X Updates Recommendation Algorithm to Prioritize Dwell Time and Completion Rates

X has updated its home timeline recommendation algorithm to introduce a new AI ranking model that better predicts user dwell time and prioritizes posts with high completion rates. Additionally, the platform improved its retrieval stage by using vector-based matching to surface content from unfollowed accounts. This change shifts content visibility on social media toward deeper user engagement metrics rather than superficial clicks, potentially changing how creators and media outlets optimize their posts. It aligns with broader industry trends where platforms leverage AI and vector search to curate personalized content feeds. The updated ranking model builds upon existing post completion rate metrics by optimizing the system's predictive accuracy. Meanwhile, the vector-based retrieval system operates independently from the separate system that supplies content from followed accounts.