How does the YouTube recommendation system work?

The YouTube recommendation system is the best tool on the video platform to generate traffic and audience.

In a recent post, Cristos Goodrow, VP of Engineering at Youtube, explains that this tool generates “a significant amount of general audience on YouTibe, even more than subscriptions or channel searches”.

The problem is that most users do not know how the recommendation system works and it is this gap that Goodrow intends to fill.

The recommendation system

In the post, the vice president of Engineering points out that “our recommendation system is based on the simple principle of helping people to find the videos they want to watch and that creates value for them. A user can find recommendations in two main places: on their homepage and on the “next” panel. The homepage is what the user sees when they first open YouTube — it displays a mix of personalized recommendations, subscriptions and the latest news and information. The “next” panel appears when you’re watching a video and suggests additional content based on what you’re currently watching along with other videos we think you might be interested in.”

When it was implemented in 2008, the recommendations system was limited to showing the videos with the most views, ranking the content based on their popularity, without taking user preferences into account, and creating a “Trends” page.

Cristos Goodrow explains that “today, our system ranks billions of videos to recommend content tailored to your specific interests . Unlike other platforms, we do not connect viewers to content through their social media. Instead, the success of YouTube recommendations depends on an accurate prediction of the videos you want to watch.”

“To do this, we assume that everyone has unique viewing habits. Our system then compares your viewing habits with those that are similar to you and uses that information to suggest other content you might want to watch. So, if you like tennis videos and our system understands that other people who like the same tennis videos also like jazz, you can recommend jazz videos to you (for categories like news and information, this may work differently – more info on this for later) — even if I’ve never watched a single one before.”

Finally, the post’s author remembers that “we also know that not everyone always wants to share these information with us. So we’ve developed controls that help you decide how much data you want to provide. A user can pause, edit or delete their YouTube searches and their viewing history whenever they want.”