15 min read 
TL;DR
- YouTube doesn't run on one algorithm. Search, Home/Browse, Suggested Videos, the Shorts feed, Live, and Notifications each have their own ranking system and their own priority metrics.
- CTR, average view duration (AVD), and satisfaction (likes, shares, survey answers, "not interested" clicks) matter most across the board, and satisfaction now outweighs raw watch time on its own.
- Shorts and long-form videos rank in different systems. A channel can run both successfully, but the playbook for each is different.
- Staying active, meaning regular uploads plus live streams, is a real signal for the Browse Features algorithm.
What Is the YouTube Algorithm?
Here's the thing worth unlearning first: "the YouTube algorithm," singular, doesn't exist. What people call by that name is really a group of separate ranking systems, each covering a different part of the platform (search results, the home page, the "Up next" column, the Shorts feed, live streams, notifications), and each leaning on a different mix of signals to decide what to show a given viewer at a given moment.
That group of systems has changed shape more than once since YouTube launched in 2005: from pure view counts (2008) to watch time (2012), then to engagement and the fight against misinformation (2016–2019). Between 2023 and 2026, it split further into separate systems built around format, with satisfaction now outweighing raw watch time.
Below is where each system operates today and what it cares about most.
| Surface | What it ranks | Key signals in 2026 |
|---|---|---|
| Search | Results for a search query | Metadata match (title, description), engagement on that query, channel authority |
| Browse Features | Home page, trending | Watch history, repeat visits, channel activity, satisfaction |
| Suggested Videos | The "Up next" column | Topic similarity, shared viewership, AVD, returning viewers |
| Shorts feed | The vertical short-form feed | Scroll-stopper signal, view-through rate, swap-away rate |
| Live | Live streams in recommendations and subscriptions | Per-viewer watch time, concurrent viewers, chat activity |
| Notifications | Push alerts and the bell icon | Subscriber relevance, notification-open history, timing |
A handful of factors show up across most of these systems:
- Click-through rate;
- Average view duration and overall retention;
- Satisfaction: likes, shares, post-watch survey answers, "not interested" or "don't recommend channel" clicks;
- How many of a channel's videos a given viewer has already watched;
- How recently that viewer engaged with similar topics;
- Recent search and watch history;
- Demographic and geographic context.
One quick clarification while we're here: impressions and views aren't the same thing, and people mix them up constantly. According to YouTube's documentation, an impression is counted once your thumbnail has been visible for more than a second with at least half of it on screen. It says nothing about whether anyone clicked. A view only counts once that click leads to an actual watch. If you're getting plenty of impressions but a weak click-through rate, that's usually a thumbnail or title problem, not a content problem.
What Changed in the YouTube Algorithm in 2025–2026
Four things explain most of the gap between how ranking worked a couple years ago and how it works now. YouTube's own Creator Insider channel, run by staff from the Creator Liaison team, is a decent way to keep up with these shifts directly from the source rather than through secondhand recaps.
Satisfaction beats raw watch time
Watch time used to be almost the whole story. Not anymore. YouTube's help documentation now lists satisfaction surveys as their own signal, separate from watch history, built specifically to capture whether a viewer actually enjoyed a video rather than how long they sat through it. A five-minute video with strong satisfaction can beat out a twenty-minute video with weak satisfaction, even if the longer one technically racked up more watch time.
A more AI-driven recommendation model
YouTube's recommendation system leans more heavily on machine learning models that connect what a viewer does across formats, long-form, Shorts, and live, into one preference profile instead of treating each format as its own audience. For channels that publish across formats, that means a Short that does well can actually pull new viewers into your long-form videos, not just circulate inside the vertical feed.
Shorts and long-form: different ranking systems
Shorts and long-form videos get judged by different logic, and YouTube's own guidance backs that up: short-form is ranked on view-through behavior and post-watch reactions, not on watch time the way long-form is. That has a real downside. A viral Short that's off-topic for your channel can attract viewers who never touch your other content, which pulls down your channel's overall engagement numbers even though the Short itself did great. One channel can run both formats well, but each needs its own plan.
