LinkedIn's Feed Algorithm Switched to LLM-Based Recommendations
LinkedIn confirmed in 2026 that it has deployed Generative Recommenders -- large language model-backed recommendation systems similar to Instagram and TikTok -- replacing its prior rule-based algorithm. The system now measures relevance and depth rather than raw engagement volume.
The practical consequences for B2B marketers are significant and directional.
The 60% External Link Penalty: What It Means
Posts containing external links -- to blog posts, landing pages, case studies, webinar registration pages -- now receive approximately 60% less distribution than equivalent posts without external links. LinkedIn is optimizing for users staying on-platform, and any post that pushes users off-platform is penalized at the distribution layer.
This has immediate implications for how B2B marketers promote events, content, and campaigns on LinkedIn.
What no longer works:
- "Register for our webinar [link]" posts from company pages
- "Read our latest case study [link]" posts
- Link-in-comments workarounds (now detected and penalized similarly)
- Hashtag stuffing to boost reach
What the algorithm rewards:
- Posts that generate deep engagement: comments with multiple sentences, replies, extended reading time
- Native LinkedIn content: carousels, documents, polls with context, and native video
- Personal profiles (8x more engagement than company pages)
- Original perspective and named expertise rather than recycled insights
The Depth Score: LinkedIn's New Core Ranking Signal
LinkedIn introduced "Depth Score" as a primary ranking signal in 2026. It measures how long users engage with content -- dwell time, multi-sentence comments, saves, and private shares all factor in. A post that gets 500 reactions but no substantive comments scores lower than a post with 50 comments averaging four sentences each.
For B2B marketers, this means optimizing for quality of engagement over volume. One genuine "this changed how I think about X" comment from an ideal customer is worth more algorithmically than 50 reaction-bait thumbs-ups.
What This Means for B2B Event Promotion on LinkedIn
Most B2B event promotion on LinkedIn currently relies on a formula that is now broken: company page posts with event registration links, scheduled during business hours, tagged with industry hashtags.
The new formula has to look like this:
1. Promote events from personal profiles, not company pages. Personal profiles deliver 8x more engagement. Founders, executives, and team members posting about event topics outperform company page announcements by a wide margin.
2. Post topic content without links first. Build a week of topical posts about the subject your event covers -- in native LinkedIn format, no external links -- before posting the event invite. The algorithm will have established your topic authority before the promotion post lands.
3. Use LinkedIn native events rather than external registration. LinkedIn's native event format does not carry the external link penalty because it keeps users on-platform. This is not always possible when your registration system is external, but where you can use LinkedIn native events, do it.
4. Let attendee conversation do the promotion. Posts by attendees saying "just signed up for this event" or sharing what topic they are most interested in generate organic algorithmic amplification that outperforms company announcement posts.
Why LinkedOtter's Event-Led Motion Is Built for This Algorithm
At LinkedOtter, we invite target accounts to events -- we do not pitch them. Invitations go by direct message and personalized outreach, not mass LinkedIn posts with external links. The event creates the intent signal. The follow-up starts with attendees who have already self-selected.
That motion is not dependent on LinkedIn feed reach. We generated 38 C-level attendees at RSA and 754 webinar signups in 26 days using direct outreach and targeted invite lists, not viral LinkedIn posts.
The LinkedIn algorithm change matters for teams whose pipeline depends on organic LinkedIn reach. For teams running event-led outbound with direct invite infrastructure, it changes the content strategy but not the core motion.