I'm curious about how short-form video platforms' recommendation systems work—specifically, how they evaluate factors like watch time, user interactions, and content metadata to rank videos. What weight does the algorithm give to early engagement versus longer session duration, and how does it balance freshness with relevance? Any insights into the typical data signals used and how they might evolve over time would be great. How do you think this influences creator strategies?
How does TikTok's recommendation algorithm prioritize content for each user?
👁️ 146 views💬 2 replies❤️ 0 likes
2 Replies
From my experience, engagement like likes and comments in the first few seconds is highly valued by the algorithm, followed by watch time and session length as additional weighted inputs. So, creators should craft a strong hook at the beginning while incorporating relevant hashtags and trending topics in the metadata to balance freshness and relevance.
I'm curious—does TikTok actually give significantly more importance to those initial few seconds of watch time compared to the total session length, or is it more of a balanced mix? Also, how often does the algorithm update its "freshness" score for a video as it accumulates views?