What changed in March 2026: the 360 Brew system
In March 2026, LinkedIn replaced its entire feed algorithm with an LLM-based system. The announcement was made publicly by Tim Jurka (VP Engineering, LinkedIn) and Hristo Danchev (LinkedIn Engineering Blog). Internal name of the system: 360 Brew.
This change is not cosmetic. It is not a tweak to the weights of like, comment and share — it is a complete overhaul of how LinkedIn evaluates content before distributing it. The 360 Brew system now analyses the semantic substance of every post, not just its engagement signals. And it compares that substance to two things: the declared editorial profile of the account, and the semantic history of its last 30 publications.
The consequences are measurable. According to the 2025 van der Blom study on 1.8 million posts, organic reach dropped 50% on average over 12 months, and engagement by 25%. Reach declined for 98% of users. And critically, it dropped most for those publishing "universal" content without a clear thematic anchor. Conversely, accounts that publish consistently and substantively within a precise niche saw their reach increase — sometimes dramatically.
A.U.T.H. is the method that decomposes the 4 signals 360 Brew evaluates and that determines your reach. It is not an algorithmic "hack" — it is rational alignment with what the system now rewards.
Who this method is for
A.U.T.H. is the natural complement to VPCEC and P.R.O.F.I.L.S. VPCEC builds the editorial strategy, P.R.O.F.I.L.S. optimises the profile as a conversion funnel, A.U.T.H. ensures the content you produce is actually distributed by the 360 Brew algorithm.
It is designed for:
- B2B leaders and creators active for 6+ months on LinkedIn who observe an unexplained reach drop on their recent publications even though the content appears high-quality.
- Profiles that already have a defined editorial angle (via VPCEC) but want to verify and adjust their algorithmic alignment.
- B2B marketing teams managing multiple leader accounts and needing to understand why some perform and others do not, despite seemingly equivalent content.
It is not designed for beginner accounts (under 30 publications): 360 Brew needs a minimum history to properly classify a profile. Before 30 posts, follow VPCEC and do not worry about A.U.T.H. yet.
Foundations: the 4 signals 360 Brew analyses
Why 4 levers and not 3 or 5? Because LinkedIn Engineering's public documentation (Tim Jurka, Hristo Danchev and the LinkedIn Engineering Blog team) explicitly identifies 4 signal families that the 360 Brew system uses as main inputs for its LLM scoring model.
These 4 families are:
- Content originality — semantic embedding detection of copied, paraphrased or AI-generated content with no personal contribution
- Perceived informational value — measured by saves, dwell time and content structure (fact density, presence of data)
- Temporal signals — engagement velocity in the first 60-90 minutes, 5-day engagement curve, repost rate
- Semantic match with profile and history — thematic consistency between content, profile declarations, and the last 30 publications
A.U.T.H. maps exactly to these 4 signals:
- Authenticity = content originality
- Usefulness = perceived informational value
- Timing = temporal signals
- Hub = semantic match
Skip a single signal and the whole system collapses: content can be authentic, useful and well-timed — but if it is not thematically aligned with the account's history, 360 Brew will treat it as noise. That is why A.U.T.H. must be activated as a system, not a checklist.
The 4 levers of A.U.T.H. in detail
Each lever is described with: the question it answers, the underlying algorithmic mechanism, the action to take, the KPI to measure and the mistake to avoid.
Authenticity
Being recognised as an original source, not a relay
The question: are your publications original content, or restatements of what everyone else is already saying?
The mechanism: 360 Brew uses semantic embeddings to measure the distance between your content and the 50,000 closest LinkedIn publications from the last 30 days. If the distance is too low, your content is classified as "restatement" and its initial reach is drastically reduced before it even hits a first real engagement test.
The action: impose a simple rule — every publication must contain at least one element that exists nowhere else: a personal quantified experience, a field observation only you have witnessed, an original synthesis of sources no one has crossed yet, or an explicit position on a polarising topic.
The KPI: initial reach rate (impressions in the first 60 minutes / number of followers). Benchmark: above 3% for 360 Brew-aligned content. Below, authenticity is a problem.
The mistake to avoid: using LLM output directly (ChatGPT, Claude, Gemini) without reworking the form and adding personal content. 360 Brew's embeddings detect LLM generation patterns with accuracy that improves every month. A 100% AI post, however well-written, is now identified and penalised.
