Rapidly Viral

Virality, explained.

Platform mechanics

How a For You Feed Decides What You See

Launch library · evergreen read

Photo: 7413b Image of Curatorial Object (54689337551) by San Diego Air & Space Museum Archives (Public domain), via Openverse

A For You style feed is built to predict, moment to moment, which piece of content is most likely to keep a particular viewer watching. It does this by comparing your past behaviour against the behaviour of people who reacted similarly to earlier content, building a kind of running profile that updates continuously as new signals arrive from whatever you do next.

Instead of showing everything from accounts you follow in order, the feed is constantly re ranking a much larger pool of possible content, testing small amounts of it with you before deciding whether to show more like it. Your scroll speed and watch time matter more than any single like, since those signals arrive constantly rather than only occasionally.

This is why two people can open the same app and see almost entirely different feeds. The system is not showing a shared timeline, it is running a continuous, personalised experiment on each viewer, adjusting its guesses with every video you linger on or skip, which is part of why describing a feed as unfair or biased misses how individually tailored the whole system actually is.

Back to the library

Share

Sharing opens the network in a new tab. No tracking scripts are loaded on this page.

Printed from Rapidly Viral. Sources for this article are listed at the end of the page.