How to Understand The Youtube Recommendation Algorithm

It would be wonderful if YouTube would send us a message when uploading a video saying, "Great job, your video is excellent, it will be at the top of search results!" or "Try harder, pay attention to video optimization." In reality, it's a bit more complicated.

YouTube's algorithms often communicate with us in a complex language, this makes it extremely complicated to determine what it wants from our content. We need to consider the rules, recommendations, user experiences, and do research to determine what exactly the platform pays attention to when it likes a video.

This is called signals, a kind of trigger for YouTube that help assess the efficiency, authority, and quality of a video.

YouTube Search Signals: Relevance, Engagement, and Quality

With over 500 hours of content uploaded every minute, algorithms and ranking tools are essential for organizing and distributing such a vast amount of information. These tools aim to assist creators and help users find the best content they desire, benefiting the platform by processing and distributing data.

YouTube values transparency between creators, the platform, and viewers. To ensure creators' efforts are not in vain, YouTube strives to provide abundant information about platform rules, offer content creation and promotion advice, and present it in a user-friendly manner.

Understanding how YouTube search works is crucial to grasp what signals are significant in evaluating your video. The platform's algorithms primarily focus on three key factors: relevance, engagement, and quality.

In addition to the three main factors, YouTube also considers additional indicators:

- Viewing and search history

- Relevance and popularity

- Authority and trustworthiness of information

The specific additional indicator that YouTube takes into account depends on the type of content. For example, for music or entertainment content, relevance and popularity are crucial due to rapidly changing trends. News, medical, and political videos require a special assessment of source reliability and authority.

YouTube ranking signals in search results

➔ Relevance

First and foremost, relevance is about matching search results to the user's query. When searching for information, users should receive specific, relevant, up-to-date, and comprehensive content.

If you're searching for something very unique or exploring a narrow topic, you'll likely see all relevant search results. In other cases, the platform aims to show the best results first. But what does "best" mean?

On YouTube, relevance is determined by the alignment of the video's title, description, tags, and content with the user's query. This is why optimizing your video is crucial. A well-chosen title that accurately reflects the search query and keywords solves most issues with search ranking.

YouTube, like the Google search engine, generates results based on keyword queries. If you're unsure about your video's title, there are analytical tools that can assist you.

You can use browser extensions like VidIQ or online keyword analysis tools like Google Trends. This way, you can identify what users are searching for and how they formulate their queries on YouTube. Pay attention to how other creators format their titles within your niche. This will provide insights into what content is more popular and what keywords are trending.

The more accurately your video matches a user's query, the higher its relevance. Remember that YouTube offers many built-in analytics tools. In YouTube Studio, navigate to Analytics, go to the Content section, and check "How viewers find your content." Under the Traffic source category, select YouTube search to see which queries viewers use to find your videos. You can compare these queries with the ones suggested by analysis tools like Google Trends and assess their effectiveness.

The "Research" tab is also available, replacing keyword analysis tools. Here you can find out what users are searching for on the platform, including your viewers. Currently, the feature is available for Australia, Canada, India, the United States, and the United Kingdom.

To ensure a video meets user expectations, YouTube utilizes viewing and search history (if enabled). By default, it is enabled, but can be manually disabled. Consequently, search results for the same query may differ. For example, if two different individuals search for the word "cricket," which has two meanings - insect and the sport, the results may vary. However, if you frequently search for sports-related content, you will likely see cricket matches rather than insects.

It is important to bust the myth that videos with optimized metadata will always appear at the top of search results. Optimization is indeed crucial and the first step towards popularity, but not the only factor. All the elements that YouTube responds to must be considered, leading us to the metric of engagement.

➔ Engagement

Engagement is a metric that is evaluated based on signals from viewers, specifically how users interact with a video. This primarily includes the watch time in hours and the duration of views.

