This document captures research findings on LinkedIn's content distribution algorithm, drawn from multiple large-scale analyses. It is structured in two parts:
Part 1 — Early Engagement Mechanics covers the engagement signals, golden window and format multipliers that determine how posts are distributed once they pass the initial quality filter. This research was conducted in late 2025 and the underlying mechanics remain valid.
Part 2 — The 360Brew Shift (March 2026 update) covers LinkedIn's fundamental algorithm overhaul: the replacement of task-specific ranking models with a single 150-billion-parameter AI model called 360Brew. This changes what gets through the quality filter in the first place, particularly for AI-generated content, and introduces semantic analysis, topic authority scoring and AI content detection.
The engagement mechanics in Part 1 still apply — once a post clears the 360Brew quality gate, the golden window, comment threading and dwell time dynamics are unchanged. But 360Brew means that template-following, AI-patterned posts may never reach the engagement test at all.
Research basis: 1.8M+ posts (van der Blom Algorithm Insights 2025), 50,000+ posts (LinkIntel), 1M posts (Socialinsider), 621,833 posts (AuthoredUp), and 40+ additional sources.
LinkIntel's analysis of 50,000+ posts found that the first 30 minutes now influence 75% of total reach, up from 60% in 2024. Posts where the author responds to comments within 30 minutes receive:
- 64% more total comments
- 2.3x more views
This is the single highest-leverage intervention point after the content itself. The algorithm uses early engagement velocity to decide whether to push the post to a wider audience beyond the initial 2–5% sample.
Under 360Brew, LinkedIn now calls this "Sustained Relevance Scoring" rather than the golden hour. The concept is similar, but the window for high-quality posts has extended: posts that maintain genuine engagement over 48–72 hours can achieve 340% longer active distribution time.
LinkedIn's algorithm does not treat all engagement equally. The weighting from highest to lowest value:
- Share with caption — highest value. Someone endorsing your content to their own network with added context.
- Save — strong signal of reference value. Under 360Brew, saves have become one of the most critical ranking signals. A post with 200 saves now dramatically outperforms one with 1,000 likes.
- Repost — similar to share but without added commentary.
- Long comment (15+ words) — worth roughly 2.5x a like.
- Standard comment (5+ words) — worth roughly 2x a like.
- Reaction (emoji reactions beyond thumbs up).
- Like — lowest value. Likes carry significantly less weight under 360Brew than they did previously.
A single substantive 15+ word comment is worth approximately 2.5 passive likes in algorithmic terms.
Posts that spark 3+ comment exchanges between different participants (not just the author replying to individuals, but people replying to each other) receive 5.2x amplification versus comparable posts without that discussion depth.
This is why contrarian takes and genuine questions outperform informational posts — they create debate between readers rather than a series of isolated reactions.
Under 360Brew, the algorithm also evaluates comment quality and lexical diversity. Comment sections filled with "Great post!" or "Agree!" are identified as low-entropy noise. If comments share similar phrasing or originate from a tight cluster of users who always engage with each other, the model downgrades reach.
Comments from relevant industry experts carry 5–7x more algorithmic weight than interactions from random connections.
One comment from a senior decision-maker in your target audience is worth five or six from people outside the industry. The algorithm assesses professional background alignment between the commenter and the post topic.
Under 360Brew, engagement from relevant industry experts carries even greater weight. An MIT Sloan study of 1,200 B2B profiles found that the top 30% with highest network relevance (not network size) achieved 210% higher content performance.
Dwell time (how long someone spends reading a post) is the algorithm's most important ranking signal because it is nearly impossible to fake.
- Posts with 61+ seconds of dwell time average a 15.6% engagement rate
- Posts viewed under 3 seconds average just 1.2%
This is why well-structured content with visual breathing room outperforms short hot takes in the long run. Carousels are particularly strong, averaging 2–3 minutes of dwell time versus 15–30 seconds for text posts.
