The platform listed various positive and negative signals that define post reach, and Business Insider reporter Truman Dickerson dug into those in order to offer a better sense of what users should and should not be posting in the app.
According to Dickerson, X’s Phoenix algorithm aims to display posts to each user based on predictions around what they’ll engage with, based on past behavior.
This audio is auto-generated. Please let us know if you have feedback.
The platform listed various positive and negative signals that define post reach, and Business Insider reporter Truman Dickerson dug into those in order to offer a better sense of what users should and should not be posting in the app.
According to Dickerson, X’s Phoenix algorithm aims to display posts to each user based on predictions around what they’ll engage with, based on past behavior.
In terms of weighting, the Phoenix algorithm is primarily focused on driving shares, with follows and likes now being less of an indicator of perceived value to X users.
Dickerson said that according to X’s source code:
A share in which the user copies the post URL is worth around 20 points, which equates to 40 times the value of a like.
Replies, quotes and shares via internal DMs are weighted at around 5 points per action.
A follow is weighted at around 4 points, which is 8 times that of a like.
A repost is weighted at around 1 point.
A like is valued at 0.5 within X’s engagement metrics, and is one of the lower indicators of positive user engagement.
Based on these numeric indicators, and in combination with X’s other mathematical factors (who a user follows, recency, etc.), the X algorithm comes up with a ranking number for each of the posts that could be displayed to every user every time they log into the app. This then dictates what they see in their feed.
X wants to incentivize re-sharing of X posts in other formats, via users copying and re-sharing post links. In other words, the best way to maximize X post reach is to optimize for re-sharing.
In terms of negative indicators, Dickerson said the Phoenix algorithm will penalize posts that it predicts people will report, mute, or which will prompt users to tap the ‘not interested’ button.
Interestingly, the numerical penalty for these actions is far more significant than if a user blocks somebody.
A report is weighted at -234
A mute is weighted at -59
Not interested is weighted at -43
A block is weighted at -31
So while posting divisive comments to spur more response still works, it’s a risky bet, because if users start reporting or blocking an account, it’s going to be hit with significant reach penalties.
However, the algorithm is personalized to each user. These penalties will only apply to that viewer and won’t reflect broader reach and engagement. Reports and blocks do carry a negative signal, but X has seemingly lessened this as an impact for other users, making it more of an isolated impact for each user.
The insights here don’t reveal any major change in approach from the platform, though they do provide some pointers on what X is doing to show users more of what they want to see, and less of what they don’t.
In formulating a strategy based on this, the main focus for brands should be audience research, and understanding what target audiences are more likely to respond to and share with their connections, both on and off the platform.
