The LinkedIn algorithm now uses AI language models to match posts with each member’s professional profile, not only with their network. On 9 November 2026, LinkedIn engineers explained a system called Profile Relevance that reads your headline, skills, work history, education and country, then recommends posts from outside your network. In our view, this makes a clear profile and a focused topic more useful for creators than chasing connections.
Key takeaways
- New on 9 November 2026: LinkedIn described Profile Relevance, an LLM-based system that recommends posts from outside your network based on professional context (LinkedIn Engineering).
- What it reads: profile highlights (headline, summary, skills), work history and education, plus post text, titles and transcripts. LinkedIn says it excludes name and gender.
- Local matters: LinkedIn adds the member’s and author’s country to the model, so local professional posts are more likely to surface.
- Early proof needed: a post from this source needs at least one like, comment or share from an unconnected member before it is recommended.
- LinkedIn’s own test results: hide rate down 18.8%, viral rate up 0.94%, and viral engagement up 17.7% for members with little activity history.
What changed in the LinkedIn algorithm in November 2026?
On 9 November 2026, LinkedIn’s engineering team published a detailed post about Profile Relevance, a retrieval system that finds posts from people you are not connected to that could appear in your feed. LinkedIn describes it as a key source for this content, alongside others. LinkedIn calls these posts Unconnected Content. Social Media Today covered the update the same day.
LinkedIn explains the problem plainly: traditional recommendation systems lean on network proximity and engagement, which makes content beyond your immediate network harder to discover, “particularly for creators who are still building their audience.” Profile Relevance is LinkedIn’s answer. It uses LLM-generated embeddings (mathematical representations of meaning) to compare a member’s professional context with the content of a post.
This builds on a bigger change LinkedIn announced on 12 March 2026: a new feed ranking system powered by LLMs and GPUs that, in LinkedIn’s words, “better understands what a post is actually about and how it relates to a member’s evolving interests and career goals.” LinkedIn says the feed serves more than 1.3 billion professionals.
How does the LinkedIn feed decide what you see?
Based on LinkedIn’s two engineering posts, the feed works in two broad stages. First, retrieval picks a set of candidate posts from millions. Then ranking orders those candidates for each member.
Stage 1: Retrieval (finding candidate posts)
LinkedIn says it replaced several separate retrieval sources, such as trending posts, collaborative filtering and topic-based systems, with a unified system built on LLM embeddings. For posts, LinkedIn’s prompts include the format, author information (name, headline, company, industry), engagement counts, article metadata and the post text. For members, they include profile information, skills, work history, education and a list of posts the member engaged with.
One detail matters for creators: LinkedIn converts engagement counts into percentile buckets (for example, a post in the 71st percentile of views) so the model can understand popularity. In LinkedIn’s own words, popularity is “such a strong relevance signal.”
Stage 2: Ranking (ordering your feed)
For ranking, LinkedIn built what it calls a Generative Recommender model. Instead of scoring each post in isolation, it processes more than a thousand of a member’s past interactions as a sequence, to understand where their interests are heading. LinkedIn lists the actions it reads as long dwells, likes, comments and shares, and it separates passive actions (click, skip, long dwell) from active ones (like, comment, share). LinkedIn says the ranking model relies on professional signals and engagement patterns, “never demographic attributes.”
What is LinkedIn Profile Relevance and what does it read?
Profile Relevance starts with a “member prompt”: a structured summary of the professional information in your public profile. According to LinkedIn, it covers:
- Profile highlights: headline, summary and key skills.
- Work history: companies, titles, durations and descriptions of past and current roles.
- Education history: schools, fields of study and degrees.
LinkedIn says it does not include personal attributes such as your name or gender. On the content side, it generates embeddings from post text, titles and transcripts, with author metadata such as location when available.
LinkedIn also shared two design decisions that every marketer on the platform should know:
- Country is added on purpose. LinkedIn found that its AI summaries sometimes lost geographic cues, so it appended member and author country names to the prompts, based on the location in each profile. The result, LinkedIn says, is that members are more likely to see posts that reflect their local professional context.
- An engagement threshold. Topic match alone wasn’t enough to compete in the feed. So a post from Profile Relevance must first receive a single like, comment or share from an unconnected member before it is recommended.
To judge quality, LinkedIn built an AI evaluation system (AutoEval) that checks how well a post matches a member across topic, location, job title, company, industry, seniority and education.
What results did LinkedIn report?
These are LinkedIn’s own figures from online experiments with the final version of Profile Relevance, compared with its previous retrieval source:
| Metric (LinkedIn test) | Reported change | What it suggests |
|---|---|---|
| Feed updates viewed | +0.54% | Members viewed slightly more posts |
| Viral rate (likes, comments, shares) | +0.94% | Members took more like, comment or share actions |
| Hide rate | −18.8% | Fewer posts were hidden, so the feed felt less noisy |
| Viral engagement for cold-start members | +17.7% | Members with little history engaged more |
| Daily active users (cold-start members) | +1.8% | Newer or less active members came back more |
In our view, the hide-rate drop is the most important number for brands. It suggests fewer off-topic posts are reaching people, so focused, specific content has a better chance than generic posts.
How do you increase reach on LinkedIn under the new algorithm?
