Latest update: May 12th, 2022.
Since the performance of videos and channels is closely tied to the YouTube algorithm, understanding how it works is extremely helpful.
In this article, we will do exactly that. We will reveal how the algorithm works, based on different reliable sources including YouTube itself.
While other platforms have always been very reserved about their algorithms, YouTube is quite open. This is to help its users to maximize their YouTube channel potential and provide better content. Of course, the algorithm is not known in each one of its aspects, but its fundamental points are known. And we can take them into consideration when creating content and optimizing our YouTube channel.
Let’s start by what Google itself said a few years back, in 2016. (And, don’t worry as recently Google confirmed that all the points from that document are still valid). So, back then Google published a study explaining how its complex video recommendation system works. We have read this long study carefully and have scooped out the key points for you, that are valid also in 2022.
By the way, did you know that 70% of YouTube views come from recommended videos? For this reason, understanding the algorithm is fundamental for the growth of your channel!
So, let’s get started and by the end of this article, you will be a new person, knowing all the leaked secrets about the YouTube algorithm.
How does the YouTube algorithm work?
Generally speaking, the YouTube algorithm shows to the users a mix of the videos they “like” the most, and the most relevant according to the video the user has just seen (his watch history). It’s therefore customized according to the videos that the user likes and the channels he follows the most.
The reason for this, is that by helping users to find videos they are more likely to like and watch, YouTube can keep viewers on the platform as long as possible and motivate them to return to the site regularly.
This is why YouTube takes seriously its algorithm, and the goal is clear: try to offer each user the best, customized experience.
Our recommendation system is built on one simple principle: helping people find the videos they want to watchCristos Goodrow, VP of Engineering At YouTube
At this point, you are probably wondering how knowing how the algorithm works can help you. But, don’t worry. There are major factors that are under your control and that you can use to trigger the algorithm and help your channel. And we will see them one by one.
But before we start, here is something you need to know.
The videos are shown in five different sections of the platform, which in order of importance are: Recommended videos, homepage, search, subscriptions and trends.
The order of importance is established based on the section’s potential to generate new visits. A video that is shown repeatedly in recommended videos reaches many more new visitors than a trending video for a few days. Therefore, the order we will discuss is also the order in which each source gets more traffic on YouTube; as we said, suggested video is the king with over 70% of the total traffic.
Signals that YouTube’s Algorithm uses to evaluate videos (confirmed by YouTube)
YouTube uses a number of different signals to evaluate videos and rank them in their algorithm system. These signals build on one another to help inform the system about what users find satisfying. These signals, include…
- Watch time
- Survey responses
- Shares, Likes and dislikes
- Returning visitors
- Channel’s niche, overall health and authority
1. Video clicks (CTR)
Reason: “Clicking on a video is a strong indicator that you will enjoy it as well. After all, you wouldn’t choose to watch something you didn’t want to see”, explains YouTube.
What we learn from it: CTR is therefore very important, and so is the choice of engaging titles and thumbnails.
CTR can be a metric which can be measured on any internal YouTube traffic source, from suggested videos to YouTube search. The type of CTR – meaning on which traffic source – will have a different impact on a video ranking. For instance, a good CTR from YouTube search will increase the ranking of a video for specific keywords, while CTR from suggested videos will lead to YouTube suggesting your video even more.
But, of course, clicks (CTR) can’t be a standalone signal. This is why, since several years, YouTube implemented another crucial signal. See below.
2. Watch time (audience retention)
Reason: “Your watch time – meaning which videos you watched and how long you watched them for – provides personalized signals to the algorithm about what you most likely want to watch” So, if a tennis fan spent 20 minutes watching NBA highlights and only a few seconds watching match analysis video, the algorithm can safely assume that user(s) found the highlights more valuable.
What we learn from it: Content is crucial. Focus on providing engaging and relevant content, that will keep your viewers as much as possible.
