The TikTok algorithms are subject to constant discoveries because of their uniqueness and innovativeness. They manage to capture peoples attention for at least 10 minutes after opening the app. While Instagram can not show off such results – only 3 minutes. This is a piece of strong evidence proving the great advantage given by recommendation algorithms rather than “like&follow” algorithms. So how do they work?
From the creator’s point of view:
After the video is uploaded the TikTok analyzes every kind of content related to this video (images, videos, text, hashtags). Algorithms capable of doing that helps creators to popularize content by recommending it to the global audience based on their preferences. This is thanks to the technologies provided by Bytedance Corporation, the company that deals specifically with algorithms for video-platforms such as TikTok or its Chinese counterpart Douyin. As a rule, TikTok analyzes the duration of the video, hashtags, theme, music used, location, time of publishing etc.
From the user’s point of view:
Each of these factors is individually considered by TikTok’s For You recommendation system, which means that every For You page will be fully unique up to the interests of a particular user and his level of interest:
- Users interaction: liked/shared videos, followed accounts, the comments you post, and the content you create.
- Video you like/share information: This may include details such as captions, music and hashtags.
- Device and account settings: language settings, country settings, and device type.
For instance, TikTok reveals that a strong indicator of interest may be long videos (up to 60 sec) watched be a user. This will be more significant than a weak indicator, for example, the viewer and video creator are in the same country. Videos on the page “For you” are classified according to these “indicators of interest” and on the probability of user interest in certain content.