How Does YouTube Algorithm Work?
Have you always wondered how YouTube understands you better than anyone else on the planet? It gives you relevant videos for topics you’re browsing or previously searched for. It’s even your therapeutic teacher who can help you console in terms of crisis.
But, how can someone read your mind like a psychic? In short, YouTube Algorithm does the entire job for you to visit the platform again and again. So what exactly is YouTube Algorithm?
YouTube Algorithm in simple terms can be defined as a set of guidelines or instructions followed to enhance the user experience by making them binge content continuously based on their watch pattern, likings, and browsing history.
So, was this Algorithm introduced with the invention of YouTube? No, it was gradually discovered in course of bettering the reach of the platform. Let’s dive into its history to understand the working of the Algorithm better.
YouTube Algorithm Recap
2005–2011: Views and Clicks Priority
With the introduction of the platform, YouTube was not keen on making it a popular and eye-catchy platform for the public. Hence, YouTube Algorithm randomly started showing the videos on the homepage which had more views and clicks. This resulted in bad quality videos which had fake comments, likes, misleading titles, and click baits showing up on everyone’s homepage.
2012–2015: Watch Duration
Here, the main motive of YouTube was to increase the retention time of users on the platform by upgrading the quality of the content. To enhance the user’s experience, YouTube started to show short and long format videos on the homepage. Short videos had more views and completion time. Long videos increased the overall watch duration of the platform. Even Likes, Dislikes, Shares, and Not Interested options were taken into consideration.
2016: Deep Learning Optimization
As there were millions of videos generated each day, YouTube found it daunting to recommend videos that were high in quality and relevance. As a result, they adopted an integrated approach of Machine Learning and AI (Artificial Learning) in their algorithm. This system uses a funnel approach, where videos are filtered based on viewer count, history patterns, and then reaches the ranking funnel. After reaching the ranking funnel, they’re presented a score and are shown to the user based on their score value on the homepage.
2017–2020: Reforming Community Guidelines
YouTube has made serious efforts to remove borderline content by 70% as per 2019 reports. This borderline content would include any misleading, fake, misinformed, extremist propaganda, violent and another demeaning form of content to be demonetized if the creator fails to keep up with the community guidelines.
How exactly does the YouTube Algorithm work?
The YouTube Algorithm primarily works on 3 factors:
(a) History and Preference Pattern: This depends on the history (search and watch) and most likely videos to be watched by the viewer on a routine basis. This gives a mapping idea to the algorithm and recommends videos as per the user’s choices.
(b) Engagement Parameters: This includes the video’s performance in terms of the channel’s affinity with the user, likes, shares, watch time, and the Seasonal Inclination of the audience towards the video. Seasonal Inclination here would mean creating videos on the current trend or scenario.
E.g. during summers, viewers would be interested in watching
- Outfits to wear during summer
- Summer Drinks that you can try at home
- How to protect your skin from harsh summers
- Must visit hill stations in India
(b) Other Factors: The channel’s reputation with the audience can impact the overall market. This could be from other sources such as Social Media Channels, Blogs, and other sources.
So, the goal from 2021 for YouTube Algorithm has been to customize videos for each viewer as per their choice, so this makes them get hooked to the platform and be a long-time consumer of their platform.
Let’s jump into the 5 user groups; YouTube Algorithm uses to optimize their content on the platform.
(1) Homepage
(2) Suggested Videos
(3) Search
(4) Trending
(5) Subscriptions
According to YouTube reports 2021, Homepage and Suggested videos are the prime source of traffic generation for the majority of channels. Tutorial or instruction-based videos generate traffic from the search bar of YouTube.
Now, let us look at how YouTube determines its Algorithms for the above-mentioned 5 groups.
(1) Homepage: This gives a generalized idea, as the Algorithm is still not able to figure each viewer’s customized choice. The homepage usually displays the viewer the videos based on their watch history, subscriptions, and video success. Again the video success would include the likes, dislikes, views, CTRs (Click through Rates), average view percentage & duration, and feedback through surveys. This factor is highly unstable, as the audience’s behavior keeps fluctuating with time.
(2) Suggested Videos: Here the Algorithm decides based on the viewer’s (a) Personalization behavior such as watch history, (b) videos that are frequently watched together or in a loop, (c) related topics, or similar to the video currently viewed (d) videos from a particular channel. Due to these factors, each user’s Home and Suggested tab displays different results.
YouTube has also observed people are more attracted to watching videos recommended from other channels. This helps YouTube expand its results page for the suggested tab.
(3) Search: To rank in the video search, make sure to follow Relevant Keyword Optimization and Engagement metrics. YouTube is also like a search engine for your video platform, hence you should think in terms of what the audience would type in the search bar. This is where partial SEO becomes necessary in your optimizing your keyword data in your title, description, metadata, and content as per the user’s query. Make your YouTube descriptions customized and should explain the content in-depth as much as possible with feature keywords so that your video pops up for the relevant keywords typed by the user. Make sure not to just blindly stuff your keywords unnecessarily.
Engagement metrics will be evaluated based on the popularity in terms of views, likes, and shares. If the user has subscribed or viewed a particular channel in the past, the chances of that channel (related to the topic) ranking in the search query are higher.
(4) Trending: Popular and new types of videos are displayed on the trending page. These videos are ranked based on their view count and rate of view growth for every video. This can be accessed by clicking on the explore tab on the homepage and then choosing various categories from the Trending page. The trending page has been divided into 4 categories: “Now”, “Music”, “Gaming”, and “Movies”. The trending page is specific to the particular country.
(5) Subscriptions: The audience gets notified every time their favorite YouTuber uploads a new video by subscribing to their channel. YouTube Algorithm doesn’t regard the number of subscribers to rank a channel. But, it relies on a factor called ‘view velocity’ for ranking a video from the subscription data. View velocity computes the number the viewers who watch the channel’s videos consistently when new content gets uploaded. Greater the view velocity, the greater the channel ranks. YouTube Algorithm also measures the number of active Subscribers for assigning the channel ranking.
Final Thoughts
The present online space is experiencing saturation as many people are entering the market with new videos and ideas every second. You need to keep reinventing yourself with the latest trends every day and stay consistent in your journey. Always focus on providing high-quality, innovative, and solution-oriented videos to your users. Remember YouTube is a people-centric platform, where the Algorithms are dictated by the users and not the platform.
Zakiya Fathima
