YouTube’s recommendation system connects billions of people around the world to content that uniquely inspires, teaches, and entertains. Given everyone has unique viewing habits, the core job of recommendations is to meet user needs by helping people find the videos they want to watch and that will give them value.
Recommendations on YouTube
At the highest level, our recommendations system is designed to anticipate and meet a user's needs to drive value through relevant and satisfying viewing experiences. Understanding a user's viewing behavior, likes, dislikes, subscriptions, and feedback, including from satisfaction surveys, helps us do this most effectively. And just like on Search, we also consider the reputation and the quality of a channel to determine how, when, and to whom the content can be surfaced to, and use external evaluators to help us understand how an average viewer might perceive the content to ensure users have a high-quality experience.
We also know not everyone wants to always be logged in or share their information with us, so we’ve built controls that help people decide how much data they want to provide. Users can always pause, edit, or delete their YouTube search and watch history whenever they want.
You can find recommendations at work in two main places on YouTube:
How YouTube Elevates High-Quality Information
For topics where accuracy and high-quality are key — like news, personal finance, and medical and scientific information — we elevate high-quality voices using a variety of tools. Here’s how it may look:
How YouTube Provides Additional Context
To provide additional context about certain events, topics, and publishers, we use information panel across YouTube, which may look like:
For topics subject to misinformation, we may link to third-party sources to give more context.
Information panels on the video Watch page make it easier for users to assess an organization's background.
We require that creators disclose realistic content that is made with altered or synthetic media, including content made with GenAI, and then labels appear on the video or in its description.