How the LinkedIn Algorithm Works in 2026: What B2B Teams Need to Know
How the LinkedIn algorithm works in 2026, sourced to LinkedIn: the signals its feed ranks on, what it limits and which popular claims lack a source.
TL;DR: The LinkedIn algorithm in 2026 retrieves candidate posts for each member and ranks them by the predicted chance that the member will read, react, comment or share. LinkedIn has said it uses dwell time and skips, values conversation over likes, can show posts a week or two old, and limits reach for engagement pods, automated comments and generic AI-looking posts. In traxy's study of 1,965 B2B posts, comments made up 19% of engagement.
This guide is part of our LinkedIn content strategy series. Every claim about the feed links to LinkedIn's own engineering papers, help pages and newsroom, or to a named LinkedIn employee quoted by the press. Claims we could not trace to LinkedIn are listed separately near the end.
How Does the LinkedIn Algorithm Work in 2026?
LinkedIn's feed works in two steps. First it retrieves a pool of candidate posts for a member, then a ranking model scores each one. In LinkedIn's February 2026 paper on its feed ranking model, the authors describe the process this way: "candidate posts are first retrieved from multiple sources. Then a ranking model predicts the likelihood of each action" (LinkedIn, arXiv, February 2026). The actions it predicts include long dwell, clicks and skips, along with likes, comments and shares.
In March 2026, LinkedIn's engineering team described a rebuilt feed that uses large language models to understand posts. Tim Jurka, LinkedIn's VP of Engineering, said LinkedIn had moved to large-scale LLMs to understand content in much more depth. LinkedIn's blog post says its new ranking model "processes more than a thousand of your historical interactions" and relies on professional signals and engagement patterns, "never demographic attributes" (VentureBeat, March 16, 2026, reporting on LinkedIn Engineering).
Step 1: Quality classification
LinkedIn screens posts for spam and quality. Its engineering team wrote that "our online and nearline classifiers label every image, text, or long form post as 'spam,' 'low-quality,' or 'clear' in near real time" (LinkedIn Engineering, 2017). The same post says LinkedIn continuously predicts whether a share is likely to go viral.
Step 2: Ranking by predicted actions
The ranking model estimates what each member is likely to do with each post. LinkedIn's LiRank paper lists "like, comment, share, vote, and long dwell and click" as the action probabilities it predicts (LinkedIn, arXiv, February 2024). LinkedIn has also studied dwell time directly, looking at "aggregated per-update dwell time" to improve feed ranking, and built a model that demotes posts members are likely to skip (LinkedIn Engineering, 2020).
Step 3: Relevance over recency
The feed does not show posts in time order. Rishi Jobanputra, LinkedIn's senior director of product management for the feed, said in 2025 that a post a week or two old can still be relevant and keep sparking conversations (PR Daily, August 25, 2025). In July 2025 LinkedIn rolled back a test that had shown many weeks-old posts. Bhairavi Jhaveri, a LinkedIn communications lead, called it testing "to strike the right balance between prioritising relevant content vs recent content in your feed" (Mediaweek, July 17, 2025).
LinkedIn Algorithm Signals: What LinkedIn Has Said
Signal | What LinkedIn has said | Source |
|---|---|---|
Predicted actions | The ranking model predicts the likelihood of like, comment, share, vote, long dwell and click | |
Dwell time and skips | Per-update dwell time is used to improve feed ranking, and posts likely to be skipped are demoted | |
Member history | The ranking model processes more than a thousand of a member's past interactions and uses no demographic attributes | |
Post quality | Classifiers label every post as spam, low-quality or clear in near real time | |
Conversations vs likes | "Likes signify good direction, but authentic conversations and connection signify great engagement" | |
Post age | Posts a week or two old can still be shown if they are relevant | |
"Read more" clicks | LinkedIn no longer counts "read more" clicks as positive engagement the way it once did | |
Automated comments | Detected automated comments are left out of "Most Relevant" comments and sometimes not shown outside the commenter's network | |
Engagement pods | Pods are "not allowed on LinkedIn," and LinkedIn says it actively limits the reach of inauthentic activity | |
Generic AI-looking content | Content that seems AI-generated and lacks a clear perspective "is less likely to be widely distributed beyond a person's immediate network" | |
Saves and sends | Post analytics show saves (people who bookmarked the post) and sends (shares in a LinkedIn message) |
Last verified: November 12, 2026.
LinkedIn has not said how much each signal weighs. Saves and sends appear in post analytics, but we found no LinkedIn statement that they count for more than likes in ranking.
What Does LinkedIn Limit or Demote?
