LinkedIn Just Rewrote the Rules. Most Brands Haven't Noticed Yet.
In late 2024, LinkedIn quietly retired the system that had governed content distribution on the platform for years. No press release. No announcement post. Just a gradual, then sudden, collapse in reach for the majority of creators and brands — and a corresponding surge for a small cohort who happened to be doing exactly what the new system rewards.
The replacement is called 360Brew. It is a 150-billion-parameter, decoder-only foundation model trained on LinkedIn's own professional data, described in a research paper published on arXiv in January 2025 by LinkedIn's engineering team. It replaced not one algorithm but an ensemble of thousands of specialised ranking models — one for feed, one for job recommendations, one for people you may know, and so on — with a single adaptive AI that handles all of them simultaneously.
The implications for B2B marketers are significant, and most brands are still operating under the assumptions of the old system. This guide explains what 360Brew actually is, what it rewards, what it penalises, and the specific content strategy adjustments that restore and grow reach in 2026.
The Scale of the Shift: What the Data Shows
AuthoredUp's analysis of more than 3 million LinkedIn posts tracked the before-and-after with precision. Between June 2024 and May 2025, median impressions per post fell from 1,211 to 636 — a 47% decline in 12 months. That is not a seasonal dip. It is a structural redistribution of attention.
The distribution of that reach is the more important story. The top 2% of creators went from capturing 15% of all feed share to 31%. Regular users dropped from 57% to 28%. The algorithm did not simply reduce overall reach — it concentrated it. The question for every B2B marketer is which side of that concentration you are on.
| Metric | Before 360Brew (Jun 2024) | After 360Brew (May 2025) |
|---|---|---|
| Median impressions per post | 1,211 | 636 |
| Top 2% creators' feed share | 15% | 31% |
| Regular users' feed share | 57% | 28% |
| Company page follower reach | ~5% | 1.6% |
| Personal vs company page reach | ~3x advantage | 561% advantage |
Source: AuthoredUp analysis of 3M+ LinkedIn posts, June 2024 vs May 2025
How 360Brew Actually Works
The old LinkedIn algorithm was a factory with thousands of specialised machines. Each one had a specific job: count reactions, track keyword frequency, measure connection proximity, score recency. Engineers spent years hand-tuning each component. The system was powerful but brittle — it could be gamed because it measured proxies rather than substance.
360Brew is architecturally different. It is a decoder-only transformer — the same class of model as GPT-4 — but trained specifically on LinkedIn's professional data. Rather than counting signals, it reads your content like an editor. It understands semantic relationships between topics. It can infer that "Gong's revenue intelligence platform" and "Salesforce CRM integration" are related concepts without you ever using those exact terms together.
The feed ranking process now works in two stages. Stage one narrows millions of posts to a few hundred candidates based on your network and past behaviour — this part is largely unchanged. Stage two is where 360Brew takes over. It reads each candidate post, evaluates the author's profile for expertise alignment, assesses the quality of engagement the post has received, and scores relevance to the specific reader. The research paper describes this as "many-shot in-context learning" — the model creates a temporary personalised version of itself based on your last two to three months of activity, then applies that understanding to rank content for you specifically.
The New Engagement Hierarchy
The most consequential change under 360Brew is not what it reads — it is what it counts. The engagement hierarchy has been completely reordered.
| Signal | Algorithmic Weight | Why It Matters |
|---|---|---|
| Save | 5× a like | Signals lasting value — someone wants to return to it |
| Meaningful comment (15+ words) | Equivalent to 15 likes | Signals the post sparked genuine thinking |
| DM share | Second strongest signal overall | Private endorsement to a specific person |
| Dwell time 61+ seconds | 15.6% engagement rate | vs 1.2% for posts under 3 seconds dwell |
| Delayed engagement (24–72 hrs) | 4–6× better in Suggested | Signals evergreen value, not just novelty |
| Like | Baseline | Surface-level; still counted but weighted low |
| High likes + low dwell time | Negative signal | Flagged as engagement bait |
Source: AuthoredUp signal weight analysis; meet-lea.com dwell time research
The logic is straightforward once you see it. A like takes one tap and zero thought. A save means someone found your content valuable enough to return to it. A 200-word comment means someone read your post carefully enough to form a response. 360Brew is built to distinguish between these, and it does so with considerable accuracy.