The "shown in feed" metric
YouTube Analytics tracks a "shown in feed" number for Shorts, separate from general impressions. Impressions cover thumbnail views across the home page, search, and suggested panels. "Shown in feed" only counts how often a Short showed up inside the Shorts feed itself. Comparing the two tells you whether a Short is getting discovered through the vertical feed or through search and channel pages, and that points you toward a different fix depending on which one's lagging.
The Browse Features Algorithm (Home Page and Trending)
Browse Features sends the most traffic for most channels. It builds the home page and the trending section from:
- Watch history and repeat visits;
- Subscriptions and past interactions with a channel's content;
- How regularly a channel publishes and stays active, including live streams;
- Recent satisfaction data;
- Regional popularity, specifically for the trending section.
In practice, videos that pick up strong engagement in the first 24 to 48 hours have the best shot at sticking in regular viewers' feeds. Consistent uploads paired with live activity tell this system a channel is alive between main releases. That's where a tool like Gyre comes in: it turns pre-recorded video into a continuous 24/7 live stream on YouTube, Twitch, or Kick, so a channel stays visibly active in the gaps between scheduled uploads, which is exactly the kind of activity this system rewards.
The Suggested Videos Algorithm
The "Up next" column pulls from:
- How closely a video's topic matches the one currently playing;
- AVD for the current video and others like it;
- Content that did well with viewers who share a similar profile;
- Returning viewers, meaning the share of an audience that comes back to a specific creator after their first watch.
Returning viewers carry a lot of weight here in 2026. A viewer who comes back the next day for more of your content is worth far more than a one-time new viewer.
The Shorts Feed Algorithm
The Shorts feed judges content on different terms than long-form watch time. It looks at:
- Scroll-stopper: whether a viewer stops on your video mid-swipe;
- View-through rate: the share of a 15 to 60 second clip watched in full;
- Swap-away rate: how quickly a viewer swipes past, which counts against you;
- Likes, shares, comments, and post-watch survey answers.
For a closer look at this system specifically, check our guide to the YouTube Shorts algorithm.
The Live Algorithm
Live streams get ranked on their own and get an extra push in Browse and Notifications. The main factors are how long viewers stick around during a stream, how many people are watching at once, and how active the chat is. Channels that stream regularly, including running 24/7 through something like Gyre, tend to land higher in subscribers' feeds while they're live.
The Search Algorithm
Ranking in YouTube search comes down to:
- How precisely keywords show up in the title, description, and tags;
- The channel's authority behind the video;
- How long viewers who arrived through that specific search term keep watching.
The Notifications Algorithm
Notifications run on their own logic, deciding which subscribers with the bell on actually get a push alert for a new upload. It looks at how often a subscriber opens that channel's notifications, their history with the channel's content, and the best time to reach them based on their time zone. Ignore a channel's notifications a few times in a row, and you'll stop getting them, bell icon or not.
7 Tips to Increase Reach on YouTube
Each tip below targets a specific system.
Use precise keywords in titles and descriptions (Search)
Keywords help the search system figure out what a video is actually about and match it to the right queries. Google Trends, Ahrefs, or just YouTube's own search bar are all fine ways to find what people are searching for. Descriptions and tags get scanned too, so put keywords there as well, not just in the title.
Hold attention from the first seconds (Browse + Suggested)
The opening seconds of a video work almost like their own checkpoint. If a lot of viewers bail early, the system reads that as a sign the video isn't matching what people expected, and pulls back on distribution. A strong hook, a surprising fact, a clear promise about what's coming, helps here. Just don't overpromise: if the payoff doesn't match the setup, satisfaction tanks along with it. Our roundup of the most popular videos on YouTube is a decent source of ideas for what a strong opening looks like.
Ask for engagement without overdoing it (Satisfaction)
Likes, comments, and shares all count toward satisfaction, so a clear call to action near the start and end of a video still earns its keep. But the flip side matters more than people think: a "not interested" or "don't recommend channel" click hurts more than a new like helps, so pushing too hard can backfire.
