Our UK podcast listening statistics page is the place we keep these as they land.

Radio, for scale, per Global's RAJAR release on 29 October: 51.2 million UK adults a week, 87% of the population, which Global calls an all-time high. Podcasts are a long way behind radio in reach and a long way ahead of where they were. Both things are true, and I would rather pitch a sponsor with the honest pair than with the 35% on its own.

UK: the midterms, live, with the camera on

Global announced on 2 November that The News Agents and LBC were joining forces for the US midterms on Tuesday 3 November. Emily Maitlis and Jon Sopel in Washington DC from 1am to 5am, Lewis Goodall in Texas on a Senate race, Simon Marks on either side of them, Nick Ferrari taking the morning. LBC's night ran on radio, in the app, and streamed live on YouTube and Global Player. The News Agents recorded a dedicated midterm episode on Wednesday 4 November, with more from DC over the rest of the week.

Goalhanger did the same with The Rest Is Politics US. Per Podnews on 29 October, the show live-streamed its coverage from midnight Eastern.

Here is what I read in that. Two of the biggest UK podcast publishers treated an American election as a live television event, built around their podcast presenters, with the podcast episode arriving the morning after as the considered version. The live stream is the raw material; the episode is the edit. For a smaller show there is a usable shape in there. You do not need an election. You need one moment a year where being live makes sense, and then a properly cut episode from it. The live stream brings the audience; the edit is what they share.

Have the copied shows come down yet?

Last week's edition covered Podnews finding 1,778 shows copied into Spreaker. The follow-up, on 2 November, is slower than anyone would like. Per Podnews, Spreaker has cleaned up 72 shows from that list, and 1,706 shows remain in Apple Podcasts from Spreaker that are also in Apple Podcasts from another host. Originals include shows from the BBC, History Hit, NBC News and the New York Times.

The useful development is the data. Podcast Index now publishes daily "lookalike feeds" files, pairs of live feeds that share episodes, each with a likely label and a confidence score. The Index's own page is careful to call each pair something to verify, not a verdict. Podnews says that of the top 500 lookalikes, more than 290 are on Spreaker. If you run a show of any size, someone on your side should look at your name in that file once a week. It is a CSV; it takes two minutes. The feed lock still goes on first.

Is a download a listen? Amazon says an unseen ad is not an impression

Not a podcast story, and I will be clear about that. Per PPC Land, Amazon Ads has told advertisers that from 1 January 2027 it will stop counting a display impression when the ad starts downloading and will count it only when it appears on the shopper's screen. Amazon expects impressions on those placements to be about 63% lower on average, with bigger drops further down the page. This is display advertising on Amazon's own store pages. Audio and podcasts are not mentioned.

Podnews, on 4 November, drew the parallel with podcasting's oldest argument, downloads against plays, and pointed to its own estimate that about 13% of podcast downloads are never played. That estimate is built on a 2021 Apple bug and the article says so; it is an informed guess, not a measurement.

My opinion on the parallel: the gap is much smaller in podcasting, and the direction is the same. A download counts a file that arrived, not a person who listened, and the whole industry has agreed to call it a listen for twenty years because it is the only number a host can see. Amazon just told its advertisers that a number everyone had agreed to count was 63% air. If a sponsor asks you the same question about downloads, have the honest answer ready: here are downloads, here are Spotify and YouTube plays, which are counted at the play button, and here is the gap between them. The show that volunteers the gap is the one that keeps the sponsor when a platform forces the change.

Who pays whom to put your YouTube on Spotify?

Flightcast announced on 3 November a tool that takes an existing YouTube channel, a playlist or chosen videos, and publishes them to Spotify as video, then keeps doing it for new uploads, with titles, descriptions, thumbnails and chapters adapted. Podnews notes that earning from Spotify through it is subject to Spotify's eligibility rules. It is not free: the press release prices it at $100 a month, with Flightcast's hosting included. Flightcast was co-founded by Steven Bartlett, so this week's YouTube story and this one come from the same desk.

