Recommendation algorithm
Also known as: the algorithm, feed ranking
The ranking that decides who sees a video, driven mainly by how people behave in the first minutes after it is published.
Updated
What it means
The ranking is a prediction rather than a judgement. A platform estimates whether a given viewer will watch, and everything it uses to decide that comes from how other people behaved first.
Two signals dominate everywhere. Audience retention on the people who started watching, and Engagement rate among the people who finished, which together describe whether the video held anybody.
It is not a single system. Every platform ranks differently, and the same video can be pushed on one and ignored on another for reasons that neither of them ever explains to anybody.
How it works in practice
The opening is the only part under real control. A strong Hook decides retention, retention decides ranking, and nothing later in the edit recovers an audience that left in the first moments.
Consistency does more than any individual video. A channel publishing regularly accumulates a picture of who its audience is, which is what makes a new upload reach anybody at all.
Posting at a time the audience is present helps because of the first hour. That window is weighted heavily, which is the real argument for Post scheduling rather than publishing when an edit finishes.
What to watch out for
Chasing it produces worse videos. Material shaped for a ranking rather than for an audience performs once and builds nothing, and the ranking adjusts faster than anybody can follow it.
A single video proves nothing either way. The variance between uploads is large enough that a success and a failure can be the same piece of work published a week apart.
Reach that collapses is read as punishment. A quiet upload is usually a weak opening rather than a Shadowban, and the two are almost impossible to tell apart from outside the platform.
How RawClip handles it
Nothing in the analysis targets a platform ranking, since it works from what is in the footage. A sequence with its strongest moment early and no dead passages is what these systems reward, which happens as a consequence of the edit rather than as an aim of it.