Monday, September 14, 2020

How TikTok Algorithm Works

        TikTok, a Chinese owned viral video sharing platform, has been the fastest growing social media app in 2020. Recently, U.S. government began to consider TikTok as a national security concern and warned that the platform will be banned in the U.S. this September. TikTok’s only opportunity to remain in the country is to find a U.S. buyer, which would store all sensitive data from U.S. users locally going forward.             

        With acquisition proposals for TikTok U.S. from prominent corporations, such as Microsoft and Oracle, most disagreements began to revolve around one TikTok attribute in particular: the algorithm. U.S. buyers want TikTok’s secret algorithm to be a part of the deal, while ByteDance, TikTok’s owner, is not keen on selling it.

        Many attribute TikTok’s rise due to its unique algorithm, which is loved by both TikTok creators and viewers. TikTok creators find it as the easiest platform to find fame while viewers are entertained by personalized video suggestions. Listed below is a summary of how the algorithm operates from creators and viewers point of view as partially revealed by TikTok and speculated by many prominent news sources.

 

TikTok Algorithm: Creators

  • When any new video is uploaded, TikTok shows it to a small number of users in between popular videos
  • TikTok tests the performance of the video with those users (based on likes, comments, shares, and number of followers gained) and if the video meets the performance threshold, TikTok continues to show the video to more and more viewers
  • Any video can get unlimited number of views if it continues to perform well based on Tiktok’s performance metrics, thus providing an opportunity for any creator to find fame on the platform 


TikTok Algorithm: Viewers

  • The system recommends content by ranking videos based on a combination of factors – starting from interests expressed by a new user and adjusting for things a new user indicates they are not interested in
  • TikTok continues to evaluate the user’s account based on their video preferences (what they like, comment on and share) and each user’s interaction with a video is accounted for in ranking videos and showing them to the user
  • TikTok confirmed that each viewer’s video suggestions are personalized and two users can never have completely the same video recommendations on their ForYou pages

No comments: