What I Learnt about TikTok’s Recommendation Algorithm Through Creating Content

TikTok’s Core Competency and Business Model

TikTok recommends content to users through an in-house recommendation algorithm. It tries to predict what someone likes to watch based on the user behavior. The aim is to provide more customized content and greater user satisfaction.

This is different from Medium blog where human curators and publication editors do the main heavy lifting for content selection.

This is also different from YouTube’s recommendation algorithm. As YouTube revenue is advertisement driven, it prioritizes contents with advertisement. As TikTok is largely Pre-Revenue, it is still trying to guess what the users will like and trying to push the most attractive and customized content to them.

What the algorithm prefers

• Trending content is preferred. It has its own community of fashion, value and vocabulary and therefore the trending content often can be different from hot Google searches.
• Recently uploaded Content is preferred. Not uncommon for content platforms. It can be measured in hours, not days. If you can targeting the peak period at night when users finished school or work, having the videos published in the morning of the same day is probably a little early.
• Prioritizes the use of hashtags. Hashtags help in effective categorization of content. Creators can…