TL;DR
Niche beats network. LinkedIn's new algorithm, 360Brew, matches posts to people interested in your topics instead of just showing them to your connections - which means your content reaches buyers in your ICP who've never heard of you.
Every post runs a gauntlet: a quality gate, then a test batch in roughly the first hour, then expansion if the test goes well. Comments and dwell time drive expansion; engagement bait and generic AI writing kill it.
The winning strategy hasn't changed since the update - it's just finally rewarded: consistent, specific, expert content from real people, posted where your team can engage in the first hour.
The big change: from your network to your niche
For most of LinkedIn's history, the feed was a connection graph: post something, and it got shown to the people you were connected to. Reach was a function of network size.
That era is over. The new system has a name: 360Brew, the foundation-model ranking system LinkedIn rolled out across the feed (there's a white paper, if you enjoy digging). Instead of scoring posts with hand-built connection heuristics, 360Brew is a large model that reads everything as context: your profile, what you post about, what you've liked and viewed, and what every potential viewer cares about - then matches posts to readers the way TikTok or YouTube matches videos to viewers. The feed became an interest graph: the algorithm works out what your post is about and serves it to people engaged with that topic, whether or not they follow you.
Emil broke this down on LinkedIn when he noticed his own reach dipping and traced it to the rollout - the post is a good two-minute version of this article:
For B2B this is the whole ballgame. It means a security engineer posting about compliance gets shown to CISOs, not to their college friends. On a sales call last week, a GTM engineer evaluating Postbeam summarized his plan in exactly these terms: have each AE own a niche - his example was an AE selling into private equity, posting only about how PE firms use the product - and let the algorithm build that AE a following inside the accounts he sells to. That plan only works because of this change. It's also why our webinar panel kept hammering niche consistency: the algorithm can only match you to an audience if it can tell what you're about.
Your profile is part of the algorithm now
The under-appreciated consequence of a model that reads everything as context: it's not just scoring your posts, it's scoring you. Your headline, your about section, your past posts, even what you like and view all feed the model's picture of what you're an authority on - and that picture decides whose feeds your posts can reach. Four moves follow:
- 1. Declare 2-3 clear topics you want to reach people about, and stay in that lane.
- 2. Rewrite your profile to back them up. Headline, about section, job descriptions - all building the case for why you have authority on those topics. Don't keyword-stuff; write it the way you'd write your company's landing page.
- 3. Engage in your lane too. What you like, comment on, and dwell on teaches the model your topics as much as what you post.
- 4. Post 1-5 times a week about those topics, consistently. Volume without the topical spine doesn't compound; the spine without volume never gets tested.
The silver lining Emil called out: the feed is still flooded with generic content. If you sound like a person and are genuinely helpful, the bar for out-of-network distribution is lower than it looks.
How a post actually gets distributed: the three stages
Stage 1: the quality gate
The moment you hit post, LinkedIn classifies it: spam, low quality, or clear. Obvious spam signals (link farms, follow chains, pods) get filtered. Low-quality signals - engagement bait, walls of hashtags, content that pattern-matches to templated AI writing - don't get removed, they just start with a handicap. Clear content moves on at full strength.
Stage 2: the test batch (the golden hour)
Your post is then shown to a test audience: a slice of your followers plus people who engage with your topics. What happens in roughly the first hour decides everything. The signals that matter, in rough order: comments (especially ones long enough to be real, and your replies to them), dwell time (people actually reading, which is why documents and longer posts punch above their weight), shares, then reactions. This is why the best time to post is less about the clock and more about when your first-degree audience is around to engage - and why teams set up Slack pings so colleagues see a new post while the test batch is live.
Stage 3: expansion into your niche
Pass the test and distribution widens in waves - to more of your network, then further into the interest graph: non-followers in your niche, people in adjacent roles, the buyers you actually want. Strong posts keep resurfacing for days (sometimes weeks) instead of dying in an afternoon. Fail the test and the post quietly stops - which, as we covered in our breakdown of Kate Tollo's virality rules, is the good news: failure is invisible, so the rational move is to post more and read the signal.
What the algorithm rewards (and punishes) in 2026
Rewarded:
- - Demonstrated expertise. Specific, first-hand knowledge in a consistent niche. The algorithm is explicitly tuned for it, and so are buyers.
- - Conversation. Posts that start real comment threads, and authors who reply within the first hours.
- - Dwell time. Content people actually stop and read: strong hooks, documents and carousels, formatting that pulls the reader down the page.
- - People over pages. Personal profiles consistently out-reach company pages - the foundation of every employee advocacy program.
- - Consistency. Regular posting in one lane teaches the algorithm who to show you to. Going dark for a month means rebuilding from zero.
Punished (or quietly suppressed):
- - Engagement bait. “Comment YES and I'll DM you the guide” is explicitly demoted.
- - Generic AI slop. Templated, could-be-anyone writing underperforms both algorithmically and with humans - one reason templated-post tools stopped working. AI drafting works when it's fed your real context and voice, not a generic prompt.
- - Link-first posts. Posts whose only job is pushing people off-platform earn less reach; put the link in the comments or make the post stand on its own.
- - Automation flags. Scraping-based tools and bot engagement risk more than reach. Publishing through LinkedIn's official partner APIs avoids the whole category of risk.
The B2B playbook: make the algorithm your distribution
Put the three stages together and the strategy writes itself:
- 1. Give everyone a lane. Each founder, AE, and engineer owns a niche and posts in it consistently, so the algorithm learns exactly which buyers to serve them to.
- 2. Win the quality gate with real context. Draft from actual work - deals, builds, customer conversations - in each person's own voice. That's what separates expert content from slop, and it's what Postbeam's MCP is for.
- 3. Win the test batch as a team. Post when people are around, get Slack notifications out the moment a post goes live, and have teammates comment (genuinely) in the first hour.
- 4. Harvest the expansion. The whole point of reaching non-follower buyers is knowing who they are: engagement analytics plus ICP-filtered warm leads turn algorithm reach into pipeline instead of vanity impressions.
- 5. Repeat what the algorithm confirms. A post that beats your average is a signal. Iterate on it instead of reinventing - the repeat-your-winners rule works because the test-batch mechanic resets with every post.
If you run a team, this is exactly the system we built Postbeam for GTM teams around: 15 minutes per person per week, one sitting - review the draft, tweak, approve.
Built for the 2026 algorithm
Postbeam drafts in each person's voice from their real work context, publishes through LinkedIn's official APIs, pings your team in Slack for the golden hour, and shows you which engagers match your ICP - so the algorithm's reach becomes your pipeline.
Book a demo →