X reveals what makes posts go viral, and likes are not the biggest factor

If you regularly post on X, you may think likes are the clearest sign that your content is performing well. But this might not be the case. 

X says post recommendations play a major role in engagement. It has made key parts of the source code behind its recommendation algorithm public, giving you a closer look at how posts are selected and ranked on the “For You” feed. At the centre of the system is Phoenix, which predicts how likely you are to interact with a post and uses those predictions to calculate its ranking.

What can give your X post a bigger boost?

The numbers reveal a significant difference between engagement types. If Phoenix predicts that you will share a post by copying its URL, that action carries an 80-times weight relative to a like, or roughly 40 times when compared with the like’s positive score of 0.5.

Must Read: SpaceX buys Cursor for $60 billion: Elon Musk’s biggest AI bet yet to take on OpenAI, Anthropic

Replies, quote posts and sharing through direct messages each have a weight of 5, while a predicted follow has a weight of 8. A repost has a weight of 1, compared with 0.5 for a like.

So, if you want greater visibility, simply collecting likes may not be enough. Shares, replies, quotes and follows can have a stronger influence on the ranking score.

Negative reactions can reduce visibility

Phoenix also factors in actions that signal users do not want to see certain content. A predicted report carries a negative weight of -234, while mute is weighted around -59.

Must Read: AI may reshape Indian jobs, but physical work could offer some protection, Goldman Sachs says

“Not interested” and blocking carry negative weights of -43 and -31, respectively.

Compared with the 0.5 weight assigned to a like, these negative signals can therefore have a much larger impact on a post’s ranking.

X tests ‘Under the Hood’

X has also introduced Under the Hood, a pilot that lets selected users see aggregate information about labels that may affect the visibility of their posts or accounts.

The feature is currently limited to a randomised group, with wider availability depending on user feedback.

For you as a creator, the takeaway is simple - a like is only one of many signals X considers when deciding what appears in your For You feed.