
How 360Brew Is Rewriting Visibility, Value, and Survival for People Who Make a Living From Thought

In the 360Brew foundation model
Average LinkedIn views year-over-year
Weight of a Save vs. a Like
To build recognised topic authority

360Brew is not a routine algorithm update. It is the largest structural change to professional content distribution in LinkedIn's history: a 150-billion-parameter foundation model replacing thousands of specialised ranking systems with a single AI engine.
LinkedIn has become an authority market. The scarce asset is no longer attention. It is credible legibility: the ability to appear coherent, useful, and professionally aligned enough for an AI-mediated system to trust you with distribution.

Most commentary treats 360Brew as a tactical event. That reading is too small.
LinkedIn has redesigned visibility around a richer interpretation of relevance, intent, and credibility, using a large language model to compress a professional identity, a piece of content, and a reader's prior behaviour into a single ranking judgement.

Compensate for weak substance with posting frequency, engagement tricks, network games, or formulaic audience capture.
Does this person reliably produce content that fits who they claim to be and delivers value that readers demonstrate they want to retain?
Van der Blom's research: drawn from 400,000+ LinkedIn posts, confirms: creators posting 5–7 times per week see –27% average reach per post.
Those at 3–5 posts per week recovered reach 38–42% faster after 360Brew's rollout.
Volume is no longer a hedge. It is a tax.

A fine-tuned LLaMA-3 dual encoder generates 3,072-dimensional embeddings by processing full profile text and content, grouping posts by interest clusters.
A sequential transformer processes a member's most recent 1,000 feed impressions as a causal sequence, learning behavioural patterns to personalise distribution.
360Brew is a 150-billion-parameter decoder-only foundation model developed by LinkedIn's Foundation AI Technologies (FAIT) team, replacing hundreds of separate task-specific ranking models with a unified architecture.

–56%
–29%
+31%
+42%
The algorithm has shifted from the social graph (who you know) to the interest graph (what you know about).
This is the most consequential structural change for creators.

Year-over-year across the platform
Across all content types
For average creators
These numbers describe an average: including the enormous volume of low-quality, generic, and engagement-bait content 360Brew was specifically designed to suppress.
Creators who adapted are telling a different story.

"Dwell time has become the silent kingmaker."
Posts generating 61+ seconds average 15.6% engagement rates.
Posts under 3 seconds average 1.2%. Dwell time is weighted 2.8x heavier than likes.

Creators at this frequency recovered reach 38–42% faster after 360Brew's rollout.
Posting twice within 24 hours cannibalises reach by approximately 20%.
Only 5% of posts that underperform in the first 60 minutes ever recover.


"A post might reach 5,000 people this week. An article reaches 500 people per week for years."
Articles underperform for immediate organic reach. But they are the only LinkedIn content format indexed by Google, meaning a well-crafted article continues generating traffic and authority signals for months or years, making them a long-term strategic asset, not a vanity format.

A specialist with 2,000 followers posting consistently about one topic can now outreach a generalist with 50,000 followers.
The algorithm has shifted from a social graph to an interest graph. Posts from random peers in the same topic are up 42%, while off-topic posts from connections are suppressed.
"AI made what was valuable terribly easy. The effort to start and a reasonable strategy, both are now worthless." — Ruben Hassid
When anyone can generate competent-looking content, the differentiator is no longer competence. It is taste, consistency, and the willingness to say something specific enough to be wrong about.

360Brew builds a pattern profile based on topic consistency, language use, and engagement history over a sustained period. Once established, topic authority delivers up to 78% higher distribution. LinkedIn's SSI scores above 75 correlate with 78% more profile views. Strategic pivots must be gradual transitions, not sudden shifts.

The Causal LLM retrieval system processes full profile text and content together to generate a 3,072-dimensional embedding.
If your headline says "B2B Sales Leader" but you post about personal development, the system detects the mismatch and restricts distribution.
Profile architecture is no longer a branding exercise. It is an editorial strategy.
Headline, About section, experience language, featured items, and recent posts must tell the same professional story.

Posts generating 61+ seconds of sustained attention average 15.6% engagement rates vs. 1.2% for posts under 3 seconds.
Posts of 800–1,000 words outperform short posts. Extended dwell time is weighted 2.8x heavier than likes.
Posts with sustained engagement 24–72 hours after publishing perform 4–6x better.
"The hook gets you the click. The substance gets you the save. And saves are worth 5–10x more than the like you would have gotten from a clever opening alone." - Sarra Richmond

The optimal strategy is layering all three. A post gets 5,000 impressions this week. An article gets 500 per week for three years. The newsletter guarantees you reach 100% of subscribers every time, regardless of algorithm changes.