Also, this is a very important aspect to consider if you are thinking to increase your video’s reach by purchasing YouTube views. Since audience retention is such an important metric, make sure to either use advanced targeting – i.e. with Google Ads – or use high retention views.
3. Survey responses
Reason: “This is what we, at YouTube, call “valued watch time”, hence the time spent watching a video that you consider valuable, to ensure that viewers are truly satisfied with the content they’re watching. We track valued watch time through user surveys in which you rate the video on a scale of one to five stars, and how it made you feel: was it inspiring or meaningful? Only videos with four or five stars that you rate highly are counted as valued watch time’.”
What we learn from it: YouTube is moving in a direction where they want to provide real value and emotions to viewers. Therefore, providing videos that “hit the right cords” in viewers, meaning emotional response or value, will result over time in higher exposure. On top of that, such video also tend to have more shares, which is another signal which the YouTube algorithm takes into consideration.
4. Shares, likes and dislikes
Reason: “People are more likely to be satisfied by videos that they share or like, on average. Therefore, this data (shares, likes) is used by the algorithm to predict the likelihood that you will share or like additional videos. If you dislike a video, it’s a sign that you probably didn’t enjoy watching it.”
What we learn from it: There has always been a debate, but YouTube confirmed many times, last one being September 2021, that indeed likes are taken into consideration when ranking a video. For this reason, also asking for likes, or sharing a video all over social media will have a positive benefit on our channel’s ranking.
Therefore, don’t be shy to ask for likes, shares and also comments. And even if you don’t use voice-over in your videos, no worries. You can use YouTube call-to-action animations which, in a fancy way, will visually remind your viewers to engage with your video.
If you want to invest on your channel, you can also think of purchasing shares and real engagement. Just make sure to keep it real, by getting real social media shares which can both drive traffic and have a positive impact on the algortihm.
However, YouTube also explained that when it comes to likes and dislikes, not likes are the same. For instance, a like from a person who likes every single video who watched on YouTube, will not have the same value to the algorithm as a like from someone who rarely leaves a like on a video.
Signals that YouTube’s Algorithm uses to evaluate videos (not confirmed by YouTube)
- Channel’s frequency of upload
- Prolonged experience on the platform
Signals that YouTube’s Algorithm uses to evaluate channels (confirmed by YouTube)
- Niche relevancy
- Average video quality, measured by various video signals (watch time, shares, etc)
- View per visitor
- New visitor to subscription rate
Now that we learnt about the overall most important signals, it’s time to breakdown what ranks a video in every specific traffic source of YouTube, from YouTube search to the trending tab. And, after that, we will debunk some of the most common myths on factors that actually do NOT affect the YouTube algorithm, contrarily as contrary to what most people believe. So keep reading!
How the algorithm works for each specific YouTube source
As you know, videos can reach viewers from different sources and tabs on YouTube. And, for each one of them, the algorithm uses slightly different formula’s to decide which videos to suggest or rank. The traffic sources we will see more in detail now, are:
- Recommended videos (70% of traffic, including home feed suggestions)
- Home feed
- YouTube search results
- Trending videos by country
1. Recommended (suggested) videos algorithm
Two different users will never have the same YouTube experience. To each user, YouTube shows the personalized and most relevant content and to do so, the algorithm analyses hundreds of videos that the user has previously watched and how the user interacted with them to find new content that may be of particular interest to him.
Parameters deciding what if a viewer liked the video
YouTube needs to find out what a viewer likes in order to recommend videos to him. For this, YouTube considers how a viewers interacts with a video based on the following parameters:
- How long he watched the video (audience retention)
- If he liked the video and most importantly, he shared it or subscribed to the channel afterwards.
- If he saved it in to a playlist
- If he came back to watch it (if he watched the video more than once, also known as returning visitor)
- If he engaged with other videos of the channel author of the video (and, for this reason, uploading regularly is super important!)
