---
title: 'YouTube Algorithm Changed Forever: 4 Updates You Missed'
source: 'https://youtube.com/watch?v=t2y9pxGggdQ'
video_id: 't2y9pxGggdQ'
date: 2026-07-12
duration_sec: 482
---

# YouTube Algorithm Changed Forever: 4 Updates You Missed

> Source: [YouTube Algorithm Changed Forever: 4 Updates You Missed](https://youtube.com/watch?v=t2y9pxGggdQ)

## Summary

This video reveals four major YouTube algorithm updates driven by Google's Gemini AI, shifting focus from retention to viewer satisfaction, semantic understanding, and cross-platform data. The creator, Romero, claims these changes are unprecedented and explains how they affect video recommendations and channel growth.

### Key Points

- **Unique Insight** [00:00] — The information in this video cannot be found elsewhere because other YouTube coaches focus on titles and descriptions, missing the real update that happened on Google, not YouTube.
- **Gemini and Semantics ID** [00:55] — Google's Gemini AI now analyzes YouTube videos and assigns a semantic ID that captures the energy, vibe, and intent of the video, using cross-platform data from Google Drive, Gmail, and search to predict user needs.
- **Retention No Longer King** [03:18] — Good abandonment means viewers leave after getting their answer quickly, which is positive. Bad abandonment occurs when viewers leave unsatisfied and search for another tutorial. Satisfaction now matters more than watch time.
- **Co-visitation** [04:58] — YouTube now looks at who else your viewers watch in the last 31 days. If your audience overlaps with big creators like Mr. Beast, your videos get recommended to similar audiences. Sharing links to family or buying bots can harm your channel.
- **Negative Comments** [06:38] — The algorithm now penalizes videos with negative comments (arguing, complaining). A boring video with happy comments can outperform an entertaining video with bad comments. Rage bait is no longer effective.

### Conclusion

The YouTube algorithm has fundamentally changed due to Google's Gemini AI, prioritizing viewer satisfaction, semantic understanding, and audience behavior over traditional metrics like retention and tags. Creators must adapt by focusing on satisfying viewers and fostering positive engagement.