How to evaluate audience engagement data

It helps to understand how well the video meets the audience's expectations and whether it should be recommended in search results. Other engagement metrics include the number of views, likes and dislikes, and the number of subscribers. There is debate about the impact of subscriber count. While YouTube considers this metric, it is not the sole determining factor.

How to evaluate the engagement rate yourself?

To do this, there is the “Overview” tab in YouTube Analytics. Select a specific video in the Content section and navigate to its Analytics.

The key metrics to pay attention to are watch time (in hours) and average view duration. It's important to note that this engagement data doesn't appear immediately after the video's publication but takes a few hours to populate. After the initial data calculation, the statistics will update more regularly.

Evaluating these metrics right after publishing doesn't make sense; the video needs time to find its audience.

Also, pay attention to the "Audience Retention" report in the "Overview and Engagement" tab. It highlights four key moments in the video: introduction, best moments, peaks, and drop-offs. This helps identify which parts resonated with viewers and which parts they may have wanted to skip.

➔Video quality

The third key signal that YouTube considers for video ranking is its quality. 

Quality is a wide concept, but YouTube has its own definition. Quality refers to the presence of valuable and truthful information on the channel about a specific topic. The video should align with its topic, address the viewer's questions, and provide value. This signal evaluates the video's content itself, rather than its platform presentation. 

For creators who consistently produce content and aim to reach a larger audience, it's important to perceive quality slightly differently. Authority and reliability of information are the foundation of creating a quality video. This is where its value lies. Use trusted sources, question the data you receive, and ask questions. If your channel has a clearly defined theme, quality videos will enhance the authority of the entire channel. The platform aims to reward creators who consistently produce high-quality content on their chosen topic.

➔Extra signals

Depending on the type of content, additional analysis tools may be required for video evaluation. For music or entertainment-related searches, YouTube focuses on popularity and relevance.

Sometimes, the same song title may appear in the works of different artists. Films may have the same title but be produced 20 years apart. Users may search for a piece by a well-known artist who has been in the industry for decades. In such cases, YouTube needs to determine what the viewer is more likely to prefer and what they intended to find when creating the search query. YouTube will prioritize more relevant and popular search results.

Certain categories of videos place a strong emphasis on accuracy and reliability, such as news, medical, and political content.

Only videos that use trusted sources can rank high in search results. Sometimes, when watching a health-related video, a special panel may appear below the player with additional information about the medical source.

To determine the reliability of a source, YouTube refers to guidelines developed by organizations such as the World Health Organization (WHO) and the National Academy of Medicine. These nonprofit organizations focus on disseminating scientific information related to health and U.S. legislation in this field.

In 2021, a group of experts from the National Academy of Medicine published a report on how to determine reliable sources of medical information shared on social media.

The study resulted in the development of the following principles:

1. Scientific basis: Information should align with the best scientific standards available at the time.

2. Objective: Sources used should have a scientific objective rather than a financial or political one.

3. Transparency and accountability: Sources should provide the full context of research and data, including limitations and possible errors, allowing users to make informed conclusions.

If a video is based on such sources, it may display a panel with a link to an official source. This practice is gaining popularity in the United States and the United Kingdom and is likely to become a global standard for this type of content.

What will happen if the sources are unofficial?

The visibility and ranking of a video will depend on the information presented in the video itself. If the sources used are unofficial but the video aligns with the principles of scientific truth, it is likely to appear further down in the search results, following videos with links to official sources. No specific labeling will be applied to your video in such cases.

However, if the data presented in your video is highly controversial or deemed as disinformation, YouTube may take measures to limit its promotion or apply sanctions accordingly.

Signals for YouTube to get into Recommendations

The YouTube recommendation system is designed to help users discover videos they will enjoy. It provides personalized recommendations based on your search history, interests, and recent news, which are displayed on the main page.

When you watch a specific video, the "Up Next" panel appears after the video, showing you related videos that may interest you. Additionally, there is a list of recommended videos on the right side of the currently viewed video.

