Made On YouTube 2026: what a podcast running clips should plan for

YouTube published its Made On YouTube 2026 announcements on 23 September, and one example prompt in the post was about podcasts.

"Witty podcasts for my daily 30-minute commute."

That is a viewer typing a sentence and getting a home feed built from it. YouTube calls these custom feeds, and says a viewer can save several and they refresh themselves.

I read the full post and the Podnews write-up from 25 September on 28 September, because at VALORAE Arc every campaign ends up on YouTube somewhere. Arc has delivered 500+ videos, and a change to how a viewer asks for a show changes how a clip should be labelled.

There is a problem with all of it, though.

Nothing in the post has a date. The language is "soon" and "coming". So this is a planning note, and I will say plainly which parts are worth acting on before anyone can use them.

Change 1: viewers will describe what they want in a sentence

Custom feeds turn the home page into something the viewer writes.

Today the home feed is built from what you watched. The announcement describes a feed built from what you say, like "evening wind down with long documentaries" or the commute prompt above.

Nobody outside YouTube knows how a prompt gets matched to a channel. The post does not say, and I am not going to guess at a ranking signal.

What I can say is the cheap part. If a person has to describe your show in a sentence to find it, your title, description and channel page should already contain that sentence.

Most podcast channels do not.

They are named after the host, the episode titles are guest names, and nothing on the page says "a founder interview show about hiring" in words a person would type. A clip titled with a guest's name is invisible to a viewer who does not know the guest.

So write the description of the show the way a listener would ask for it. That helps search today and costs nothing if custom feeds arrive late.

Change 2: Shorts can be grouped into seasons and episodes

The post says Shorts can now be organised into dedicated seasons and episodes, "making it easier to watch microdramas and series."

It says microdramas. It does not say podcasts.

But a podcast clips channel that runs a recurring format already is a series. "The hiring question" every Tuesday. "One founder mistake" from every guest. Those are episodes of something, and YouTube is building a shelf for exactly that shape.

Series beat random clips.

That holds independent of this feature. A viewer who liked one clip from a recurring format has a reason to find the next one, and a viewer who liked a random highlight has no thread to pull.

So the work is to decide your two or three recurring formats now, name them consistently, and cut to them. If the seasons feature reaches your channel, you drop the clips in. If it never does, you still have a channel a viewer can follow.

Change 3: three cuts, tested against each other

YouTube Studio, per the announcement, will allow video A/B testing "with up to three different cuts."

That is a bigger deal than thumbnail tests. A cut is a different edit, which means a different opening.

The post does not say whether Shorts are included, and I would not assume they are. For long-form episodes, though, it means you can finally test the first thirty seconds instead of arguing about them.

Here is the complication. Three cuts means three edits, and edits are the expensive part. A test only helps when the cuts differ in one decision you care about: cold open on the best line versus the host's intro, or a guest's claim first versus the question first.

Test one decision at a time. Three cuts that differ in everything tell you nothing.

Packaging still starts with the thumbnail and title, and VALORAE Media's guide to thumbnails that get clicked covers the part a cut test will not.

Change 4: an editor that suggests trims

The announcement describes a conversational editing tool for Shorts and the YouTube Create app that "suggests trims, and reorders content for better pacing."

Trims are useful. Selection is the job.

An assistant that tightens a clip you already chose will save minutes. It has no idea which of the ninety minutes was the one worth clipping, because that depends on who the show is for and what the guest has said elsewhere. We made the same argument about third-party AI clippers in what AI tools do well and where they stop, and a built-in tool changes the convenience, not the argument.

So use it for pacing. Keep the choice of moment with a person who has watched the episode.

Change 5: likeness detection is coming to mobile

This one is for the founder whose face is the show.

The post says likeness detection is coming to mobile, with enrolment, match alerts and actions from the phone, and that speaking voice detection will be combined with facial detection for better accuracy.

If you clip a founder, a fake clip of that founder is a real risk. A cut that puts words in their mouth, or an AI voice over their face. Detection that runs from a phone lowers the bar for the person most likely to spot it, which is the founder.

When it reaches your account, enrol. It costs nothing, and at VALORAE Arc we would rather the person being clipped is the first to hear about a fake, before a client does.

Change 6: the podcast feature most shows cannot plan for

Podnews reported on 25 September that YouTube Music Premium subscribers will get Your Podcast Lineup, a weekly spoken AI preview of recommended podcasts that then plays suggested episodes hands-free.

It is subscriber-only, it has no date, and there is no published rule for how a show gets picked.

So I would not plan around it. Keep an eye on it, and if it launches, check whether your show appears for the listeners you would expect. That is the extent of what anyone can honestly do today.

The short version

• Made On YouTube 2026 was published on 23 September. Nothing in it has a launch date yet.
• Rewrite your show description the way a listener would ask for it, before custom feeds arrive.
• Pick two or three recurring clip formats and name them consistently. Series beat random clips.
• When three-way cut tests arrive, change one decision per test.
• Let the built-in editor trim. Keep moment selection with a person.
• Enrol the founder in likeness detection the day it reaches the account.

Clips cut from your own footage and posted on a schedule is what we run at valoraearc.com.