If you noticed your LinkedIn feed showing you posts from complete strangers talking about your exact job function, you are looking at the platform’s new Profile Relevance Model.
Historically, social feeds prioritized personal networks: if John followed Sarah, John saw Sarah’s posts. Then came the engagement-bait era, where viral stories about crying CEOs got pushed everywhere regardless of who you were.
LinkedIn's new model ditches both.
By applying LLM-generated semantic embeddings, the system reads your personal headline, current job description, past work history, and recent reading behavior. It then matches those profile embeddings against the actual thematic core of new posts—not just broad hashtags.
To rank under this system, your publishing strategy needs to change:
- Pick a Tight Lane: You cannot talk about crypto on Monday, fitness on Wednesday, and cloud architecture on Friday. When you post consistently within a distinct domain, the model classifies your profile as an authority node.
- Write for Substance, Not Engagement Tricks: Asking "Agree or disagree?" does not trigger the semantic relevancy check. Clear, actionable breakdowns that retain reader attention (dwell time) tell the algorithm your piece solved a problem for that specific job title.
Did you like our blog?
Comments (0)
All comments are reviewed and approved by an admin before appearing publicly.
No comments yet - be the first to share your thoughts.
