The leak of "TikTok Algo 101"
In a document leaked by TikTok itself called "TikTok Algo 101", it explains how TikTok's algorithm works systematically to make you want to use the app more and more. The document has been covered by many different news media, including The New York Times. They write that the document addresses both the mathematics of the algorithm and how it "reads" people. Here, we will briefly go through how the algorithm creates increased use of the app, step by step.
Step 1 - Viewing time
When you use the app, the algorithm will first analyze how much time you spend on a video (watch time) and whether you come back to watch more (retention).
Step 2 - "Down the rabbit hole"
If you stay, the app steers you towards more similar videos. It steers you towards videos you might like so that you keep scrolling. This is how TikTok leads you down a rabbit hole.
The last step is split up to target Content Consumers and Content Creators:
For consumers:
Step 3 - Engagement
In addition to analyzing how you watch videos, TikTok evaluates how you engage with the app. It looks at likes, comments and play-throughs and steers you towards similar content. In this way, the app uses users' "viewing habits" to promote videos it believes are relevant and entertaining to you. This is in addition to displaying paid content in the form of ads.
For content creators:
Step 3 - Content assessment
When you post videos yourself, the algorithm will analyze how the videos you post are doing based on the parameters discussed, including watch time and engagement. In addition, the content you create is assessed based on how frequently you post videos. All the parameters are factors that together measure how your videos contribute to keeping people on the app for longer. The highest scoring content is shown to the most people.
In summary
TikTok's goal is to increase the proportion of active users, and the algorithm that ensures this evaluates what types of content different users engage with in different ways. The content that is engaging is shown most often. This is useful to be aware of, firstly as content creators, because we gain insight into how to use the algorithm to our advantage. Secondly, as users, it's good to have an understanding of how our user habits affect and are affected by the algorithm.