## Transcript

going to see in our lifetime. So, the information you're going to learn in this video, you cannot find anywhere else, okay? And that's because every other YouTube coach is telling you that this update has to do with titles and
descriptions or how long your video is. It's all [ __ ] They were all looking in the wrong place. This update was so big that it wasn't done on YouTube. The actual update that changed the YouTube algorithm forever was done on Google.
Now, the only reason I know this is because I went through Google developer algorithm changed. But first, if you don't know who I am, my name is Romero. I've been running faceless YouTube channels since 2018. And across all of
my YouTube channels, I have made over $2 million in ad revenue. Um, and I do other things, but I don't think you give a [ __ ] [laughter] So, let's just get started. Update number one, that is Gemini and Semantics ID. For those of
you that don't know, Google is Gemini's AI system. It's like their own version of Chad GPT. And what it does is that it analyzes all of the YouTube videos that are posted and it makes a summary about them. But on January 14th of 2026,
changed the settings on Gemini. And what this change did is that it gave Gemini the ability to reason across platforms. Meaning that now Gemini knows what you're doing on Google Drive, what you're sending on Gmail, and what you're
searching for on Google. And the purpose of that is because Google wants to know how you're feeling. They want to know what you do and what you're trying to do. Because in my opinion, what Google is trying to do is that they're trying
to go from a search engine to an answer engine. And they're using Gemini to do What does that have to do with my YouTube channel? Well, because of this new ability Gemini has, it now gives every single YouTube video something
called a semantic ID. And what a semantic ID is is essentially a summary of your video, but now it understands the energy of your video, right? The vibe, what the intent [music] was. So, let me give you an example. Let's say
pull out your phone, and you search code trucks near me. And then you search how stressed. You're in a situation and you're looking for answers fast. Now,
let's say that when you're sitting on the freeway, you open up YouTube. Well, guess what? The first video you're going to see on your homepage is a tutorial on here's the craziest part. That video tutorial that you would see on your
homepage about how to change a tire, it wouldn't be a 10-minute video. It would be a 30 second to [music] three minute video because again Google told Gemini that you were in a situation that you were stressed and Gemini controls the
YouTube algorithm. So when you open up YouTube, it used common sense and tried to predict what you were looking for. And not only that, but it also knows wouldn't show you a 10-minute video. It wouldn't go based on which uh tutorial
the most views. it would be the shortest, straight to the point video. So that's what a semantic ID is. It's what Gemini uses to predict what a user wants to watch based on their Google activity. Moving on to update number
two, and that's retention is no longer king. Now, if you've watched any YouTube tutorial ever, you know that retention is the most important part where [music] how it works anymore. And that's because on January 14th, Google also implemented
something called good abandonment. And essentially what this is is that they want to make sure that the user is satisfied. So here's an example. Let's say you make a 10-minute tutorial video on how to change a tire. And the watch
time of that video or the average view duration is 2 minutes. Meaning that after 2 minutes the majority of the viewers click off. You know, before that assume that your video was just [ __ ] But now Gemini investigates. It tries to
find out why people left after 2 minutes. And let's say with uh this tutorial, the first 2 minutes of your video was just you explaining it and that and it's like okay people didn't click off after 2 minutes because the
video was bad. They clicked off because they got their answer. That would be what Google considers good abandonment. But bad abandonment is when Gemini goes and investigates same scenario. It saw that the majority of the users would go
back on YouTube and searched for another tutorial on how to change a tire. It would then reason and think, okay, clearly this video did not answer their question. It left the viewer unsatisfied, so they had to go and find
their answer somewhere else. That is bad abandonment because now Google sees what viewers do after they watch your video. Now, it's not necessarily how long people watched for, it's about whether they were satisfied or not. Moving on to
update number three, that is co-visisitation. And if you're a small YouTube channel looking to grow, this is probably the most important one. YouTube performance overall or or your content. It also looks at who else your viewers
the last 31 days of a user's activity and they're trying to see how the viewer ended up on your channel. for example. So, let's go ahead and say that 1,000
people watch uh one of Mr. Beast's videos and then after that they click on 1,000 people clearly like Mr. Beast and they also watched, you know, your
algorithm recommends you in the same neighborhood [music] as Mr. Beast. thing. So now YouTube will start testing out your video to other people that
watch Mr. beast and you know it builds this snowball effect. Now how a lot of channels die is because of this. If the wrong people watch your video, you're never share the link to your video to any family or friends or put it in a
random, you know, Reddit thread or, I don't know, buy bots to watch your video, which you should never do because again, now this is all viewer based. [ __ ] what you have in your tags or your description because Google doesn't need
more data, right? It has every single possible YouTube video in their system. So now it can start recommending videos based on the audience's pattern. Update number four is negative [music] comments. The old YouTube algorithm saw
ton of comments, that was a good thing. But now, since it's mostly based on viewer satisfaction, if your comments are people arguing or people complaining about the video or disagreeing with you in some way or form, the algorithm is
lot of YouTubers have is to rage bait essentially. Well, that just got patched. A boring video with happy comments is now going to outperform a
really entertaining video with bad comments. Now, this update did send an entire ripple effect to YouTube, right? is a lot of things are going to change. I only covered four of them in this video, but in the Google files I went
sit here and make an hourong video to explain it all, you can click the link school community. Once you're in, you're going to see 80 plus video tutorials
where I take you step by step on how I managed to make over $2 million with YouTube, or at least what my strategy is. But you're also going to find the full report of this new YouTube update. And obviously you're also going to see
plan on posting a part two very very soon. Just kind of depends how this community, go ahead and leave a positive comment down below for Gemini. Like the
video and subscribe to the channel. Besides that, that's pretty much it. Y'all have a good day.