Under 360Brew, dwell time has been formally elevated to the #1 ranking factor, ahead of all other engagement signals.
- Posts continue distribution for 48–72 hours under 360Brew (up from 2–3 weeks via Suggested Posts in 2024–25). High-performing content can circulate for longer, but consecutive posts now cannibalise each other more aggressively.
- Posting multiple times within 24 hours causes the newer post to receive less reach.
- More than 3 consecutive same-format posts decrease reach by 20%+.
- Minimum gap between posts should be 12 hours.
- The optimal posting frequency is now 3–4 posts per week, not daily. Each new post cuts the distribution lifecycle of the previous one.
- Optimal posting times for B2B: Tuesday to Thursday, 9–11 AM in your target audience's timezone.
- Content from individual profiles receives 3x higher engagement than company page posts. Under 360Brew, this gap has widened further.
- Recycled or duplicate content achieves 84% less reach than original content.
Research shows different post formats have measurably different engagement levels:
| Format | Engagement Multiplier |
|---|---|
| Polls | 1.64x |
| Carousel/Document | 1.45x (6.6% engagement rate) |
| Multi-image | Highest raw engagement (6.6%) |
| Single image | 1.18x |
| Native video | 1.10x |
| Text-only | Baseline |
Under 360Brew, single-image posts underperform text-only content by 30%, reversing 2024–25 patterns. Videos under 30 seconds achieve 200% higher completion rates than longer formats. LinkedIn Live videos receive 7x more reactions than standard video uploads.
Research basis: LinkedIn 360Brew engineering documentation, Richard van der Blom Algorithm Insights 2025 (1.8M posts), AuthoredUp (3M+ posts, Q3 2025), Originality.ai (3,368 posts, January 2026), Brixon Group (500+ B2B profiles), and 20+ additional sources on 360Brew mechanics and AI content detection.
In late 2024, LinkedIn began replacing its entire content ranking infrastructure — previously a collection of thousands of task-specific models — with a single unified AI system called 360Brew. This is a decoder-only transformer with 150 billion parameters, trained specifically on LinkedIn's networking data.
The old algorithm tracked metadata: clicks, hashtags, connection graphs, post timestamps. 360Brew reads meaning. It understands that "AI governance framework" and "EU AI Act compliance strategy" are related concepts even if those exact terms never appear together. It interprets whether someone's profile, expertise claims and content actually align.
The practical consequence: LinkedIn now evaluates you, not just your posts. Your profile, posting history, topic consistency and engagement patterns are assessed holistically.
360Brew does not use a simple binary "AI detector." It evaluates a combination of signals that, together, make low-effort AI content easy to identify:
Lexical diversity. Human writers naturally vary vocabulary, sentence structure and rhythm. AI falls into patterns — repeating similar phrases, using predictable transitions, maintaining an unnaturally consistent tone.
Template pattern recognition. Posts that follow the standard AI template (Hook → Body → CTA, with single-sentence paragraphs, bullet points and a rhetorical question close) are structurally identical to millions of other AI-generated posts. 360Brew's semantic engine recognises this conformity.
Profile-content alignment. The algorithm checks whether posts match the creator's stated expertise. AI-written content often lacks the specific, contextual language that comes from real professional experience in a particular domain.
Engagement authenticity. Comment velocity, account relationship patterns and engagement timing are analysed to identify coordinated or automated engagement.
The data on AI content suppression is consistent across sources:
- AI-generated content receives approximately 30% less reach and 55% less engagement compared to human-written posts (van der Blom, 2025).
- Posts with recognisable AI patterns achieve up to 47% less organic reach (Brixon Group, 500+ B2B profiles).
- Overall, median post reach dropped 47% year-over-year after the 360Brew rollout (AuthoredUp, 3M+ posts, Q3 2025). This partly reflects the relevance-first design: fewer impressions, but higher-quality impressions.
Posts that could have been written by anyone — or any AI — are distributed to no one in particular.