On the same day Podnews carried an Audion and YouGov study of France: 58% of people there say they have started a piece of content on one platform and finished it on another, which the study calls liquid content, and YouTube dominates French digital audio. One master, published everywhere, with the audio good enough to carry on with the screen off, is the practical version of that finding. It is the argument we make in video podcast or audio podcast: you do not choose, you make one thing that works as both.

In brief

Apple's fourth-quarter results: per Podnews on 4 November, Services revenue was $31.6 billion, up 9.8%, and podcasts were not mentioned on the call. Six Colors called the near-10% growth the lowest in three years. Two platforms in two weeks have reported to investors without saying the word.

The Podcast Show London 2027 has tickets on sale, per Podnews on 4 November, for 19 and 20 May, with a new stage focused on video. The organiser's release, same day, puts first-release day passes from £65 until prices rise on 9 February.

Podscribe's October rankers, released 4 November: Crime Junkie reached the most people, about 16.4 million; The MeidasTouch Podcast had the most plays, 61.2 million; The Diary of a CEO was second on plays at 48.1 million; iHeartMedia's monthly plays rose 63%. Podscribe's figures are estimates, and the plays include Rumble.

Ausha, per Podnews on 4 November, switched on automatic monetisation worldwide for shows with 5,000 downloads in the past 30 days. Captivate Studio, per Podnews on 3 November, is an invite-only remote recording beta for up to seven participants. And PodcastOne, per Podnews on 4 November, lost a board member and its interim finance chief in a week, the third finance chief to leave in under two years, with filings expressing "substantial doubt" about trading beyond July 2027.

The numbers

From the edit desk: when should you stop an interview, and how do you stop it well?

The live election nights are a reminder that most of you have a luxury the LBC desk did not: you can stop. Almost nobody does. I watch the last twenty minutes of a lot of recordings, and the pattern repeats. The host felt the conversation go flat at minute 48, kept going to minute 75 because the calendar said 90, and then asked me to find the episode inside it. There is one, and it ends at minute 48.

Three signals that a recording has run past its best. The answers get shorter: a guest who was giving you two minutes is now giving you twenty seconds and a glance at the clock. The repeats start: a story you have already had comes back with different words, and the guest does not notice. The energy drops: the voice sits lower, the gestures stop, and on video you can see the shoulders go. Any one of those is a note. Two together is the end.

What to say, out loud, with the recording running:

1. To stop on a high: "Let me stop you there, that's the one I wanted to end on." Then thank them by name and stop. Do not add a question after it.
2. When the guest has nothing: "I think we have it. Is there one thing you came here to say that I have not asked?" One, by name. If the answer is no, you are done.
3. When you have lost the thread yourself: "Give me ten seconds." Breathe, look at your notes, and go again. Silence is free to cut; a panicked question is not.

Stop and re-record when the problem is technical (a camera dropped, audio clipping, a doorbell) or the guest is clearly unwell or distracted. Push on when the problem is a dull guest, because a second session will be dull as well, and a short episode beats two long ones.

Flag things for the editor as you go, with your voice. "Cut that, again" before a retake. A single clap when something has gone wrong: it is a spike on the waveform and I find it in seconds. "That's the clip" after a line you want as a short. Each marker saves me ten minutes of hunting and saves you a round of notes.

Which number will you stop quoting this week?

Pick one. The 35% on its own, now you know what it counts. Downloads as if they were listens, now Amazon has shown what happens when a platform stops pretending. Or the 90-minute runtime you keep hitting because the diary said so, when the episode was finished at 48. Whichever you choose, the edit gets easier, and if you want that edit done by someone who has made the stop-at-48 cut more than a thousand times, a full episode edit is £280. Billing is every four weeks, there is no minimum, no VAT is added, and we revise until you are happy. Send us the show and tell us where it went flat. Back next Thursday.

If you'd like a hand with your own show, here's our full episode edit and what it costs.