Profile-Content Alignment: The Requirement Most Brands Miss
Here is the part of 360Brew that most B2B marketers have not yet absorbed: the algorithm checks whether your content matches your stated expertise before distributing it.
360Brew reads your profile — your headline, About section, experience, and skills — as text, then checks whether your posts fall within the same semantic territory. If your profile says "Head of Growth, B2B SaaS" but your recent posts cover crypto, travel photography, and political commentary alongside occasional marketing content, the algorithm cannot classify what you are actually expert in. So your content does not get matched to relevant audiences. It does not get penalised. It simply does not get distributed.
The research paper confirms this. The model uses profile data as a prior for understanding what expertise an author possesses, then evaluates each post against that prior. Topic-hopping resets the prior. It takes approximately 90 days of consistent, aligned posting for 360Brew to fully categorise you and begin optimising your distribution.
For B2B brands, this has a direct implication for employee advocacy programmes. The most effective LinkedIn strategy in 2026 is not a company page posting daily — it is three to five senior practitioners posting consistently within their defined expertise areas. The algorithm is designed to amplify exactly that pattern.
What Gets Penalised (And By How Much)
360Brew does not just reward good behaviour. It actively suppresses specific patterns that the old algorithm either ignored or inadvertently rewarded.
Engagement pods are detected with 97% accuracy. If caught, the account receives a shadow ban lasting 60 to 90 days. The model is trained on LinkedIn's full dataset of authentic versus inauthentic engagement patterns, and it has seen every pod format that exists. Short comments under 15 words from pod members carry reduced weight regardless of whether the pod is detected.
External links in the post body carry a 25 to 40% reach penalty. The workaround — placing the link in the first comment — is well-established and effective. If you are still putting URLs in the body of your posts, you are paying a significant distribution tax on every piece of content you publish.
AI-generated content receives 45% fewer interactions than human-written posts. This is not because LinkedIn can always detect AI writing — it is because AI-generated content tends to produce lower dwell time and fewer saves, which the algorithm interprets as low-value content regardless of origin.
Posting more than once per day triggers a reach penalty. The algorithm interprets high-frequency posting as a signal of volume-over-substance behaviour. The sweet spot is four to five posts per week for personal profiles.
Format Performance in 2026: The Numbers
Not all content formats are equal under 360Brew. The format performance data from 2026 is striking in how decisively it has shifted away from video — which received a significant algorithmic boost under the old system — toward formats that generate dwell time.
| Format | Reach Multiplier | Notes |
|---|---|---|
| Polls | 1.64× | Only 1.4% of content — massively underused |
| Carousels (documents) | 1.39× | 278% more engagement than video; 8–10 slides optimal |
| LinkedIn Newsletters | +48% reach | Google-indexed; lands in inbox; the sleeper weapon |
| Short vertical video (30–90s) | +80% reach | Up significantly; long video still underperforms |
| Text-only posts | 0.88× | Below average; weakest format in 2026 |
| Long-form video | 0.86× | Down 72% year-over-year |
Source: AuthoredUp format analysis, 2026
The carousel finding deserves particular attention. Carousels generate 596% more engagement than text-only posts and 278% more than video. The mechanism is dwell time — a 10-slide carousel requires genuine reading engagement. Each slide swipe is a passive signal of continued interest. The algorithm interprets this as high-quality content regardless of the reaction count.
LinkedIn Newsletters are the most underused high-leverage format on the platform. They reach your connections' inboxes directly, are indexed by Google, and carry a 48% reach advantage. If you are already writing long-form LinkedIn posts, converting them into a newsletter is the single highest-leverage format change you can make.
