The TikTok Algorithm: How it Works

Matt Wurst

VP, Client Management

TikTok uses a system of metrics and video evaluation tools to better understand its community members and offer them content with a high engagement rate.

With billions now forced to shelter in place, finding new creative outlets and means for passing the time have shifted digital behaviors yet again. We’ve talked about the emergence of TikTok for a while now, but the steady ascent of app downloads, usage, and videos created are up (on average) nearly 20% over the past 6 weeks.

Brands that may have been skeptical even a few months ago have seen the rules of engagement change, as well. Consumers’ tolerance, openness, and acceptance (nee, appetite for?) branded content on TikTok is far past the tipping point now. Omnichannel means every channel, and TikTok reaches audiences on a global scale. And like other platforms that brands have learned how to optimize content creation and distribution for, it all starts with mastering the algorithm.

TikTok’s platform serves one primary objective: to show and offer content with high engagement rates, taking into account which members/profiles of the community have seen, will see, and interact with it. But how is content specifically curated “For You?” Let’s break it down how it works:

1.  Each video uploaded to the platform is analyzed using artificial intelligence using natural language processing and computer vision technology

2. Information is extracted from each video to better understand the context and content as follows:

  • The Visuals: people, frame, decor, atmosphere… all aspects of the video elements are analyzed.
  • The Audio: music, lyrics and other additional elements to better balance the video.
  • Meta Data: all keyword and manually created aspects of the video are taken into account: name of the video, description & the hashtags used.

3. Once the content has been analyzed from every angle, it is boosted by TikTok with a small number of users in order to assess its virality. User reactions are evaluated using a point system:

  • Revisit rate: 10 points
  • Completion rate: 8 points
  • Shares: 6 points
  • Comments: 4 points
  • Likes: 2 points

If the score of a video exceeds a certain threshold, only then will be pushed to a larger number of users until it reaches mass spread on TikTok. 

From Many, Comes One

Personalization of content is also a priority, and TikTok wants to get to know individual users better in order to offer content that is more in line with their interests and expectations.

It starts the first time they even open TikTok: a video stream is submitted to each user without them even having to register. That’s when the personalization learning begins. Not only does the social network learn as much as possible about you and your consumption habits, but it also tries to make you stay as long as possible on its platform by offering videos with high engagement rates.

The point system works not only to study the virality of videos, but also to analyze user profiles. Each time you watch a video, your “score” will change and the recommendations for videos prioritized within your experience will be adapted. Moreover, it is quite natural that the more videos you watch and consume, the more your profile preferences can and will be optimized by TikTok. Other factors that impact the algorithm include day/time of viewing, geolocation, and your friend/follower network.

Bottom line: the more you consume, the more TikTok will adapt and give you more of what you like. So stick around and help them to better help you.