Video recommendations are not solely based on your personal preferences. The algorithms also consider the interests of all users on the platform and try to find commonalities. This allows YouTube to create mini-communities based on shared interests.

For example, if you frequently watch tennis-related videos and the system notices that other users who watch tennis videos also enjoy jazz music, it may recommend jazz to you as well. That's why you may sometimes see videos in your recommendations that seem unrelated to your previous interests.

YouTube continually strives to improve the quality of content on the platform, and user feedback plays a crucial role in achieving this. Feedback can be obtained through surveys or by analyzing statistics.

One important metric for YouTube is "valuable watch time." Users are asked to rate the video they watched on a scale of 1 to 5 stars, and the watch time of videos that receive 4 or 5 stars is considered valuable.

Of course, not all users fill out the survey after every video. To address this, YouTube has trained a machine learning model to predict responses to such surveys. To verify the accuracy of these predictions, YouTube excludes ratings already provided by users. This helps assess how well the predicted data aligns with actual responses. By making such predictions, YouTube can determine the quality of a video and how well it suits individual users.

It takes some time for YouTube to understand the content of a video in order to include it in recommendations. This determines which users will see the video.

At this stage, metadata plays a crucial role. Clear and comprehensive metadata enables the platform to quickly understand the topic and content of the video. Some users have noticed that recommendations often feature videos from popular channels that they could have discovered themselves, as those channels are already popular. Why, then, does YouTube not actively recommend videos from smaller channels?

The reason is that smaller channels have less information available simply because they have fewer videos. The more data YouTube processes from your channel and the more videos it analyzes, the clearer it becomes for the platform to identify your target audience.

Another signal of your interests that helps understand which videos you like is subscribing to a channel. However, research has shown that subscriptions are not the sole determining factor for YouTube. Many users subscribe to a channel based on just one viewed video, as a way to express gratitude to the creator. Analyzing recent views and search history is much more important in understanding what appeals to the viewer.

Signals for the “Trending” tab

Many creators wonder how their videos can make it to the "Trending" section. Let's first understand why it is important to be featured there.

The "Trending" section showcases videos that are interesting and popular among a large audience. To be featured in this section, the content should not only attract a significant number of views but also be new and relevant to current events. These videos are not personalized and appear the same for all users within a specific country. The "Trending" section is updated every 15 minutes.

YouTube has its own algorithms and ranking principles to determine the content that makes it to the "Trending" section. The following factors are considered when selecting videos for this section:

- Appeal to a broad audience.

- Avoidance of shocking or misleading content.

- Reflection of the breadth of events happening in the world.

- Demonstration of creative and cultural diversity among creators.

- Uniqueness and non-standard presentation.

YouTube aims to find the optimal balance of these factors. Additionally, quantitative data is taken into account, including view count, the speed at which views are accumulating, where the videos are being watched (including views outside of YouTube), time since publication, and the popularity of the specific video compared to other recently published videos on the channel.

It is important to note that controversial videos cannot be featured in the "Trending" section or recommended. Controversial videos refer to those that may not violate community guidelines but are on the edge of acceptability. YouTube cannot legally remove such videos, so they do not promote them in search results or recommendations.

The main goal of every creator on YouTube is to create high-quality content that will find its audience.

The algorithms of the platform work to help users quickly and easily find what they are looking for in a sea of videos. Even better, they aim to help users discover something truly interesting and valuable, rather than a random assortment of videos with similar themes.

There are specific signals and indicators that YouTube relies on to determine how one video is better than another. Relevance, engagement, and quality are the main factors considered. These parameters take into account both the quantitative metrics of a video (which can be seen in the "Analytics" section) and the qualitative content of the video.

However, it is important to remember that the most crucial aspect is your creativity as a creator. If you create content that is important to you and understand that it is important to others, you will always find your place in the hearts of your viewers and like-minded individuals.