Automated comments and engagement pods
LinkedIn defines automated comments as "comments posted to LinkedIn using a browser extension, script, or third party tool." It describes an engagement pod as "a coordinated group of people that like, comment and share posts" and says "this form of inauthentic activity erodes trust and is not allowed on LinkedIn" (LinkedIn, March 13, 2026). Its Help Center adds that LinkedIn may limit how many comments a member or Page can make in a period, and may limit the visibility of comments made with an automation tool (LinkedIn Help, via Social Media Today, August 2025). Our LinkedIn engagement pods guide covers the risks in more detail.
Generic AI-generated posts
LinkedIn does not ban AI-assisted writing. Laura Lorenzetti, LinkedIn's Executive Editor, wrote: "It's ok to use AI to help you write, but your posts and comments need to represent your voice and your perspectives." Posts that look AI-generated and lack a clear perspective are less likely to be shown beyond the author's immediate network, and LinkedIn said its early testing identified generic content correctly 94% of the time (LinkedIn, May 20, 2026).
Spam and low-quality posts
Jobanputra said LinkedIn will "actively detect and limit the reach of spammy or low-quality content, including bot-generated posts and comments" (PR Daily, August 2025).
How Does the Algorithm Treat Different Content Formats?
LinkedIn has said little about reach by format. Its public statements cover four formats.
Video. LinkedIn said video uploads grew 36% year over year and video viewership was up 36%, and video created on the platform grew 100% (TechCrunch, February 4, 2025).
Polls. In 2022 LinkedIn said it had "heard feedback that there are too many polls in the feed" and would show fewer polls from people members don't know (LinkedIn, May 5, 2022).
Hashtags. In July 2025 Jobanputra said LinkedIn was removing the hashtag feed because people were not using it much (Digital Information World, July 19, 2025).
Links. We found no LinkedIn source for a fixed reach penalty on posts with external links. LinkedIn has not published how links affect ranking.
For a format-by-format comparison by pipeline goal, see LinkedIn content formats compared, and for formatting mechanics, how to post on LinkedIn.
What Does Engagement Look Like on B2B Posts?
LinkedIn does not publish engagement benchmarks, so we measured them. In our LinkedIn engagement benchmarks study of 1,965 posts published by 48 B2B profiles from August 6 to October 28, 2026, the median post got 125 engagements.
Metric | Value |
|---|---|
Median engagements per post | 125 (88 reactions, 27 comments, 1 repost) |
Share of engagement that is reactions, comments and reposts | 79%, 19% and 1.5% |
Comments per reaction on the median post | 1 comment for every 3.8 reactions |
Median engagement per 100 followers | 0.23 |
Engagement per 100 followers, profiles under 10K followers | 0.64 |
Engagement per 100 followers, profiles over 100K followers | 0.09 |
Share of engagement won by the top 10% of posts | 45% |
Source: traxy, LinkedIn Engagement Benchmarks (2026), 1,965 posts from 48 tracked B2B profiles, extracted November 4, 2026. Engagement per 100 followers = (reactions + comments + reposts) ÷ followers × 100. Last verified: November 12, 2026.
Comments are the scarcer signal, at about one for every four reactions, and conversation is what LinkedIn's feed team calls "great engagement." Smaller accounts earned more engagement per follower than large ones.
Company Pages vs Personal Profiles
LinkedIn's own material points B2B teams toward both. On Company Pages, LinkedIn says "companies that post weekly see a 2x lift in engagement with their content" (LinkedIn Marketing Solutions). On employees, LinkedIn's Elevate material says "on average, a company's employees have 10x as many connections as the company has followers" (LinkedIn Elevate).
For B2B teams, that makes founder and employee profiles the main way to reach buyers, with the Company Page in support. See our guides to LinkedIn employee advocacy and LinkedIn company page optimization.
LinkedIn Algorithm Claims We Could Not Source
These figures are widely repeated, including in an earlier version of this page. We could not find them in any LinkedIn source, so we removed them.
"Company Page posts reach 1.6% of followers" and "Company Page content is 1–2% of the feed." No LinkedIn source found.
"Posts with external links get 30–50% less reach." No LinkedIn source found, and LinkedIn's public statements focus on the value of the post rather than a link penalty.
"Employees' networks are 12× larger than a Company Page's following." LinkedIn's own figure is 10×, above.
"LinkedIn shows posts up to two or three weeks old." LinkedIn has said "a week or two," and it rolled back the 2025 test that showed older posts.
"Saves and sends weigh more than likes." LinkedIn added both to analytics but has not said how they are weighted.
"Every post is first shown to a small test audience." LinkedIn has described retrieval and ranking, not a fixed test-audience stage.
What This Means for B2B Pipeline
1. Write for the people you sell to
The ranking model uses each member's history of more than a thousand interactions, so posts reach people who have engaged with similar topics. Posting regularly on the problems your buyers have builds that audience over time.