360Brew assigns topic authority based on sustained content within a defined area. The mechanics:
- Post consistently about 2–3 related topics for 60–90 days and 360Brew starts to recognise you as a credible voice in that space.
- Scatter posts across random topics and the algorithm cannot determine who should see your content.
- If your profile says "Marketing Director" but you post about cryptocurrency, the algorithm sees a mismatch and suppresses distribution.
- Content that aligns with the creator's stated expertise receives higher confidence scoring and broader distribution.
This is why broad, generic "motivational" content is increasingly crowded out — it provides no topical signal for the algorithm to match against relevant audiences.
360Brew evaluates your profile before distributing your content. This is what practitioners call the "profile-content audition":
- Headline and About section define your content niche in the algorithm's interpretation.
- Professional experience is matched against post topics. Domain- specific language that could only come from a practitioner carries more weight than generic advice.
- Engagement history — who you comment on, what you react to — tells the algorithm which "table you sit at." Engaging broadly with random content prevents the algorithm from placing you in a relevant content community.
Tactics that were common under the old system now actively harm visibility:
- Engagement bait. "Comment YES if you agree" and similar patterns are detected and downgraded. LinkedIn recognises 70+ engagement bait patterns.
- Engagement pods. LinkedIn's AI identifies reciprocal engagement patterns with 97% accuracy. Pods now suppress rather than boost reach.
- External links in post body. Posts with external links experience approximately 60% reach reduction. LinkedIn prioritises keeping users on platform.
- Excessive hashtags. More than 5 hashtags triggers a 68% reach reduction. Hashtags have had no measurable positive impact on reach since late 2024.
- Topic scatter. Posting across unrelated topics prevents 360Brew from classifying expertise.
- Post-and-ghost behaviour. Publishing AI content and not engaging with comments. 360Brew interprets this as low commitment and reduces distribution.
- Post editing after publishing. Significant edits reset the distribution clock.
- Deletion and reposting. LinkedIn's algorithm remembers deleted content and may penalise the repost.
The algorithm is explicitly designed to surface:
- Genuine expertise. Content that demonstrates real domain knowledge through specific examples, practitioner language and concrete results.
- Topic consistency. Sustained posting within a defined professional niche over 60–90+ days.
- Save-worthy content. Frameworks, checklists, original data and actionable specifics that people bookmark for later reference.
- Substantive engagement. Comments that add context, disagree thoughtfully or extend the conversation.
- Dwell time. Content dense enough to reward careful reading.
- Authentic human voice. Posts that vary in rhythm, include genuine uncertainty, and contain the rough edges that AI tends to smooth away.
Every AI post generator uses the same structural formula:
Hook → Body → CTA
The hook is constrained to 7–12 words or 210 characters. The body uses single-sentence paragraphs, bullet points and white space. The CTA is a rhetorical question or engagement prompt.
This template is now so ubiquitous that it functions as an AI fingerprint. Over 53% of long-form LinkedIn posts are classified as "Likely AI" by detection tools (Originality.ai, January 2026), and the structural conformity is a primary detection signal.
The expanded version adds more layers (re-hook, end-of-body teaser, second CTA for reposts) but the underlying architecture is identical. A post that follows this template — regardless of how good the individual sentences are — signals to both human readers and the algorithm that it was assembled from a standard AI framework.
The backlash against AI-generated LinkedIn content is now widespread. Audiences have developed sensitivity to AI-sounding posts, and content that follows recognisable AI patterns gets scrolled past or called out in comments.
Common audience complaints:
- Identical structures across unrelated posts
- Hollow enthusiasm: "I'm thrilled to announce..."
- Dramatic openings that deliver mundane conclusions
- Generic "lessons learned" that could apply to any industry
- Rhetorical questions that serve the algorithm, not the reader
- Emoji overload as personality substitutes
The professionals gaining ground under 360Brew are those with clear expertise, consistent themes and content that reads as if a specific human being wrote it about a specific subject they know well.
Last updated: March 2026