LinkedIn Algorithm & 360Brew
Competitive Intelligence Dashboard · Last updated April 15, 2026
Scan complete
7-day freshness window
Freshness note: No LinkedIn official announcements found in the April 9–15, 2026 window. Most current practitioner data: Socialinsider benchmark report (~Apr 1, 2026), Limelight/Extramile analysis (~Apr 1–8, 2026). Sections with no new weekly data are marked accordingly.
Median Reach Drop (YoY)
-47%
AuthoredUp, 3M post study — Dec 2025
360Brew Model Size
150B
Parameters — arXiv / LinkedIn Engineering
Avg Platform Engagement Rate
5.20%
+8% YoY — Socialinsider, 1.3M posts, Apr 2026
Carousel vs Text Engagement
3.7×
Carousels outperform text — meet-lea.com, Feb 2026
No new 360Brew model announcements in the last 7 days (Apr 9–15, 2026). Background and recent practitioner analysis below.
What 360Brew Is — Architecture Confirmed
LinkedIn replaced thousands of specialized ranking models with one massive AI system called 360Brew — a 150-billion parameter language model, the same technology class that powers ChatGPT, but trained specifically on professional network data. The introduction of 360Brew is not an iterative update to ranking logic — it replaces the platform's central nervous system, shifting from a deterministic calculator of user actions to a probabilistic reasoner of user intent.
ConfirmedBackgroundSource: AuthoredUp / pettauer.net — Dec 2025 / Jan 2026
Deployment Status: Partial Rollout, Expanding
LinkedIn published the 360Brew research paper in January 2025, describing a "pre-production model V1.0" developed over 9 months. They have not announced deployment percentages or which surfaces are fully using it — but creators started reporting visibility changes in summer 2024, which is LinkedIn's typical pattern of quiet, gradual A/B testing. Current educated estimate: 40–100% deployed across different feed surfaces.
RumoredBackgroundSource: AuthoredUp — Dec 2025
Four Pillars: How 360Brew Scores Content
360Brew assesses content based on four key pillars — the first being Profile Coherence: whether a post aligns with a creator's professional identity and expertise. The "360" refers to a full, holistic view of professional activity — posts, comments, profile, and network all count. The "Brew" aspect is how the AI blends hundreds of signals (timing, relevance, relationships). It doesn't just scan keywords — it understands the meaning behind words.
ConfirmedRecentSource: upgrowth.in / mdcomms.co.uk — Apr 2026 / Nov 2025
Many-Shot In-Context Learning: Per-User Personalisation
360Brew creates a temporary "personalized version" of the model for each viewer by analysing their last 2–3 months of activity. The research calls this "many-shot in-context learning" — meaning the algorithm learns what a specific user considers valuable, then applies that understanding.
ConfirmedBackgroundSource: AuthoredUp — Dec 2025
Future Trajectory: Agentic Feed, Format Agnosticism
Looking ahead, 360Brew may evolve toward a state where users do not browse a feed — an AI agent curates on demand. Some analyses argue the model is becoming more format-agnostic over time, increasingly prioritizing information density over container. Marketers should choose the format that best conveys information, not the format that once gamed distribution.
RumoredRecentSource: pettauer.net — Jan 2026
Based on Socialinsider (1.3M posts, Apr 2026), meet-lea.com (Feb 2026), justpollen.com (Mar 12, 2026). Bar = relative engagement score.
Format
Engagement & Notes
Avg Eng. Rate
Carousel / Doc
Platform avg 5.20%; native document leads at 7.00% avg (+14% YoY). Generates 15–20s dwell time vs 8–10s for simple formats.
24.42%
Multi-Image
Multi-image posts achieve 6.60% average engagement — the second-best format after carousels. They sustain dwell time better than single images while requiring less user interaction than carousels.
6.60%
Native Video
Videos achieve 6.47% average engagement, with 53% more engagement for short videos under 60 seconds. Across all LinkedIn pages there is a 36% YoY decline in video views. Videos with subtitles achieve 29% higher engagement.
6.47%
Poll
Polls achieve 4.40% average engagement — a doubling since 2023. Despite their efficacy, they remain heavily underutilised. Well-crafted polls (2–4 options, "Other" option prompting comments) can generate hundreds of responses and often trend as platform-recommended content.
4.40%
Text-Only
Text posts achieve approximately 4% engagement rate, viable primarily when writing is sharp, personal, or provides genuine insight. Long-form text posts (1,000–1,300 characters) outperform shorter variants because they generate extended dwell time.
~4.00%
Format Rotation Dividend
Accounts rotating between formats (carousel, text, video, polls) achieve 37% more follower growth and 28% more consistent visibility compared to accounts using single formats repeatedly. Audience fatigue with repetitive formats is real, making content diversity strategically important.
1. Polls — Underutilised, High-Signal Format
LinkedIn polls achieve an average engagement rate of 4.40%, representing a doubling of their performance since 2023, yet they remain heavily underutilised relative to their reach potential. Well-crafted polls with 2–4 options and an "Other" option to prompt comments can generate hundreds of responses and often trend as platform-recommended content. Competitors slow to adopt polls leave white space here.
New data — Apr 2026Confirmed
2. AI Citation Positioning — Semantic Consistency as Moat
A SEMrush study found AI responses have a 0.57–0.60 semantic similarity score with LinkedIn content — meaning AI doesn't just "read" LinkedIn posts, it mirrors them, repeating specific definitions and brand voice in its own answers. The fix: use precise, consistent terminology. If you want to own "Sustainable Architecture," don't alternate with "Green Building" or "Eco-friendly." Pick your script and maintain it so the AI knows exactly how to quote you.
~Apr 1, 2026Rumored
3. Long-Form Text Revival — Machines Now Read for Depth
Long-form writing is back in style in 2026 — not because LinkedIn wants to be Medium, but because long-form is easier to evaluate for depth and detail. In 2026, you are not only writing for a busy executive scrolling between meetings; you are also writing for machines trying to decide whether you have earned the right to show up when someone asks a professional question. Long-form text posts (800–1,000+ characters with strong formatting: short paragraphs, bold key points, numbered lists) naturally boost time-on-content.
Background — Jan 2026Confirmed
No new creator program announcements found in the April 9–15, 2026 scan window. No new updates this week.
Posting Frequency De-emphasised by 360Brew
The 360Brew update significantly reduces the importance of posting frequency. This shift allows busy professionals to focus on developing thoughtful, well-researched content rather than rushing to maintain an aggressive posting schedule. Posting more often will not automatically get more reach. 360Brew is far more interested in whether content is relevant than whether you are posting every day. It is recommended to focus on a few core topics rather than a broad spectrum — this helps the algorithm understand who you are and who should be seeing your content.
ConfirmedRecentSource: thebrightbrand.com / extramiledigital.com — Dec 2025 / Apr 2026
Reach Declines of 30–50% Widely Reported Since 360Brew Rollout
Many users report reach decreases of 30–50% following the 360Brew update. This reduction stems from two primary factors: more precise targeting (content is shown to more relevant — but narrower — audiences) and algorithm confusion about users with inconsistent profiles or content patterns.
ConfirmedBackgroundSource: thebrightbrand.com — Dec 2025