The Golden Window: Timing and Post Mechanics
The first 60 to 90 minutes after posting determine approximately 70% of a post's total reach. This has not changed under 360Brew — the initial velocity signal still matters. What has changed is that posts now have a second life.
Under the old algorithm, a post that did not gain traction in the first two hours was effectively dead. Under 360Brew, saves and meaningful comments received 24 to 72 hours after posting trigger a second distribution wave in LinkedIn's Suggested feed — and those delayed engagements perform four to six times better than initial ones. This means that evergreen, substantive content continues to accumulate reach long after the initial posting window.
The practical implication: reply to comments within 30 minutes of posting. AuthoredUp's data shows this generates 64% more total comment volume. And write posts long enough to require genuine reading — the 1,300 to 1,900 character sweet spot. Posts under 500 characters receive a 35% reach penalty as low-effort signals.
Seven Practical Adjustments for B2B Marketers
1. Audit your profile-content alignment before your next post. Read your LinkedIn headline and About section, then look at your last 20 posts. Do they fall within the same semantic territory? If not, 360Brew cannot classify your expertise and will suppress your distribution. Tighten your profile to match your actual content focus, or tighten your content to match your profile positioning.
2. Move all external links to the first comment. Every post with a URL in the body is paying a 25 to 40% reach tax. This is the single easiest fix with the most immediate impact. Write the post, publish it, then immediately add the first comment with the link and a one-line context note.
3. Build for saves, not likes. Before you publish, ask: is there something in this post that someone would want to return to? A framework, a data point, a decision matrix, a checklist. If the answer is no, the post will generate likes but not saves — and likes are now the weakest signal in the hierarchy.
4. Convert your best long-form posts into carousels. Take your highest-performing text posts from the last six months and rebuild them as 8 to 10-slide carousels. The format change alone will generate significantly more dwell time and reach for the same content.
5. Start a LinkedIn Newsletter. If you are writing substantive long-form content, a newsletter is the highest-leverage distribution channel on the platform. It reaches inboxes directly, is indexed by Google, and carries a 48% reach advantage. The barrier to starting is low — LinkedIn's native newsletter tool requires no additional setup.
6. Activate employee advocates on personal profiles. Company pages reach 1.6% of followers. Personal profiles posting the same content reach 561% more. The most effective B2B LinkedIn strategy in 2026 is three to five senior practitioners posting consistently within their expertise areas — not a company page posting daily. Build an internal programme that makes this easy.
7. Reply to every substantive comment within 30 minutes. Comment threads — where people reply to each other, not just to you — are a strong indirect engagement signal. They tell the algorithm you created a discussion, not just a broadcast. Replying quickly extends the thread and generates 64% more total comment volume.
The Quietest Implication: The Algorithm Now Reads Like a Human
The most significant thing about 360Brew is not any specific signal change. It is the underlying architecture. A 150-billion-parameter model trained on professional data can tell the difference between someone who knows a subject and someone who summarised three blog posts. It can smell the gap between expertise and performance. It has seen enough genuine expert content to recognise when something is hollow, regardless of how many people tapped a reaction.
This is the reset that the image shared at the top of this post describes. The loud creators who spent 2024 and 2025 accumulating vanity engagement with carousel slop and "Agree?" bait are not being penalised — they are simply becoming irrelevant. Nobody read it. The model noticed.
For B2B brands and practitioners who were always sharper in the room than on the page, this is the opening. The algorithm is now designed to find you. Your job is to make sure your profile and your content are aligned enough that it can.
The same principle applies to AI answer engines. 360Brew and ChatGPT, Perplexity, and Google AI Mode share the same underlying logic — semantic understanding of genuine expertise. Content that earns dwell time and saves on LinkedIn is the same content that gets cited in AI answers. A coherent content strategy aligned to your core expertise pillars simultaneously improves your LinkedIn reach and your AI search visibility. They are not separate problems.
If your LinkedIn presence does not yet sound like you — if it sounds like a summary of what you think you should be saying rather than what you actually know — that is the gap 360Brew was built to expose. And it is also the gap that the right content strategy closes.
