A brand came to us recently with an ad they’d made for YouTube. Fully AI generated, start to finish. Script, voice, visuals, all of it. It cost them almost nothing and it looked… fine. But results were less than fine, so they wanted to know what we thought went wrong.

Technically, nothing. Practically, everything.

The issue was that they’d made an ad that talked about community and togetherness while looking like AI slop. On a platform that runs on people, not on ads. I said something on that call I’ve been saying more and more to both creators and brands lately: to connect on YouTube, you have to be a YouTuber. You don’t get to show up with a handful of slop and expect the room to care.

Different conversation, same week. A review-focused channel, well researched, legitimately useful videos, flattening out. We went looking for the format problem and eventually landed somewhere less comfortable: a viewer can get that exact information out of Claude in about nine seconds. The information was never the reason to watch, the host was. But the videos were built like corporate explainers, so the host wasn’t clicking with the audience.

I had four conversations over the last two weeks about roughly the same concept. So here’s where I’ve landed:

The flood of cheap, competent content is doing your differentiation for you. Let it.

The clearest signal I’ve seen came from our creative partner Matt Wolfe, who lives deep in the AI space. Matt knows the tools better than anyone. His plan for this year isn’t more output. It’s fewer videos, rawer, documentary-style, bigger swings. Slower and more expensive on purpose.

The guy with the best access to the shortcut is deliberately walking past it. How ‘bout that?

(Also, a quick round of applause for Matt who is hitting 1M subscribers in the next week or so!)

For most of YouTube’s recent history, production quality was the moat. Cameras, editors, motion graphics, the time to use all three. Thanks to AI, that’s all changed. And when a constraint disappears, the value doesn’t disappear with it. It moves to whatever’s still scarce.

What’s still scarce is a person worth spending twenty minutes with.

Which is why the slop is such an opportunity rather than a threat. Our feeds are currently filling up with polished, cheap, forgettable videos. The bar for looking professional is heading to zero. The bar for being worth watching is going up. In fact, I would argue that audiences are actively pushing back on some higher-production videos in favor of more lo-fi offerings that have higher authenticity and relatability.

I know “the human part is the valuable part” is an extremely convenient thing for me to believe. But we use AI all over our own workflow. Transcripts, research, first-pass concepts. I’m still not making an anti-AI argument.

The line I’d draw is back office versus on camera. Use the machines for everything nobody sees. Nobody has ever subscribed to a workflow. Except productivity content…

Here’s the part I think most channels won’t do anything with, unfortunately. It asks for the opposite of what the last five years of advice trained everyone to do. Don’t make more videos. Don’t aim for a faster turnaround. Don’t add another tool to the stack. Be more present in your own content - something that is harder, slower, and a lot more exposing than generating twelve assets before lunch.

So, a question about your last upload: If you strip out every fact a viewer could have gotten somewhere else in under a minute, what’s left?

If the answer is “the video,” good. If it’s “not much,” you didn’t make a video. You made a search result with a face on it.

Go get ‘em.

Share

Leave a comment

I gave three different clients the same piece of advice this week, almost word for word. When that happens, it usually means we’re on to something.

The advice? Stop posting so many shorts.

With one, I asked them to unlist the shorts already on the channel and pause new ones entirely while we build out the long-form strategy. Another had brand deals lined up specifically for short-form and wanted to keep feeding it. A third was posting a short a day, because that’s what everybody says to do.

All three were doing the thing that feels like free reach, but is a slow-acting tax on the channel they actually want to build.

Here’s the epiphany:

YouTube decides what kind of channel you are, and then serves your videos to that channel’s audience.

It decides based on what you upload. Post mostly shorts, and you’re a shorts channel now. Congratulations.

Then you drop a 14-minute video, and YouTube does exactly what you’d expect: it serves it to the audience it thinks you have. Which is a shorts audience. People in a feed, thumb hovering, in a completely different mode. They are not sitting down for your deep dive. They swipe.

YouTube reads that swipe as “this video didn’t work” and pulls back distribution. It’s not personal. It’s a simple audience mismatch.

