This Creator Got Shadowbanned on YouTube in 25

So you've heard of people trying to go viral on YouTube to prove it's not luck, right? Those "I started from zero and got monetized in 30 days" videos that flood your feed? A creator from vidIQ just did the exact opposite—and honestly, the experiment reveals more about how platforms actually work than any viral success story.

The team at vidIQ set out to answer a question that keeps YouTube's official stance and creator experience in permanent tension: Does shadowbanning actually exist? YouTube says no. Creators say yes, constantly. Someone needed to test it.

Enter "It's AI Dog Life," a channel specifically designed to fail.

The Recipe for Disaster

The setup was almost too simple. Using AI generation tools, the vidIQ team pumped out 30+ shorts featuring the same Pixar-style animated dogs doing mundane activities. Dogs playing poker. Dogs stuck in traffic. Dogs walking other dogs. Each video took minutes to generate with basic prompts, and the entire channel required maybe an hour of actual human effort.

"I don't think I've ever made a video faster than that in my entire life," the creator noted. "But to be honest, I'm not sure how I should feel about that as a creator on YouTube."

The content strategy? Consistent mediocrity. Same dogs, same format, one video per day, generic hashtags. The kind of templated, scalable content that platforms increasingly need to deal with as AI generation becomes frictionless.

The Algorithm's Three-Act Structure

What happened next reveals how YouTube's recommendation system actually evaluates new content—and it's more sophisticated than the binary "good content gets promoted, bad content doesn't" narrative.

Act One: The Testing Phase

YouTube gave the channel a genuine shot. The first video hit 2,500 views. Over three weeks, the channel accumulated 20,000 views total. People even subscribed. For AI-generated slop posted to a brand new channel, this seems... generous?

Turns out YouTube Shorts operate differently from regular videos. Since there's no thumbnail to click, the algorithm uses what the team calls a "seed audience"—1,500 to 2,500 initial viewers who essentially beta-test each short. This is why you see that characteristic spike in the first 24 hours.

But here's where the surface metrics lie. Yes, 2,500 people "viewed" that first short. But only 500 were "engaged views"—meaning they actually watched instead of just swiping past. The stay-to-watch percentage? A dismal 25%.

For context, successful shorts channels see stay-to-watch rates above 80%. The main vidIQ channel hits 70%. At 25%, three out of every four people were actively rejecting the content.

Act Two: The Suspicious Stage

YouTube's algorithm noticed. By video 17, something shifted. Instead of the typical few hundred views, it got 16. Then five views in the first 24 hours—"485 views less than typical, or just 1% of the views you would typically expect," as the creator put it.

A few videos later bounced back to normal numbers, creating this eerie pattern where the algorithm seemed to be... reconsidering? Testing whether this was a temporary quality dip or a fundamental problem?

Act Three: The Shadowban

From video 26 onwards, the channel flatlined completely. Six views. Seven views. Three views. One view. Some videos got literally zero views. The creator noted that many of those handful of views probably came from them personally checking the analytics.

The channel had achieved the goal: demonstrating what happens when you consistently upload content that audiences consistently reject.

What This Actually Reveals

The experiment exposes a tension at the heart of content moderation that platforms rarely acknowledge openly. YouTube officially doesn't "shadowban"—but something definitely suppressed this channel's reach, and that something was responsive to audience behavior.

"I know shadow banning, or shadow ban, is a filthy word at YouTube headquarters," the creator acknowledged. "But this test does seem to suggest that there is some sort of trigger, some sort of switch that suppresses the reach of content if the audience has already decided it's very unsatisfactory."

The mechanism seems to work like this: YouTube gives new content a fair trial with a seed audience. If that audience consistently rejects it (low engaged views, high swipe-away rate), the algorithm doesn't just reduce promotion—it essentially stops testing. With 20 million videos uploaded daily, the platform can't waste distribution on content that's already failed its audition.

What vidIQ calls "inauthentic content"—templated, AI-generated, easily replicable at scale—triggers this response efficiently. The channel went from 20,000 views to zero in 25 days.

The Question Nobody's Asking

Here's what makes this experiment actually interesting: Do we want the algorithm to do this?

If you're a creator who got caught in what feels like a shadowban, the answer is obviously no. But if you're a viewer scrolling through Shorts, being protected from the 47th variation of AI dogs doing human activities... maybe yes?

The platform has to solve for scale. As AI generation becomes easier, the volume of low-effort content will only increase. An algorithm that can't distinguish between "this creator is having an off week" and "this is manufactured spam" will either drown users in garbage or overcorrect and hurt legitimate creators.

The vidIQ experiment suggests YouTube's current approach leans heavily on audience signals. Not views—engaged views. Not retention—swipe-away rates. The algorithm is essentially asking: "Do humans actually want to watch this?"

And when the answer is consistently no, the content stops getting distribution. Whether you call that a shadowban or "algorithmic suppression based on quality signals" probably depends on whose side you're on.

What we know for sure: There's a threshold. Cross it consistently enough, and your reach doesn't just decline—it disappears. The experiment proved it in 25 days with 25 videos.

The real question is whether that threshold catches the right content, or whether legitimate creators get caught in the same trap designed for AI slop. That's the experiment nobody's running yet.

—Zara Chen, Tech & Politics Correspondent