YouTube, which has become the pioneer of the video platform of the internet age and a part of our lives, has ceased to maintain its system only on demand over time, like many other industries, and makes improvements to create supply. The most effective place to see this is undoubtedly YouTube’s recommendation system.
Sometimes YouTube offers surprising suggestions about our likes and content, and sometimes it can show us content that we think has nothing to do with it. However, it is known that there is an important algorithm and data work behind all this. When YouTube first launched the recommendation system in 2008, it only offered popular videos to users, but with the improvements, it took the data and recommendation algorithms to a different dimension.
Simple principle of recommendations: Videos that will add value
An article published on the YouTube Blog includes important information about the recommendations system, as well as a good example to understand its logic. The article, which states that each user has different viewing habits, includes the following example: For example, if you like tennis videos and the YouTube recommendation algorithm notices that other people who like tennis videos like you also like and watch jazz music videos, it may suggest jazz videos to you even if you have never watched them before. . YouTube says the simple principle of recommendations is to deliver videos that will add value for users.
How are recommendations personalized?
Suggestions on YouTube appear in front of users in two places. These are the home page and the “next” panel next to the videos watched. YouTube’s recommendations vary based on many different factors, such as clicks and watch times. Here are these factors:
Click:
Clicking on a video can also mean that the user finds it interesting. As a result of these clicks, the algorithm recommends similar or indirectly related content to users.
Watch time of the video:
Understanding that clicking a video doesn’t mean it’s actually watched, YouTube added the watch time feature to its algorithm in 2012. Adding watch times, YouTube says it saw a 20 percent drop in views. But YouTube says it doesn’t care because it’s more important to add value to viewers.
Survey responses:
In addition to viewing or watching times, YouTube also pays attention to feedback to understand whether its users are satisfied with the videos they watch. For this reason, YouTube offers users a 5-star survey after watching videos, asking them to rate the content. As a result of these answers, it determines the content that the user is related to.
Share, like or dislike:
According to whether users share, like or dislike the video they watch, the YouTube recommendation algorithm gets information about the content that the user is interested in. The fact that users use these buttons actually means feedback for the YouTube algorithm.
7104