LinkedIn Just Rewrote the Rules. Most Brands Haven't Noticed Yet.
In late 2024, LinkedIn quietly retired the system that had governed content distribution on the platform for years. No press release. No announcement post. Just a gradual, then sudden, collapse in reach for the majority of creators and brands — and a corresponding surge for a small cohort who happened to be doing exactly what the new system rewards.
The replacement is called 360Brew. It is a 150-billion-parameter, decoder-only foundation model trained on LinkedIn's own professional data, described in a research paper published on arXiv in January 2025 by LinkedIn's engineering team. It replaced not one algorithm but an ensemble of thousands of specialised ranking models — one for feed, one for job recommendations, one for people you may know, and so on — with a single adaptive AI that handles all of them simultaneously.
The implications for B2B marketers are significant, and most brands are still operating under the assumptions of the old system. This guide explains what 360Brew actually is, what it rewards, what it penalises, and the specific content strategy adjustments that restore and grow reach in 2026.
The Scale of the Shift: What the Data Shows
AuthoredUp's analysis of more than 3 million LinkedIn posts tracked the before-and-after with precision. Between June 2024 and May 2025, median impressions per post fell from 1,211 to 636 — a 47% decline in 12 months. That is not a seasonal dip. It is a structural redistribution of attention.
The distribution of that reach is the more important story. The top 2% of creators went from capturing 15% of all feed share to 31%. Regular users dropped from 57% to 28%. The algorithm did not simply reduce overall reach — it concentrated it. The question for every B2B marketer is which side of that concentration you are on.
| Metric | Before 360Brew (Jun 2024) | After 360Brew (May 2025) |
|---|---|---|
| Median impressions per post | 1,211 | 636 |
| Top 2% creators' feed share | 15% | 31% |
| Regular users' feed share | 57% | 28% |
| Company page follower reach | ~5% | 1.6% |
| Personal vs company page reach | ~3x advantage | 561% advantage |
Source: AuthoredUp analysis of 3M+ LinkedIn posts, June 2024 vs May 2025
How 360Brew Actually Works
The old LinkedIn algorithm was a factory with thousands of specialised machines. Each one had a specific job: count reactions, track keyword frequency, measure connection proximity, score recency. Engineers spent years hand-tuning each component. The system was powerful but brittle — it could be gamed because it measured proxies rather than substance.
360Brew is architecturally different. It is a decoder-only transformer — the same class of model as GPT-4 — but trained specifically on LinkedIn's professional data. Rather than counting signals, it reads your content like an editor. It understands semantic relationships between topics. It can infer that "Gong's revenue intelligence platform" and "Salesforce CRM integration" are related concepts without you ever using those exact terms together.
The feed ranking process now works in two stages. Stage one narrows millions of posts to a few hundred candidates based on your network and past behaviour — this part is largely unchanged. Stage two is where 360Brew takes over. It reads each candidate post, evaluates the author's profile for expertise alignment, assesses the quality of engagement the post has received, and scores relevance to the specific reader. The research paper describes this as "many-shot in-context learning" — the model creates a temporary personalised version of itself based on your last two to three months of activity, then applies that understanding to rank content for you specifically.
The New Engagement Hierarchy
The most consequential change under 360Brew is not what it reads — it is what it counts. The engagement hierarchy has been completely reordered.
| Signal | Algorithmic Weight | Why It Matters |
|---|---|---|
| Save | 5× a like | Signals lasting value — someone wants to return to it |
| Meaningful comment (15+ words) | Equivalent to 15 likes | Signals the post sparked genuine thinking |
| DM share | Second strongest signal overall | Private endorsement to a specific person |
| Dwell time 61+ seconds | 15.6% engagement rate | vs 1.2% for posts under 3 seconds dwell |
| Delayed engagement (24–72 hrs) | 4–6× better in Suggested | Signals evergreen value, not just novelty |
| Like | Baseline | Surface-level; still counted but weighted low |
| High likes + low dwell time | Negative signal | Flagged as engagement bait |
Source: AuthoredUp signal weight analysis; meet-lea.com dwell time research
The logic is straightforward once you see it. A like takes one tap and zero thought. A save means someone found your content valuable enough to return to it. A 200-word comment means someone read your post carefully enough to form a response. 360Brew is built to distinguish between these, and it does so with considerable accuracy.
