How the YouTube Algorithm Works (2023)

YouTube is an online video sharing and social media platform headquartered in San Bruno, California, United States. Accessible worldwide, it was launched on February 14, 2005, by Steve Chen, Chad Hurley, and Jawed Karim. It is owned by Google and is the second most visited website in the world, after Google Search.

The YouTube algorithm has one goal: suggest videos that’ll keep a viewer glued to their screen. For marketers and creators, the goal is getting the algorithm to pick our flicks so we get more clicks.

Once you figure out the YouTube algorithm’s video selection process, you’re on the way to growing a base of loyal subscribers. And turning more of the platform’s 2.5 billion users into paying customers.

So how does the YouTube algorithm work? And more importantly, how can you convince it that your videos are worth sharing? The answer has evolved over the years—what worked in 2010 won’t cut it today.

In this guide, we’ll explain what YouTube’s algorithm is, how it recommends videos, and how to make its discovery engine work in your favor.

What is the YouTube algorithm?

The YouTube algorithm is a recommendation system that decides which videos to suggest to a particular viewer. The algorithm’s job is to rank content based on the likelihood that a viewer will watch it using signals like relevance, quality, and user behavior.

It’s the gatekeeper to video views

Want more video views? The recommendation algorithm holds the key. A recent experiment showed the algorithm directly drives 70% of the views on YouTube.

In fact, YouTube’s computer code seems to override even user controls. Meaning, if you hit the Dislike or Not interested button on a video, the algorithm may still suggest similar content. That’s why it’s so important to understand how it works.

The algorithm is constantly evolving

The YouTube algorithm is an ever-moving target. Just when you’ve got it figured out, there’s an update that completely reorders recommendations and search results.

It might seem like chaos, but there’s a method to the media magnate’s madness. It gets a bit clearer when you see the three eras of the algorithm’s evolution.

• 2005 to 2011, cheap clicks do the trick: The algorithm recommends whichever videos get the most clicks. Creators game the system with clickbait titles and misleading thumbnails.
• 2012, it’s all about timing: To surface videos people actually watch instead of the old click-and-run, YouTube retooled the algorithm to prioritize watch time and percent-of-video completion. The shift encouraged some to create long-form video content and others to put out short snippets, but it did not solve the quality problem.
• 2015, getting personal: In 2015, YouTube published a very sciency looking paper that describes a new way of suggesting content. Instead of using overall popularity, the algorithm takes personal preferences (based on direct feedback and previous behavior) into account when it recommends videos.
• 2016, protecting the platform: Content moderation takes center stage as YouTube moves to moderate harmful misinformation. Creators now have to be aware of community guidelines or risk being relegated to the no-show list.

The history of YouTube’s algorithm is a forward march toward hosting quality content that keeps people on the platform. Let’s look at how YouTube’s algorithm does that today.

How does the YouTube algorithm work in 2023?

If you only take one thing away from this article, it should be that YouTube wants you to focus on your audience, not the algorithm.

“Our algorithm doesn’t pay attention to videos, it pays attention to viewers. So, rather than trying to make videos that’ll make an algorithm happy, focus on making videos that make your viewers happy,” says YouTube’s own explanation of how to reach your audience.

It’s an important distinction if you read our little history lesson. The algorithm isn’t simply recommending based on popularity. It’s more about the likelihood that an individual user will watch the next video, and the next. That’s why two viewers can end up with different results when conducting the same search.

There are three factors that affect how the algorithm matches viewers to the videos they can’t resist watching:

• Content characteristics
• The recommendation location
• External factors

Get these figured out and you’ll know how to play the algorithm like a well-tuned instrument.

Content characteristics

First off, forget the notion that YouTube watches videos and picks its favorites. The algorithm can’t do that. Instead, it looks for signals that tell it the content is a match for the viewer.

Relevance signals

Say you search for a tutorial on how to bake a coconut cake. YouTube quickly finds content that matches your query.

If the algorithm isn’t grabbing a bowl of virtual popcorn and watching millions of videos, how does it know what to show? It relies on metadata in the titles, tags, and descriptions.

Performance signals

The algorithm isn’t a critic, but it does get clues on what good content is from video metrics. The metrics YouTube uses to assess good content include:

• How long do people watch? How long is the average video view and what percent of the video does the average viewer watch?
• How often is a video watched? How many times is a video viewed when it’s recommended?
• Did people like the video? What’s the engagement rate, number of likes vs. dislikes, and average survey rating of a video?

Personal preference signals

Since YouTube’s algorithm is geared to show each viewer their ideal mix of content, it partly relies on each user’s behavior to surface videos.

To curate your mix, YouTube considers which videos you’ve watched and liked (or disliked) and which topics or channels you keep coming back to.

It also links videos people tend to watch together. If people click on “how to make macaroons” after viewing coconut cake content, you’ll probably see that in your recommendation list.

