Last month I pulled the numbers on a show we cut every week for a year. Same episodes, same upload day, same thumbnails, published on both platforms. YouTube brought in 87 percent of the new listeners. Spotify brought in almost none, but it kept nearly everyone who was already there. That one line explains most of what people get wrong about this question.
Hosts ask me "YouTube or Spotify?" like it is a fork in the road. It is not. It is a question about which job you are hiring each platform to do. One of them finds you strangers. The other one keeps the people who already like you. If you treat both as growth channels you will be disappointed by one of them and you will blame the wrong thing.
So let me be blunt about my position: YouTube is your discovery engine and Spotify is your retention shelf. Publish to both, but only design for one.
YouTube is a search and recommendation machine that happens to allow audio
YouTube does not care that you made a podcast. It cares whether the next viewer stays. That means everything that works on YouTube generally works on YouTube podcasts too: a title that makes a promise, a thumbnail that makes you curious, an opening thirty seconds that pays off the promise instead of introducing yourself for two minutes.
The practical consequence is that YouTube rewards editing. Real editing. Cutting the flat middle, tightening the answer that took the guest four tries to get right, cutting to the reaction on the second camera exactly when the joke lands. This is why we watch every episode end to end instead of trusting silence detection. Silence detection finds gaps. It does not find the moment where a guest says something honest and then immediately backs away from it, which is usually the best twenty seconds in the file.
It also means packaging is not decoration, it is half the job. I have watched two identical episodes perform four times apart because one had a title that named the tension and one had a title that named the guest. If your thumbnails are an afterthought, our sister studio ClickTheory does eight of them for 250 dollars, and I would rather you spend that than spend it on another microphone.
Spotify is where your existing audience listens, and that is not a small thing
Here is where people overcorrect. They read a post like this one, decide Spotify is dead for growth, and stop caring about it. That is a mistake for a different reason.
Spotify is where the person who already subscribed listens on Tuesday morning in the car. They are not browsing. They are not discovering. They opened the app because they wanted your show specifically. The metric that matters there is not impressions, it is whether they finish. A show with a loyal Spotify audience has a business. A show with big YouTube numbers and no repeat listening has a hobby with good analytics.
So the Spotify job is different. You want clean audio, chapters so people can skip the parts they do not want, a consistent run time so it fits into a commute, and an episode structure that makes sense with your eyes closed. Video on Spotify exists and is fine. It is not why anyone is there.
If YouTube is the storefront window, Spotify is the regulars sitting at the bar. You do not decorate the bar to attract people from the street.
What actually changes in the edit
This is the part nobody explains, so here is the honest breakdown of what we do differently depending on which platform a show leans into.
For a YouTube-first show:
- Cold open. Pull the best forty seconds from anywhere in the episode and put it at the front, before any branding.
- Aggressive first five minutes. If the conversation takes eight minutes to warm up, the audience will not wait for it.
- Visual variety. Camera changes carry attention that the words alone will not. This is the entire argument for why three cameras beat two once a show is past its first season.
- Chapters, because YouTube surfaces them and people genuinely use them to find the part they came for.
For a Spotify-first show:
- Keep the natural arc. The cold open that helps on YouTube can feel disorienting in audio only.
- Protect the pacing over the run time. Audio listeners forgive a slow moment far more than a rushed one.
- Tighten cross talk and mouth noise harder than you would for video, because there is nothing to look at while it happens.
- Write show notes that work as an actual document, not as a keyword dump.
We are not making two edits. We are making one edit with a bias, and that bias is a decision you should make on purpose rather than by accident. Most hosts I talk to have never actually made it.
The clips question, which is really the whole distribution question
Every growth conversation ends up at clips. Fine, let us go there.
Clips do not grow your podcast in a straight line. They grow awareness of you, and some percentage of that awareness eventually turns into someone typing your show name into a search bar. That gap is why hosts get frustrated. They post twenty clips, get a million views, and see forty new subscribers. That is not a broken funnel, that is the normal one.
What changes the math is picking clips that make someone want the full conversation instead of clips that resolve completely on their own. A clip that answers the question fully is a good clip and a bad advertisement. I wrote about a case where the clip outperformed the episode by a factor I still find annoying, and the lesson was not "make more clips", it was "make clips that leave something on the table".
We do vertical clips at 25 dollars each as an add-on, and my usual advice is three per episode, not ten. Ten clips from one conversation means seven of them are filler and filler teaches an algorithm that your account is average.
Should you run a separate channel for the video?
This comes up constantly, usually from people whose main channel is doing something other than podcasting. My answer is almost always no, not yet.
A second channel splits your signal in half at exactly the moment you need it concentrated. The exceptions are real but narrow: a genuinely different audience, a genuinely different format, or a main channel where the podcast uploads are visibly dragging the recommendations down. I laid out the specific thresholds in when to start a second channel, and the short version is that it is a problem you earn, not a problem you plan for.
If you are under fifty episodes, one channel. Put the effort into making the episodes worth returning to instead of into org charts for your content empire.
The thing both platforms actually reward
After a few hundred episodes I have stopped believing in platform tricks. The two platforms want different things on the surface and the same thing underneath: an episode people finish and then come back for.
That is a structural problem, not a distribution problem. It is about whether the conversation has a shape, whether the middle earns its length, whether there is a reason to press play on the next one. I broke down what that looks like in the anatomy of a rewatchable episode, and none of it is platform specific.
If your episode is good and your packaging is bad, YouTube will punish you and Spotify will not notice. If your episode is bad and your packaging is great, YouTube will give you one big week and then stop. Only the boring answer survives both: make the episode better.
What I would do if I were starting today
Here is the plan I give hosts when they ask, in order.
- Publish full episodes to both. Never make people choose a platform to hear you.
- Design for YouTube. Title, thumbnail, cold open, first five minutes. That is where strangers arrive.
- Keep Spotify clean and consistent. Chapters, good audio, predictable release day.
- Three vertical clips per episode, chosen to provoke rather than to conclude.
- Do this for twenty episodes before you judge anything. Twelve weeks of data is noise.
The production side of this is less complicated than it sounds. A single camera show cut properly beats a four camera show cut lazily, every time. If you shoot with one camera we cover that in single-camera shows, and if you are running multiple angles the Studio tier covers 2 to 4 cameras in 4K with unlimited revisions at 299 dollars an episode. First cut lands in 48 hours either way.
And if you are in a field where trust matters more than reach, the platform split shifts again. A show built to make one specific kind of client comfortable does not need a million views, which is the whole logic behind how we approach podcast editing for insurance agents. Sixty of the right listeners beats sixty thousand of the wrong ones, and I say that as someone who likes big numbers.
Pick the job first. The platform question answers itself after that. If you want a second opinion on where your specific show should lean, send me a link at samy@pixelsamy.com and I will tell you what I would cut.