Profile-Content Alignment: The Requirement Most Brands Miss
Here is the part of 360Brew that most B2B marketers have not yet absorbed: the algorithm checks whether your content matches your stated expertise before distributing it.
360Brew reads your profile — your headline, About section, experience, and skills — as text, then checks whether your posts fall within the same semantic territory. If your profile says "Head of Growth, B2B SaaS" but your recent posts cover crypto, travel photography, and political commentary alongside occasional marketing content, the algorithm cannot classify what you are actually expert in. So your content does not get matched to relevant audiences. It does not get penalised. It simply does not get distributed.
The research paper confirms this. The model uses profile data as a prior for understanding what expertise an author possesses, then evaluates each post against that prior. Topic-hopping resets the prior. It takes approximately 90 days of consistent, aligned posting for 360Brew to fully categorise you and begin optimising your distribution.
For B2B brands, this has a direct implication for employee advocacy programmes. The most effective LinkedIn strategy in 2026 is not a company page posting daily — it is three to five senior practitioners posting consistently within their defined expertise areas. The algorithm is designed to amplify exactly that pattern.
What Gets Penalised (And By How Much)
360Brew does not just reward good behaviour. It actively suppresses specific patterns that the old algorithm either ignored or inadvertently rewarded.
Engagement pods are detected with 97% accuracy. If caught, the account receives a shadow ban lasting 60 to 90 days. The model is trained on LinkedIn's full dataset of authentic versus inauthentic engagement patterns, and it has seen every pod format that exists. Short comments under 15 words from pod members carry reduced weight regardless of whether the pod is detected.
External links in the post body carry a 25 to 40% reach penalty. The workaround — placing the link in the first comment — is well-established and effective. If you are still putting URLs in the body of your posts, you are paying a significant distribution tax on every piece of content you publish.
AI-generated content receives 45% fewer interactions than human-written posts. This is not because LinkedIn can always detect AI writing — it is because AI-generated content tends to produce lower dwell time and fewer saves, which the algorithm interprets as low-value content regardless of origin.
Posting more than once per day triggers a reach penalty. The algorithm interprets high-frequency posting as a signal of volume-over-substance behaviour. The sweet spot is four to five posts per week for personal profiles.
Format Performance in 2026: The Numbers
Not all content formats are equal under 360Brew. The format performance data from 2026 is striking in how decisively it has shifted away from video — which received a significant algorithmic boost under the old system — toward formats that generate dwell time.
| Format | Reach Multiplier | Notes |
|---|---|---|
| Polls | 1.64× | Only 1.4% of content — massively underused |
| Carousels (documents) | 1.39× | 278% more engagement than video; 8–10 slides optimal |
| LinkedIn Newsletters | +48% reach | Google-indexed; lands in inbox; the sleeper weapon |
| Short vertical video (30–90s) | +80% reach | Up significantly; long video still underperforms |
| Text-only posts | 0.88× | Below average; weakest format in 2026 |
| Long-form video | 0.86× | Down 72% year-over-year |
Source: AuthoredUp format analysis, 2026
The carousel finding deserves particular attention. Carousels generate 596% more engagement than text-only posts and 278% more than video. The mechanism is dwell time — a 10-slide carousel requires genuine reading engagement. Each slide swipe is a passive signal of continued interest. The algorithm interprets this as high-quality content regardless of the reaction count.
LinkedIn Newsletters are the most underused high-leverage format on the platform. They reach your connections' inboxes directly, are indexed by Google, and carry a 48% reach advantage. If you are already writing long-form LinkedIn posts, converting them into a newsletter is the single highest-leverage format change you can make.