Apart from what a user likes, YouTube algorithm will also take into consideration what people that watch the same content or channels as you do, also enjoy to watch. Basically, YouTube tries to find people with common interest and then uses the information to suggest you new videos.
If you think about it, you surely have noticed a few times a notification on your watch page saying something like “users who watch channel x, are also watching channel y”. That is just another proof of YouTube implementing this mechanism into his algorithm. In fact, this aspect was also recently confirmed by YouTube’s VP of engineering.
Our system then compares your viewing habits with those that are similar to you and uses that information to suggest other content you may want to watch.Cristos Goodrow, VP of Engineering At YouTube
To make an example, let’s say you love watching videos about dogs. And, YouTube notices how most people who love watching dog videos, also love watching videos about Tennis (dummy example!). In that case, even if you have never watched a Tennis video before, YouTube will sooner or later show you one video about the topic.
But YouTube doesn’t take into account just the user’s watch history, his preferences, and the ones of the people who share common interests. It also takes into account a video’s social proof. This means how much a video has been appreciated by the average spectator, so the general liking of that video.
This is why in addition to “personalized” videos, YouTube recommends videos related to the topic (category) of the video that the user is watching or usually watches, which are viral or have been appreciated by other users, using metrics such as retention, shares, etc. to evaluate them.
YouTube recommends multiple channels
YouTube also explained how it is more efficient for their algorithm to show videos from different channels, so it tends to recommend videos from multiple channels even on the same content category. This, is great news, because it means that competition it’s there, yes, but there is also space for everyone.
And, you can surely confirm this statement yourself as well. Try to go on YouTube, and watch 3 or 4 videos from the same channel in a row. Even then, in the suggested videos, YouTube will show you just a few videos of that channel at best, and other videos from a variety of similar channels.
So, the golden advice here is to specialize your channel in a specific topic (this will increase the authority of each single new video) and every now and then, take a look at popular channels that upload videos on the same topic and try to understand what makes them popular.
2. Home feed algorithm
The second most popular source of YuoTube views, is a user’s home feed (on mobile) or homepage, on desktop. The home feed is a mix of the user’s subscriptions, videos suggested by the algorithm according to the user’s specific preferences and watch history, news, and a few trending videos. So, we can’t really break down this category as it is a mix of other sources discussed in this article, and their principles apply, especially the ones from suggested views.
Your homepage is what you see when you first open YouTube and it displays a mixture of personalized recommendations, subscriptions, and the latest news and informationCristos Goodrow, VP of Engineering At YouTube
3. Search on YouTube Algorithm
YouTube search is the third biggest source of YouTube traffic, and, despite being a small one, it is a big one. The reason for this is that YouTube search bring highly targeted YouTube visitors, who get to watch exactly what they were looking for, and not what someone else suggested them. For this reason, despite being a secondary source, it is important to understand how it works, as no other viewer converts as good as the ones coming from a YouTube search.
The two factors that most influence the position of your video on YouTube search results (but in general of any search engine) are keywords and their relevance. Keywords include tags, description and above all title. This is quite obvious in a way, since if a user types on YouTube “Funny dog compilation”, he wants to see a funny compilation about dogs, and not how to play with your cat.
It is important to use the keywords that the people will search for. Put them in your video titles and possibly in your tags and description too. That’s because, when ranking videos on YouTube search results, YouTube considers how the video’s title and keywords match the search made by users.
Another important thing is to select clear and relevant keywords. Try putting yourself in the shoes of the user who decides to do a search. For example, I made a video that explains how to cook a risotto. Now think what words a user can use to search for that video. One word he can use is “risotto recipe.” So you can use this word as a keyword and put it in the title and in the metadata of your video.
Other important factors
And what is really interesting to us are the secondary factors, which we can use to outrank all the other videos with the same keywords. What we mean is that, other than keywords, youtube also considers other factors, such as how many videos the user has watched from your channel in the past, the channel’s niche and the authority of the channel. And if these factors are in your favour, your video is going to rank better than those with the same keyword.