2. Aim for comments and reading time
LinkedIn predicts dwell time and comments, and its feed team calls conversation "great engagement." Posts that teach something specific or take a clear position give readers a reason to stay and reply.
3. Skip pods and automation
LinkedIn hides detected automated comments from "Most Relevant" and limits the reach of engagement pods. Engagement from them can lead LinkedIn to limit a post.
4. Track who engages
Saves and sends appear only as counts, but comments and reactions are public and named. A comment from a buyer at a target account is a sales signal. Our guide to LinkedIn engagement signals explains which ones to act on, and LinkedIn engagement to pipeline covers the follow-up.
Algorithm-Friendly Posting Checklist for B2B Teams
Before you post
Clear audience: name the job title or function the post is for.
Your own perspective: add experience, data or an opinion that generic AI-written copy would not have.
Worth saving: include something a reader would come back to.
Content structure
Hook: open with a specific claim, number or question.
Value in the post: give the answer in the post itself.
Invite comments: end with a question your buyers can answer from experience.
After you post
Reply to comments while the conversation is active.
Skip pods and comment automation, which LinkedIn limits.
Track who engages, especially people at accounts that match your ICP.
Check timing against our data on the best time to post on LinkedIn for B2B.
How to Cite This Page
traxy, "How the LinkedIn Algorithm Works in 2026," https://traxy.ai/blog/linkedin-algorithm-2026-how-it-works-b2b, last verified November 12, 2026.
When you quote a LinkedIn statement, please cite LinkedIn's original page as well. The engagement figures come from traxy's LinkedIn engagement benchmarks study.
FAQ
How does the LinkedIn algorithm work in 2026?
LinkedIn retrieves candidate posts for each member and ranks them with a model that predicts how likely the member is to read for a while, click, skip, like, comment or share. LinkedIn says the model draws on more than a thousand of the member's past interactions and no demographic attributes. Posts are screened for spam and quality, and relevant posts a week or two old can still appear.
What engagement does the LinkedIn algorithm value most?
LinkedIn has not published weights, but its feed team has said "likes signify good direction, but authentic conversations and connection signify great engagement." Its ranking papers list comments, shares, likes, clicks and long dwell time among the actions it predicts. In traxy's study of 1,965 B2B posts, comments were 19% of engagement and reactions 79%.
Does LinkedIn penalize posts with external links?
We found no LinkedIn source for a fixed link penalty such as the often quoted 30–50% drop in reach. LinkedIn has not published how links affect ranking. If you share a link, make sure the post is useful on its own, and test link placement against your own analytics.
Does the LinkedIn algorithm penalize AI-generated content?
LinkedIn says using AI to help you write is fine, but posts must represent your own voice and perspective. Content that looks AI-generated and lacks a clear perspective is less likely to be distributed beyond the author's immediate network. LinkedIn said its early testing identified generic content correctly 94% of the time.
Do engagement pods work on LinkedIn?
LinkedIn says engagement pods are not allowed and that it actively limits the reach of inauthentic activity. It also leaves detected automated comments out of the "Most Relevant" view and sometimes hides them outside the commenter's network. The extra likes and comments a pod adds can lead LinkedIn to limit the post instead of spreading it.
How often should you post on LinkedIn?
LinkedIn says Company Pages that post weekly see a 2x lift in engagement. For personal profiles, Buffer's study of more than 2 million posts found that, compared with posting once a week, posting 2–5 times a week was linked to about 1,182 more impressions per post, and 6–10 times a week to about 5,001 more. Our best time to post on LinkedIn guide has a weekly schedule.
The Takeaway for B2B Teams
LinkedIn ranks each post by what it predicts each member will do with it, and it has said it favors conversation and limits pods, automation and generic AI posts. The people who comment on posts in your market are the ones worth a sales conversation. traxy finds them, scores them against your ICP and sends the best-fit leads to your team, and you can try it free for 7 days.
Related reading:
Keep reading
All LinkedIn Content & Posting guides
LinkedIn Content Strategy (2026): A B2B LinkedIn Marketing Strategy That Generates Leads
LinkedIn content strategy for B2B: a 7-step LinkedIn marketing strategy, content pillars tied to pipeline, how to measure it and every LinkedIn content guide.
How to Post on LinkedIn (2026): Post Format, Character Limits, Bold Text, Hashtags, Tagging and Scheduling
How to post on LinkedIn step by step: post format, the 3,000-character limit, bold text, hashtags, how to tag someone and how to schedule LinkedIn posts.
Employee Advocacy on LinkedIn (2026): What It Is, Examples and How to Build a B2B Program
What employee advocacy is, LinkedIn employee advocacy examples, how to build a program in 6 steps, best practices and how to measure it for B2B pipeline.