The Golden Window: Timing and Post Mechanics
The first 60 to 90 minutes after posting determine approximately 70% of a post's total reach. This has not changed under 360Brew — the initial velocity signal still matters. What has changed is that posts now have a second life.
Under the old algorithm, a post that did not gain traction in the first two hours was effectively dead. Under 360Brew, saves and meaningful comments received 24 to 72 hours after posting trigger a second distribution wave in LinkedIn's Suggested feed — and those delayed engagements perform four to six times better than initial ones. This means that evergreen, substantive content continues to accumulate reach long after the initial posting window.
The practical implication: reply to comments within 30 minutes of posting. AuthoredUp's data shows this generates 64% more total comment volume. And write posts long enough to require genuine reading — the 1,300 to 1,900 character sweet spot. Posts under 500 characters receive a 35% reach penalty as low-effort signals.
Seven Practical Adjustments for B2B Marketers
1. Audit your profile-content alignment before your next post. Read your LinkedIn headline and About section, then look at your last 20 posts. Do they fall within the same semantic territory? If not, 360Brew cannot classify your expertise and will suppress your distribution. Tighten your profile to match your actual content focus, or tighten your content to match your profile positioning.
2. Move all external links to the first comment. Every post with a URL in the body is paying a 25 to 40% reach tax. This is the single easiest fix with the most immediate impact. Write the post, publish it, then immediately add the first comment with the link and a one-line context note.
3. Build for saves, not likes. Before you publish, ask: is there something in this post that someone would want to return to? A framework, a data point, a decision matrix, a checklist. If the answer is no, the post will generate likes but not saves — and likes are now the weakest signal in the hierarchy.
4. Convert your best long-form posts into carousels. Take your highest-performing text posts from the last six months and rebuild them as 8 to 10-slide carousels. The format change alone will generate significantly more dwell time and reach for the same content.
5. Start a LinkedIn Newsletter. If you are writing substantive long-form content, a newsletter is the highest-leverage distribution channel on the platform. It reaches inboxes directly, is indexed by Google, and carries a 48% reach advantage. The barrier to starting is low — LinkedIn's native newsletter tool requires no additional setup.
6. Activate employee advocates on personal profiles. Company pages reach 1.6% of followers. Personal profiles posting the same content reach 561% more. The most effective B2B LinkedIn strategy in 2026 is three to five senior practitioners posting consistently within their expertise areas — not a company page posting daily. Build an internal programme that makes this easy.
7. Reply to every substantive comment within 30 minutes. Comment threads — where people reply to each other, not just to you — are a strong indirect engagement signal. They tell the algorithm you created a discussion, not just a broadcast. Replying quickly extends the thread and generates 64% more total comment volume.
The Quietest Implication: The Algorithm Now Reads Like a Human
The most significant thing about 360Brew is not any specific signal change. It is the underlying architecture. A 150-billion-parameter model trained on professional data can tell the difference between someone who knows a subject and someone who summarised three blog posts. It can smell the gap between expertise and performance. It has seen enough genuine expert content to recognise when something is hollow, regardless of how many people tapped a reaction.
This is the reset that the image shared at the top of this post describes. The loud creators who spent 2024 and 2025 accumulating vanity engagement with carousel slop and "Agree?" bait are not being penalised — they are simply becoming irrelevant. Nobody read it. The model noticed.
For B2B brands and practitioners who were always sharper in the room than on the page, this is the opening. The algorithm is now designed to find you. Your job is to make sure your profile and your content are aligned enough that it can.
The same principle applies to AI answer engines. 360Brew and ChatGPT, Perplexity, and Google AI Mode share the same underlying logic — semantic understanding of genuine expertise. Content that earns dwell time and saves on LinkedIn is the same content that gets cited in AI answers. A coherent content strategy aligned to your core expertise pillars simultaneously improves your LinkedIn reach and your AI search visibility. They are not separate problems.
If your LinkedIn presence does not yet sound like you — if it sounds like a summary of what you think you should be saying rather than what you actually know — that is the gap 360Brew was built to expose. And it is also the gap that the right content strategy closes.