Usefulness
Delivering operational value the reader can apply
The question: after reading your publication, what can the reader concretely do that they could not do before?
The mechanism: 360 Brew gives 5× more weight to saves than to likes since late 2025. The reason: a save is the only signal that demonstrates the content has deferred useful value — the reader will come back to apply it. A like is an affective signal, not a value signal.
The action: every publication must be structured around one applicable action. "Here is how we cut our sales cycle by 30% in 6 months, in 4 precise steps" is useful. "Personal branding is essential in B2B" is not. The first will be saved, the second will float by and be forgotten.
The KPI: saves to impressions ratio. Benchmark: aim for 3% minimum. Accounts above 5% see their reach multiplied by 2 to 3 over the following weeks via algorithmic compounding.
The mistake to avoid: publishing inspirational or motivational content with no applicable action. This kind of content sometimes gets many likes and warm comments but no saves — and 360 Brew will understand it as low-durability content, limiting its redistribution.
Timing
Mastering the golden hour and the 5-day engagement curve
The question: does your content get enough engagement in the first 60 to 90 minutes to trigger phase 2 of distribution?
The mechanism: 360 Brew uses a three-phase scoring system. Phase 1 (0-15 min): test distribution to 5-10% of the audience sample. Phase 2 (15-90 min): if the signal is positive, extension to the full follower audience plus expanded audience. Phase 3 (90 min to 5 days): progressive redistribution based on cumulative engagement. If phase 1 fails, phase 2 never happens.
The action: two operational rituals.
- Publish at your niche's optimal times — not "in general". For a French B2B audience of SME leaders: Tuesday 9-10 am and Wednesday 8-10 am (Metricool 2025).
- Prepare 3 personal follow-up comments to post yourself in the first 60 minutes to prime the conversation. Not empty "great post" — real clarifications, counter-examples, open questions.
The KPI: cumulative impressions at 90 minutes. Benchmark: at least 10% of your audience reached at 90 minutes to validate phase 1.
The mistake to avoid: scheduling multiple publications in a row on the same day. 360 Brew allocates a 24-hour attention budget per account — two close publications cannibalise each other and halve the reach of each.
Hub (semantic match)
Maintaining thematic consistency across 30+ publications
The question: do your account's last 30 publications cover a coherent, identifiable thematic universe?
The mechanism: "semantic match" is the central concept of 360 Brew. The system builds a semantic vector for each account from its last 30 publications. That vector determines in which user feeds the account will be recommended. If the vector is coherent (all posts in the same universe), the account is associated with a precise niche and strongly recommended to people interested in it. If the vector is scattered (posts on various topics), the account is associated with no niche and its overall reach collapses.
The action: audit your last 30 publications with a binary question per post: "Does this post clearly belong to my main semantic hub?" If the answer is no for more than 20% of posts, your vector is diluted. Fix it by: either deleting off-topic posts (history loss but alignment gain), or imposing a strict 100% rule for your next 30 posts (no exception).
The KPI: percentage of the last 30 posts belonging to your main thematic hub. Benchmark: minimum 80% (ideally 95%+).
The mistake to avoid: publishing "news hook" posts outside your hub just to ride a wave (e.g., an HR consultant posting about Mistral AI because it is trending). These posts usually get little engagement AND destroy the account's semantic alignment for the following weeks. A single off-hub post can dilute the vector for 10 to 15 days.
The 14-day activation procedure
A.U.T.H. deploys in 2 weeks: 1 week of audit and fixes, 1 week of validation with the first algorithmic signals.
Days 1-3 — Semantic hub audit
- List your last 30 publications (title, topic, format).
- Classify each post as in or out of the main thematic hub (binary).
- Calculate the alignment percentage.
- Identify off-hub posts that generated engagement — these are the most tempting to reproduce, and precisely the traps to avoid.
Days 4-5 — Score the 4 signals on the last 5 posts
- For each of the last 5 posts, score on 4 criteria: Authenticity (unique element or not), Usefulness (applicable action or not), Timing (0-90 min engagement peak or not), Hub (within niche or not).
- Identify the weakest signal on average.
- This signal becomes the correction priority.
Days 6-10 — Publish 3 A.U.T.H.-aligned posts
- 3 publications 48 hours apart minimum, each validated on the 4 criteria before going live.
- Each post explicitly contains: a unique element (A), an applicable action (U), is published at the optimal slot (T), and belongs to the hub (H).
