Hooked Creators

Your AI thumbnails might be killing your CTR

Dave Heuff — Wed, 01 Jul 2026 12:01:53 GMT

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.

The undisputed champs of lo-fi thumbnails

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.

If 200 channels could've made your video, you already lost

Dave Heuff — Thu, 18 Jun 2026 16:06:06 GMT

Four videos in a row underperformed for a creator we work with. And not because they were bad. They were objectively (and a bit subjectively) good. Clean edits, solid value offering, the kind of tutorials that would’ve put up strong numbers a few years ago. They got crushed by bigger channels making the exact same videos.

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.

Your comments section is lying to you

Dave Heuff — Wed, 10 Jun 2026 12:01:16 GMT

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.