Under The Hood X Report Analyzer
Engagement Signal Guide

How Reposts Affect X Reach

Reposts are explicitly modeled by the X recommendation system. Phoenix predicts the probability that a viewer will repost a candidate, and the ranking system incorporates that prediction into the overall score.

(GitHub)

Reposts Are Different from Likes

A like generally represents personal approval or interest. A repost can additionally distribute the post to another audience. The recommendation model nevertheless doesn't simply say: “This post received a repost, add 10 points.” Instead, it predicts the probability that the viewer will repost the candidate.

Post
User context
Phoenix
P(Repost)
Repost prediction × weight
Overall ranking score

Why Repost Prediction Can Matter

Suppose a post is particularly relevant to a user who frequently shares content about a specific subject. Phoenix can learn from that user's engagement history when predicting future behavior. The current X architecture explicitly describes recent viewer engagements as a major model input. (GitHub)

Post + User A → One repost probability
Post + User B → Another repost probability

Don't Confuse Repost Count with Repost Prediction

This is one of the most important distinctions. Repost count is an observed metric. Repost probability is a model prediction. The ranking system uses the latter when evaluating candidates.

Does a Repost Guarantee More Reach?

No. A repost can contribute to the ranking model, but recommendation also depends on candidate retrieval, eligibility, other predicted actions, ranking, author diversity, out-of-network adjustments, selection, and post-selection filtering. The current open-source architecture documents all of these stages. (GitHub)

Under The Hood Takeaway

Reposts are a modeled positive action, but there is no fixed “one repost = X reach” formula.

Official Open-Source References