Design thumbnails that match the content (CTR)
Contrast, legible text, and a clear focal point all help your click-through rate. The one thing to avoid is a gap between what the thumbnail promises and what the video delivers. That gap shows up later as weaker retention and satisfaction, and that's what actually hurts your standing with the algorithm.
Add subtitles (Search + Retention)
Subtitles, especially in multiple languages, reach viewers watching without sound and give the search system more text to match against. YouTube's automatic captions cover the basics; a manual edit catches what they get wrong.
Publish on a schedule you can actually keep (Browse Features)
Consistency signals an active channel to Browse Features better than volume does. One reliable upload a week beats five videos one week and silence for a month. Live streams can fill the gaps between uploads and keep a channel visible in subscribers' feeds even on quiet weeks.
Reply to comments, especially early (Satisfaction)
An active comment section is a good satisfaction signal. Ask viewers to comment directly in the video, and reply to early comments within the first few hours. That reinforces the signal during the window, roughly the first 24 to 48 hours, when the system is deciding how far to push a video.
YouTube Algorithm Myths
A few common claims about ranking don't hold up.
Shorts get an automatic ranking advantage
Shorts run on a separate system, not a friendlier one. The format isn't inherently favored; it's just judged on different terms. Any video, long or short, does well when people watch most of it and respond positively.
Well-chosen tags significantly boost rankings
Tags play a minor, corrective role at best. YouTube's own help documentation says plainly that tags mainly help when a title has a common misspelling. On their own, they're not much of a lever.
More likes means a higher ranking
Likes are one piece of satisfaction, not the whole picture. Shares, returning viewers, survey answers, and negative feedback (dislikes, "not interested," unsubscribes) all factor in too. A channel with 10,000 likes and a weak returning-viewer rate can rank below one with 2,000 likes and strong repeat viewership.
The algorithm penalizes external links
External links don't automatically hurt a video's ranking. The problem shows up when a link has nothing to do with the video's topic and reads as spam. That's what tanks satisfaction, and satisfaction is what actually moves the needle on ranking.
New channels don't stand a chance
Also a myth. The algorithm tests every new video on a small seed audience regardless of channel size. The first 24 to 48 hours are the decisive window: the system checks CTR, AVD, and satisfaction on that small group before deciding whether to expand reach, and a brand-new channel goes through the exact same test as an established one.
If you're building out a broader content strategy, our guide on how to optimize your YouTube channel covers more ground beyond ranking factors.
Key Takeaways
- YouTube doesn't have one algorithm. It runs separate systems for Search, Browse, Shorts, Live, and Notifications.
- The biggest factors are CTR, average view duration, and satisfaction.
- Shorts and long-form videos compete in different systems. One channel can succeed at both.
- Staying active, meaning regular uploads and live streams, is a good sign for the Browse Features algorithm.
- The first 24 to 48 hours after publishing are decisive. The algorithm tests a video on a small audience before deciding whether to push it further.
FAQ
Why isn't YouTube promoting my video?
The most common reasons: a low thumbnail click-through rate (under 2%), weak retention that drops below 30% in the first 30 seconds, or a topic that doesn't match what your current audience usually watches. All three show up in YouTube Analytics.
How often should I post for the algorithm?
Consistency matters more than frequency. One video a week, reliably, beats five videos one week and silence for a month. The Browse Features algorithm cares about your rhythm, not your total output.
Do likes affect the YouTube algorithm?
Yes, but less than average view duration and click-through rate. Dislikes and "not interested" clicks count too, and the system uses them to fine-tune what it recommends. Shares and returning viewers generally carry more weight than likes alone.
Do live streams help the algorithm?
Yes. Live streams get a separate boost in Browse and Notifications. Channels that stream regularly, including through a 24/7 setup like Gyre, tend to see better placement in subscribers' feeds during a broadcast and a stronger activity signal between regular uploads.