And it runs the same way in the other direction. Serve short-form to a long-form audience and it doesn’t land either. Someone who subscribed for your 20-minute breakdowns isn’t scrolling for a 30-second clip of you. They came for the sit-down version.

Two audiences that behave nothing alike. Give either one the wrong format and you get penalized twice. Once when they don’t watch, and again when the algorithm quietly updates its idea of who you are.

So this isn’t an anti-shorts position. It’s an audience problem.

If your priority is long-form, you should treat shorts as a supporting act. Two or three per week should do the trick as long as it’s tied to a long-form upload. One a day runs the risk of redefining your channel.

But there’s another key detail: make them for the shorts audience. A horizontal clip lifted out of your video, letterboxed into a vertical frame, is the worst of both worlds (unless you’re a podcast). It underperforms as a short and it muddies the signal on your main feed. I know this because we’ve tested it and seen consistent results across multiple channels.

I believe the clip channel is still a viable option. I bet you can easily name a half-dozen creators running a secondary shorts channel. But notice when they started - after they were already enormous with enough demand that a second channel quickly took off. Before that, a clip channel is just you funneling your efforts away from your primary goal.

A word on sponsor pressure, because short-form deal flow has ramped up noticeably this summer. I’m not going to tell you to leave money on the table. Take the deals. Just understand the risks and don’t let the deliverables quietly reshape your upload pattern into something you never chose.

If you want to know where you actually stand, this takes about two minutes. Open YouTube Studio, set it to the last 90 days, and look at where your new subscribers came from: shorts or long-form.

If most of them came from shorts, there’s your answer. That’s the audience YouTube is handing your next 15-minute video to, and that’s why it’s landing softer than it should.

Then look at the ratio. How many shorts went up against how many long-form videos? If you can’t say out loud what that ratio was supposed to accomplish, the algorithm can’t either.

None of this makes shorts the enemy. It just means you have to pick which audience you’re building, and then upload like you mean it.

Can I also just add that this is kinda depressing? This entire exercise makes me realize that we’re straying further away from what made YouTube beautiful to begin with: people uploading great content and easily reaching an audience organically - not having to worry about gaming an algorithm.

It’s really hard to keep up with what YouTube wants, and it’s even harder to change what you’re doing once you figure out what that is.

But that’s for another time.

Go get ’em.

Share

Leave a comment

A creator we work with had a video pop about a month ago. Roughly five times their usual numbers. It should be a time to celebrate, right?

Here’s what actually happened next: the three videos that followed felt like failures. They weren’t. They performed right at the channel’s baseline, maybe a touch above. But nobody was comparing them to the baseline anymore. The creator, the comments, and (most importantly) the manager/sponsors were comparing them to the spike.

That’s the trap. The viral number doesn’t just live in your analytics. It moves into your head and starts paying rent.

Here’s the message I’m consistently driving home:

I’d rather see you do four solid videos a month than one monster hit surrounded by flops.

Say your baseline is 30K views. Four videos at 30K and one video at 120K are the same total. Same watch time, roughly. But they are not remotely the same channel.

The four-30K channel is building return viewers, which is a highly underrated stat on the platform. It’s giving YouTube a reliable signal that this channel delivers, every time, to a predictable audience. The algorithm loves a safe bet. It distributes safe bets.

The one-120K channel taught the algorithm nothing except that one video, one time, connected with a bunch of strangers who mostly never came back. And it taught the creator something worse: that 120K is the new bar. Every future video gets measured against an outlier, and outliers, by definition, don’t repeat on command. Nailing that number gets harder every single time you try.

So the greenlighting process quietly changes. Instead of “what’s the best version of the thing my audience shows up for,” it becomes “what might pop like that one did.” You start going for it every time, but mostly strike out, which leads to an audience that no longer recognizes your channel’s value proposition.

I’m not anti-viral. If a video takes off, wonderful, take the W and try to figure out why. Sometimes a spike is genuinely the start of a new level, but I would caution you to avoid declaring a new baseline off of one data point.