The other factors which determine YouTube search ranking are:
- CTR (Percentage of users clicking, this shows YouTube that your video is exactly what they were looking for)
- Last click (It often happens that users are unhappy about the video selected, and eventually go back to pick another one. Well, that last video, or “last click”, is an important factor as it shows that the user was satisfied, and found exactly what he was looking for)
- Audience retention
- Likes, shares, saves
- Video signals and their keywords (embedded video, shares, playlist in which it is present)
Possibly, if we were to consider two videos with similar content and targeting similar keywords, CTR and last click are the most crucial factors used by Google to decide a video’s ranking. For this reason, we developed the Keyword SEO Views, a service which does exactlty that.
4. Subscriptions page and how it can trigger algorithm
YouTube has a “subscriptions” page where the user can see all recently uploaded videos from the channels they are subscribed to (as well as posts and stories).
But subscriptions are not only a way to get views from users already subscribed to the channel, but also a way that YouTube uses to rate newly published videos through a metric called “views velocity“, which measures the number of subscribers who watch your video has just been released.
The higher the view velocity, the higher the positioning of the video.
Therefore, the subscription source is particularly important for established creators with an active subscribers base, as it can make (or break) a freshly uploaded video. But, mostly, it can help to give a little boost to new videos, if publishing is done right.
A tip is therefore to publish videos at times when subscribers are particularly active and possibly always at the same time and/or day, so subscribers know when the video is published and respond in a more reactive way. Another way is to notify users before uploading, via a post, story, or previous video.
And, of course, a catchy thumbnail! 😉 Always, always use a great thumbnail.
5. Trending videos tab
The trending page collects new and popular videos from a specific country. In this section YouTube tries to balance news videos, popular topics and content from different niches.
For trending videos, the speed of growth and sharing is therefore particularly considered. That being said, unless a video is shown at the top of the trending tab (Top 6 positions mostly), it is not really a source of traffic to aim for.
Even tho, we agree that having a video on trending tab is satisfying and no wonder many clients asked us to help them get their videos to trending.
Factors that DO NOT determine the algorithm
Now, this is important. And not just to debunk, but also to help you relax and not worry about certain things (or not obsess too much, as they are not relevant to the algorithm!).
Contrary to what many think, the following factors do not affect the algorithm
- Monetization : As specified by YouTube, whether a channel or video is monetized or not does not affect the algorithm.
In fact, YouTube says that “after a channel’s monetization gets disabled there is no difference in how the channel’s videos are promoted in search results and in the referral system” (referral system is the algorithm).
- Likes and comments : The number of likes and comments (also known as the engagement rate) has very little influence on the YouTube algorithm, unlike other social platforms. On the other hand, shares of the video are taken into consideration. What matters more than likes eventually, as we discussed earlier, is who liked the video, not how many.
- Receiving Copyright Warnings : Receiving a removal for copyright infringement or third party complaints (e.g. music featured in the video) does not harm the positioning of the channel or a video.
What YouTube says about the algorithm
In March 2021, in a blog post written by the Vice President at YouTube’s Engineering, new information was leaked regarding the algorithm with the purpose to give advice to creators to get more views. They said something that we already suspected, but now it has been officially confirmed by YouTube: YouTube does not “read” the content of the videos and the algorithm relies only on the video’s performance.
Quoting YouTube, “Believe it or not, our algorithm doesn’t have an opinion about the content you make. It actually focuses on what your viewers might enjoy based on things like videos they’ve watched, how long, shares, likes & dislikes, and feedback.”
If you want to get some more information regarding the algorithm by YouTube itself, we suggest you to watch the video below. It’s a few years old, but with up-to date information and good advice. This video is part of a whole playlist where YouTube tries to explain how their algorithm works.
This article will be periodically updated in case of news or important updates to the algorithm, so check back every now and then!