- Real-time tracking of impressions at 90 min and save rate at 24 h.
Days 11-14 — Measure and calibrate
- Compare the 4 KPIs of the 3 new posts to those of the previous 5.
- If improvement on at least 2 KPIs → the system is calibrating, continue.
- If no improvement → the weak signal identified in days 4-5 has not been sufficiently corrected, run another cycle.
- Critical patience: 360 Brew takes 10 to 14 days to recalibrate an account's semantic vector. Any measure before 10 days is noise.
Real case: from 800 to 12,000 impressions per post
Profile: CEO of a French mid-cap food company (artisanal charcuterie, 140 employees, €42M revenue), 5 years on LinkedIn, 6,800 followers.
Starting point: regular publications (2-3 per week) for 18 months, average impressions per post in decline — from 4,200 a year ago to 800 in February 2026. No apparent strategy change, no visible mistake.
A.U.T.H. audit: across the last 30 publications, only 52% belonged to the main thematic hub (artisanal food and the short supply chain challenge). The remaining 48% covered female leadership, management tips, personal inspiration — all off-hub posts. Dominant weak signal: Hub. The account's semantic vector was diluted and 360 Brew no longer knew which niche to attach it to.
Execution: firm commitment on the next 30 posts — 100% in the "artisanal food and short supply chains" hub. Leadership and management topics temporarily abandoned. Each post explicitly contained a quantified field experience (A + U) and was published at optimal slots (T).
Results measured at 60 days after A.U.T.H. activation
12,000
avg impressions / post
× 15
vs starting point
4.1%
save rate
94%
hub alignment
The change cost nothing in working time — she was still publishing 2-3 times per week. Only the thematic discipline changed. Leadership posts she wrote for pleasure were moved to a separate personal account. Her professional account refocused on its hub, and 360 Brew re-associated it with the food niche with much broader recommendation. First inbound sales meetings arrived on day 45.
The 6 common mistakes to avoid
- 1
Copy-pasting LLM output without reworking
360 Brew's semantic embeddings detect LLM generation patterns. A 100% ChatGPT post, however brilliantly written, is identified and buried. Using AI to start and structure is OK, publishing directly is not.
- 2
Publishing multiple posts the same day
360 Brew allocates a 24-hour attention budget. Publishing twice in a day cannibalises both posts: each gets roughly 50% of the reach it would have had alone.
- 3
Riding a news hook outside your hub
A single off-hub post can dilute the account's semantic vector for 10 to 15 days. The one-time engagement gain almost never offsets the reach loss on following posts.
- 4
Chasing likes rather than saves
Since the late-2025 overhaul, a save is worth 5× a like in 360 Brew scoring. A post with 500 likes and 3 saves is algorithmically weaker than a post with 50 likes and 8 saves.
- 5
Measuring too early
360 Brew takes 10 to 14 days to recalibrate an account's semantic vector after a strategy change. The first 3 A.U.T.H.-aligned posts will almost never have the expected effect — that is normal, the system has not yet realised something changed. Persevere.
- 6
Confusing A.U.T.H. with VPCEC
VPCEC is the editorial strategy (what to publish and why). A.U.T.H. is algorithmic alignment (how to get it distributed). Applying A.U.T.H. without VPCEC produces well-distributed content that serves no sales strategy. Applying VPCEC without A.U.T.H. produces strategic but invisible content. The two methods are complementary, not substitutable.
Bonus resource — 100% free
The 360 Brew LLM feed audit
Download the 7-page PDF audit grid to score your last 30 publications on the 4 A.U.T.H. levers and identify your weakest signal. Diagnostic grid, 14-day correction plan and algorithmic benchmarks.
- A.U.T.H. scoring grid for 30 publications
- Algorithmic benchmarks (reach, saves, timing, hub)
- 14-day correction plan per weak signal
Direct download coming soon
The automatic download system is being finalised. In the meantime, this resource is sent via the weekly All In newsletter — subscribe to receive it.
Subscribe to the newsletterGoing further
VPCEC — executive personal branding
VPCEC defines the editorial strategy A.U.T.H. distributes. Apply them together for maximum impact: strategy + algorithmic alignment.
Read the VPCEC methodDone-for-you execution
Our sister offer 72H CHRONO PRO applies A.U.T.H. in production on your SME's LinkedIn company page for 12 months, with a website + content + LinkedIn triptych delivered in 72 hours.
Discover the offer