Compounding math is the whole game. Consistent output at a repeatable level means more videos for the algorithm to test, more chances to convert a casual viewer into a regular, more reps improving the format itself. The spike gives you one great screenshot. The consistency gives you a business.

I know my production-company bias is showing here… of course the guy who builds repeatable pipelines thinks repeatable is the answer. But I came to this belief the honest way: by chasing outliers for a year and ending up smaller than when I started.

One thing worth doing this week: pull up your stats and cover the outlier - that’s your channel. That’s the number your decisions should be built on. If you can raise that floor by even 20%, you’ve done something meaningful and durable. The spike will come back around eventually. And when it does, it’ll be landing on a much higher floor.

Go get ‘em.

Share

Leave a comment

Three times in the past week and a half, on three completely different channels, the same thing happened. The less-produced thumbnail won.

Different niches, wildly different creators, no connection between them. The results were identical. The slick, highly produced thumbnail (two of them used AI elements) underperformed the more lo-fi version. One channel caught a decent amount of flack in the comments because of the use of AI. The lo-fi versions felt like they could have belonged on YouTube a decade ago, which is part of a thumbnail simplification trend we’re noticing platform-wide.

When the same signal shows up in three places in a short amount of time, it stops being a coincidence. So here’s the thinking: the “just use AI for thumbnails, it’s faster and cheaper” advice is starting to cost creators clicks.

Here’s why I’m not surprised, and apologies if my bias is showing…The whole pitch for AI thumbnails is speed. Ten options in the time it takes to brief a solid designer on one. While that might be true to a degree, speed is never the thing standing between us and a great thumbnail. It’s about connecting with the audience emotionally, and these days, we’re seeing that the use of AI in creative spaces is having an adverse effect.

A thumbnail has one job: win a split-second decision in a crowded feed, against ten other thumbnails, half of them on your exact topic. AI is very good at producing something that looks polished and very bad at producing something that looks chosen. It defaults to the average of everything it’s ever seen - generic, which is the one thing you can’t afford to be in a sea of competing content.

Add to that, the fact that people are looking for it, and you’re going to face push back when it’s spotted. You know exactly the kind of thumbnail I’m talking about, without naming names of course (beyond the thumbnails I shared above of course). The over-rendered, slightly-too-glossy AI look is becoming a tell, the same way a stock photo on a landing page screams inauthentic.

For any creator whose entire value is that a real person is behind the channel, looking synthetic is hurting the authority you’re working so hard to build.

I want to be careful here, because this isn’t a plea to stop using AI. We use AI all over our workflow, including to brainstorm thumbnail concepts before anyone opens Photoshop. The tool isn’t the problem. Shipping the tool’s first draft as your final answer is the problem. AI is a fantastic way to get to a starting point fast. It’s a terrible substitute for the human intuition to know what will resonate and make your audience stop scrolling.

So before your next upload, look at the thumbnail and ask, could this have come from anyone, or could it only have come from you? If the answer is anyone, then you made it faster, but you didn’t make it click.

Polish is cheap now. Everyone has it. The scarce thing is a thumbnail that looks like a human being made a decision.

Go get ‘em.

Leave a comment

We spent a while trying to fix the videos before we admitted the obvious: the videos were never the problem. The category was.

When everyone is doing the same thing, you’re not competing on quality. You’re competing on subscriber count. And you’ll lose to whoever’s bigger.

YouTube doesn’t really hand your video to your subscribers; it drops it into browse and suggested, next to a sea of other thumbnails, some on the exact same topic. When a viewer is choosing between five near-identical options, they don’t click the highest-quality one. They can’t see quality from the thumbnail. They click the channel they recognize, or the one with the bigger view count, because views and subs read as proof it’s worth their time when their time is already at a premium.

Which is why “just make better videos” is such a trap. Packaging aside - you can double the effort, sharpen the script, upgrade the graphics, and move the needle approximately nowhere because execution was never what held you back. The video was replaceable. A better version of a replaceable video is still replaceable.

The way out isn’t a better video. It’s a video only you could have made.