LinkedIn has not published a checklist for creators. The steps below are our practical reading of what LinkedIn says its systems use:
| What LinkedIn says it uses | What you can do |
|---|---|
| Headline, summary and key skills | Write a headline that names your field and audience, and list skills that match the topics you post about. |
| Work history and education | Add descriptions to current and past roles instead of titles only. |
| Country from your profile location | Set your real location. If your market is Lebanon, Saudi Arabia or the UAE, local examples help your posts match local readers. |
| Post text, titles and transcripts | State the topic clearly in the first lines, and add captions or a written summary to video posts. |
| Engagement history and percentile popularity | Post on one or two professional topics consistently, so your audience’s engagement teaches the model what you are about. |
| At least one like, comment or share from a non-connection | Write posts that invite useful replies from people in your field, not only your friends. |
Note: in Profile Relevance, LinkedIn uses profile fields to understand you as a reader. On the author side, Profile Relevance names your location plus the post’s text, titles and transcripts; LinkedIn’s March 2026 retrieval system also includes the author’s name, headline, company and industry. Completing the rest of your profile is our recommendation, not a reach factor LinkedIn has stated.
A few more steps, in order:
- Audit your profile first. Read your headline, About section and skills as if you were the AI. Would it know which topics you cover?
- Pick your lane. Choose one or two topics (for example, Meta Ads for e-commerce in the Gulf) and stay with them for a few months.
- Write for clarity, not tricks. LinkedIn’s system reads meaning, so clear, specific posts are easier to match than vague motivational lines.
- Measure in-network vs out-of-network reach. Use the new analytics as they reach your account (see below).
- Repurpose the best posts. Turn a strong post into a carousel or short video with captions, using a simple AI content calendar.
Which LinkedIn analytics show out-of-network reach?
Social Media Today reported on 3 November 2026 that LinkedIn is adding new creator analytics: a post performance table with impressions, views and watch time; the followers gained from each post; saves and sends; and In and Out of Network metrics that show whether views came from connections or non-connections. No rollout timeline was given. Remember the difference between views and impressions when you read these numbers.
Why does this matter for marketers in Lebanon and the Gulf?
LinkedIn’s country signal is good news for regional professionals. According to DataReportal’s Digital 2027 reports, published on 5 November 2026, LinkedIn had 15.0 million members in Saudi Arabia and 1.60 million in Lebanon in late 2026. DataReportal notes these are registered members, not active users.
In our view, this means a Beirut agency or a Riyadh B2B brand that posts about local market topics, with a clear profile and real location, has a better chance of reaching the right local decision makers than before. If you also use LinkedIn for lead generation, see our guide to getting more leads from social media and the LinkedIn QR code for offline events. For a comparison with another AI-ranked system, read how Meta Andromeda picks ads.
What to do this week
- Rewrite your headline so it names your field, your audience and your country or region.
- Add 5–10 skills that match your post topics, and add descriptions to your current role.
- Check that your profile location is correct.
- Plan four posts on one topic, each with a clear first line and a question that invites expert replies.
- Add captions or a text summary to every video.
- Track in-network vs out-of-network views when the new metrics appear in your analytics.
Frequently asked questions
How does the LinkedIn algorithm work in 2026?
LinkedIn says its feed now uses large language models (LLMs) to understand what a post is about and how it relates to each member’s interests. It looks at profile information such as industry, experience, skills and geography, plus how a member has engaged with posts over time. A ranking model then reads more than a thousand of a member’s past interactions to predict what they will find valuable.
What is LinkedIn Profile Relevance?
Profile Relevance is a retrieval system LinkedIn described on 9 November 2026. It turns the professional information in a member’s public profile into a structured summary, which an AI model converts into embeddings and matches with posts from people the member is not connected to. LinkedIn says it does not include personal attributes such as name or gender.
Can my LinkedIn posts reach people who don’t follow me?
Yes. LinkedIn calls these posts Unconnected Content, and Profile Relevance is built to recommend them to members whose professional context matches. LinkedIn says a post from this source needs at least one like, comment or share from an unconnected member before it is recommended.
Does my LinkedIn profile affect the reach of my posts?
LinkedIn says the profile information you choose to share, including your headline, summary, skills, work history, education and location, is used to match you with relevant posts. In our view, a clear profile also helps the system understand which professional topics your own posts belong to, but LinkedIn has not published a creator-side formula.
Does location matter for the LinkedIn algorithm?
Yes, for this retrieval source. LinkedIn added member and author country names to its prompts, based on the location in each profile, so members are more likely to see posts that reflect their local professional context, such as regional industry discussions.
Is LinkedIn’s new feed better for new creators?
LinkedIn says Profile Relevance helps creators’ compelling posts reach relevant audiences beyond their immediate network, but it published no creator-specific figures. Its reported cold-start gains (+17.7% viral engagement and +1.8% daily active users) are for members with little activity history, meaning readers, not creators.
How can I see if my posts reach people outside my network?
Social Media Today reported on 3 November 2026 that LinkedIn is adding In and Out of Network metrics to creator analytics, along with a post performance table, followers gained from a post, and saves and sends. No rollout timeline was given, so you may not see all of them yet.
Sources
- LinkedIn Engineering: Profile Relevance — Boosting Feed Relevance with LLMs at LinkedIn (9 November 2026)
- LinkedIn Engineering: Engineering the next generation of LinkedIn’s Feed (12 March 2026)
- Social Media Today: LinkedIn explains its improved content recommendation system (9 November 2026)
- Social Media Today: LinkedIn details new creator insights options (3 November 2026)
- DataReportal: Digital 2027 — Saudi Arabia (5 November 2026)
- DataReportal: Digital 2027 — Lebanon (5 November 2026)
As of 11 November 2026. LinkedIn changes its feed systems often and rolls out analytics gradually, so what you see may differ.
Learn to grow on LinkedIn and Meta with AI
Gradezilla’s Digital Marketing & AI Course covers content strategy with AI tools, ads on Meta, LinkedIn, TikTok, Snapchat and Google, and SEO with GEO, in 75+ hours of training. Live online sessions with Bashar Amin run on Tuesdays and Thursdays, and every session is recorded.
Written by the Gradezilla team · Reviewed by Bashar Amin