Educational reference-grade content, frameworks, benchmarks, tools, narrow thesis + lived expertise, field reports with data, contrarian analysis
Broad motivational opinion, AI-polished generic content, company announcements
The goal is no longer to produce something someone will like. The goal is to produce something someone will use: a framework they will bookmark, an analysis they will forward, a benchmark they will reference in a meeting.

Ruben Hassid built a 225,000-subscriber newsletter and gained 344,258 LinkedIn followers in 2025 alone, without paid advertising.
His central thesis: AI has commoditised competence. Producing grammatically correct, strategically structured content is no longer a differentiator. It is table stakes.
Unedited AI-generated content now experiences 5–7x less engagement than human-crafted content.
The practical lesson: use AI for research, outlining, and ideation, but the voice, specificity, and human fingerprint must be unmistakably present.

Does your content match what your profile claims you do?
The system detects engagement pods via comment velocity and timing patterns.
Lexical diversity and originality are actively evaluated.
Your author track record is part of every post's distribution decision.
Human comments of 15+ words provide a 2.5x reach boost and trigger three distinct distribution mechanisms. Your engagement strategy is now part of your content strategy.


Many creators say the new environment rewards authenticity. What it may actually reward is disciplined legibility.
True authenticity includes contradiction, curiosity, and drift. Algorithmically rewarded authenticity is a cleaner performance: stable topics, predictable language, strategic consistency.
The practical resolution: "I help B2B founders think more clearly about growth" can encompass strategy, hiring, culture, and AI without triggering semantic dissonance.
The trick is not narrowing your mind. It is narrowing your positioning.

Maintain 80% consistency within your pillars over a 90-day period.
Would a machine reading only your profile correctly predict your next 10 posts?
Posting 5–7x/week produces reach decline. Twice in 24 hours cannibalises your own content.
Ask: "Would someone bookmark this? Forward it to a colleague? Reference it in a meeting?"
Short paragraphs, deliberate white space, clear logical progression that rewards reading to the end.
Thoughtful 15+ word comments drive 55% increase in profile views and 20% lift in your own post reach.
Posts for daily visibility. Articles for SEO. Newsletters for direct reach. One idea, four channels.


Organic company page reach has collapsed to 1.6%. The better model is a distributed authority system in which employees, founders, and subject experts publish within a coherent strategic frame.
Align their profiles as semantic source code.
Connect individual expertise to company value propositions.
Not likes and impressions. Activate early commenting on strategic posts.

LinkedIn pursued a foundation-model architecture. The directional shift is hard to dismiss. The original arXiv paper was withdrawn, but the technical direction is confirmed.
Exact signal weights are inferred, not stated. Deployment estimates range from 40% to 100%. Too many people speak with mechanistic confidence about a system they do not directly observe.
The prudent conclusion: adapt as if quality, clarity, coherence, and usefulness are now the dominant constraints: because even if the algorithm has not fully changed, the market's expectations have.

LinkedIn is evolving from a professional networking platform into a system that increasingly decides whose expertise is machine-legible enough to deserve circulation. The old game of broad visibility, thin thought leadership, and engagement theatrics is becoming harder to sustain.
The new game favours those who can make their expertise unmistakable, their writing useful, their identity coherent, and their contribution worth returning to.
"The machine is listening. Make sure it can hear you clearly." - Sarra Richmond

Choose 2–3 durable topics and commit to 90 days of consistent, on-topic publishing.
Treat your headline, About section, and featured items as semantic source code for the algorithm.
Every piece of content should answer: would someone bookmark this and return to it?
Posts for daily visibility. Articles for SEO longevity. Newsletters to own your audience outright.


Forget generic LinkedIn advice. Sarra doesn't just write posts; she reverse-engineers how leaders truly sound, then crafts an entire content system around that voice.
When your LinkedIn presence demands precision, authenticity, and undeniable impact, there's only one choice.

[1] LinkedIn FAIT Team, 360Brew: A Decoder-only Foundation Model for Personalized Ranking and Recommendation — arxiv.org/abs/2501.16450
[2] AuthoredUp, LinkedIn 360Brew: What Actually Changed — authoredup.com/blog/linkedin-360brew
[3] Richard van der Blom, Algorithm Insights Report 2025, analysis of 400,000+ LinkedIn posts
[4] Ritchie Pettauer, LinkedIn 360Brew: The New Physics of Visibility — pettauer.net
[5–16] Cause Fokus · Trust Insights / HyperClapper · Dean Seddon · BotDog · Phil Szomszor · Chris Essey · Meet Edgar · ConnectSafely · UpGrowth · Koka Sexton · Meet Lea / Cleverly · Anas Hidaoui · Ryan Smith · Ruben Hassid — ruben.substack.com
Sarra Richmond is the founder of The Ghost, a ghostwriting and content strategy practice that helps founders, consultants, and professionals build authority through LinkedIn.

LinkedIn's Authority Machine