Creator signals

YouTube’s algorithm wants to make sure it’s not recommending dangerous or untrustworthy content. One way it does that is by reviewing the channel it came from. If the channel shows signs of topical authority and trustworthiness, then YouTube is more likely to recommend its content.

Recommendation location

We usually refer to the YouTube algorithm, singular. In reality, there are three places recommendations happen, and each has a slightly different set of criteria for surfacing content.

The homepage

The homepage is what you see when you first get to the YouTube website or app. Here, the algorithm is trying to get you to start watching something without knowing your current mood. The recommendations on the home screen are based on performance factors along with your previous watch and search history.

Suggested videos

Suggested videos show up next to the video you’re watching.

Now the algorithm has a little more context for what you’re in the mood for. It’ll use the topic of the current video along with your watch history to make its suggestions. It’ll also reference videos that were commonly watched after this one.

YouTube search results

It seems straightforward that in a search, you’re telling YouTube what you want to watch. But as we’ve seen, everyone’s results are a little different. Search for “bat” and see what happens.

You might get baseball videos. Don’t like sports? Expect some flying mammal content.

In this case, YouTube’s algorithm is balancing your behavior with relevancy signals in the titles and descriptions of the videos.

External factors

While you have a lot of control over how YouTube ranks your content (and we’ll show you how), there are some factors you just can’t control.

For example, YouTube suffers from seasonal dips in engagement like every social media platform. If your views drop in December, don’t take it personally. Or maybe consider creating some seasonally specific content.

You also can’t control the competition. Your best pasta recipe content will struggle to compete if Gordon Ramsay, Jamie Oliver, and Laura Vitale all dedicate a series to pasta.

One last factor you can’t control is topic interest. Viewers are fickle, and some topics drop out of favor in a flash (sorry, NFT content creators). Spoiler alert: you can spot the trends and ride their wave if you know where to look. Stay tuned to learn how.

Not factors

There are two things that don’t affect your content’s discoverability, even though they’re often suspected of doing so. First, the algorithm doesn’t know how or if you’re monetizing your videos. You won’t get a boost or a penalty if you’re getting ad revenue or are a part of the Shopping affiliate program.

Second, your publication frequency doesn’t directly affect your individual video views. YouTube states that in their tests, there’s no correlation between the time between uploads and an increase in views.

Ways to hack work with the YouTube algorithm

We would never recommend trying to game the YouTube algorithm (please don’t shadow ban us). After all, Google is really good at sniffing out cheat codes and blackhat tactics.

That said, there are ways you can make it easier for the algorithm to surface your content and share it with more people. Think of it as a partnership instead of a competition.

1. Use SEO keywords

In some ways, YouTube SEO operates like its cousin, Google Search (YouTube is owned by Google, ya know). People type words into the search bar and the algorithm looks for content related to those words. To grab the top spot on results pages, do your YouTube keyword research. Then place keywords in the filename, title, description, and subtitle script.

2. Add subtitles and closed captions to your video

Subtitles and closed captions help people understand your videos when they can’t, or choose not to, hear the audio.
The files that contain your video scripts also help search engines understand your content. That’s because the text in those files looks sort of like a blog post to search engines, so it tells them what a video is about.

3. Use Google Trends to find popular topics

Remember when we said you can’t control what topics are popular? It’s true. But you can know what’s popular and plan content around it. Google Trends is an easy and free way to learn what people look for and when they look for it. With a little work on Google Trends, you’ll know that April is the best time to post and promote mulch videos to your audience of lawn care enthusiasts.

4. Make irresistible thumbnails

You need video clicks to impress the YouTube algorithm. But there’s over 500 hours of content uploaded to YouTube every day! That’s a crowded space to stand out in. Your best tool in the fight for attention on YouTube is the thumbnail. Check out YouTube superstar Mr. Beast’s thumbnail collection.

Notice how he keeps the text to a minimum, includes a vibrant and relevant image, and always includes his smiling face. Get more tips for YouTube thumbnails (along with some examples) here.

5. Hook viewers with great intros

If people immediately click away from your videos, YouTube will take it as a signal that your content isn’t worth sharing. To keep viewers around, make sure you have an introduction that hooks people immediately.

That doesn’t necessarily mean making your intro big and flashy. What’s really important is that the first few seconds tell the viewer that the video will live up to the title and thumbnail. Crash Course is an educational video channel that does this well.

Click on one of their videos, and you’ll know right away that the content is high-quality and that they’ll answer the question posed in the title. You can use the same tactics in your YouTube channel trailer.

6. Add cards and end screens

To keep people binging your content, and upping your cred with the algorithm, include both a card and a CTA in each video. A card is a clickable image that leads to other videos in your library.

Think of them like internal links on a webpage. Place a card whenever you mention something that’s related to another video you’ve posted. An end screen, AKA end card or outro, is usually a static image shown for the final seconds of a video.