Same lesson, different delivery. Instead of “how to handle a pricing change” (a video 200 channels could shoot this afternoon) you tell the story of the pricing change you actually ran, the one that scared you and worked anyway. Same takeaway for the viewer. But now the lesson is welded to your specific experience, and nobody can clone it, because nobody else lived it.

The developer I mentioned is a good example. The pivot wasn’t “better dev tutorials.” It was talking-head videos about the real problems he hit building his own business. War stories that happen to teach the same things a tutorial would. Except no one without his exact résumé could make them. One of the titles we kicked around was “I lost 90% of my users — and why that’s a good thing.” A tutorial channel could cover that pricing lesson in the abstract. None of them could put their name on that sentence.

That’s the whole difference: educational content versus personal-narrative-as-educational. The first is a commodity, and commodities compete on size. The second is a moat. The lesson can be word-for-word identical. What changes is whether the video is interchangeable, and interchangeable is the thing quietly killing the YouTube middle class.

So run one quick check before your next upload. Name three things in the video that only you could have said: a specific number from your own work, a real decision you made, a mistake with your fingerprints on it. Three is the bar. If you can’t get there, you’ve made a video 200 other people could’ve made, and in that race, the only variable that matters is who showed up bigger.

You don’t want to win the size contest. You want to opt out of it.

Go get ‘em.

Leave a comment

This is the first newsletter I’ve ever written. So, I’ll take the opportunity to say welcome to the Hooked Newsletter and thanks for joining me on this brisk walk through the trenches of YouTube.

I had a call last week with a creator who’s been crashing out for a month. Mid-six-figure subscriber count, nearly a million views per video, unique style of creativity. He was convinced his content had gone stale based on feedback he was getting from the comments section. According to them, his topics were stale and he was burning his audience who were tired of hearing the same old story and themes.

Then we looked at the stats.

Eighty-plus percent of viewers on every recent video were brand new to the channel. Less than one percent were returning regulars. Fewer than 6% were actually subscribed to the channel. The thing he’d been losing sleep over was a problem to only a handful of people.

I see versions of this every week, so it’s worth saying out loud:

Your comments section is not your audience.

The people who comment are an exotic and vocal minority - they are, almost by definition, the least representative slice of the people you’re actually reaching. They are loud, and sometimes… not so nice.

Fun fact: in my 16+ years of working in YouTube, I’ve never left a comment. Ever.

YouTube doesn’t really distribute your videos to your subscribers. It distributes them to samples of new people, mostly through suggested and browse. A new test. Every time you upload. Your video shows up next to ten other thumbnails on someone’s feed. They click because the packaging worked, or they don’t. That’s the game.

Which means the creative anxiety I hear often is usually answering the wrong question - am I repeating myself, do they have a point, should I change everything based on a few complaints? The right question is whether the next sample of strangers will click and stay. They’ve never heard the bit before. They don’t know it’s your fourth video on the topic this year. They don’t care.

I’m not saying you should phone it in. Repetition for the sake of it is still bad content. But there’s a difference between making the same video twice and having a format you should keep running until the data says otherwise. Most creators who feel stuck are doing the second thing and beating themselves up like it’s the first.

A couple of things worth checking before you lay waste to a format you’re tired of:

Look at the actual returning-viewer percentage on your last ten videos. If it’s under 10%, the format fatigue story is mostly a story you’re telling yourself. Your regulars are barely the audience.

Read your top comments and your retention graph as two separate signals. The comments tell you what your most engaged 1% feels. The retention graph tells you what the other 99% did. When those two disagree the graph is the one that keeps the lights on.

And if you’re genuinely creatively done with a format, that’s a separate, real conversation worth having. But don’t let “my audience is sick of this” be the reason. Nine times out of ten, your audience just got here.

The hardest part of running a channel at scale is that the loudest feedback is usually the least useful, and the most valuable feedback is a chart on a screen that doesn’t talk back. The key is learning to weight those correctly. The risk of following the comments is rebuilding something that was working just fine.

Go get ‘em.

Leave a comment

Reach out anytime to talk about YouTube!

Visit our website to learn more.