May 15, 2026 · 10 min read

The LinkedIn algorithm in 2026: how it actually works (and what changed)

LinkedIn's feed-ranking algorithm is the most consequential black box in B2B distribution. Whether your post reaches 200 people or 20,000 is determined by signals you can't directly see. Here's what's known about how the 2026 version actually works, what changed in the most recent updates, and what that means for anyone — recruiter, sales rep, founder, consultant, job seeker — trying to get reach on the platform.

The 2024-2026 changes that reshaped the feed

Three publicly-documented changes (acknowledged in LinkedIn engineering blogs and confirmed by reach pattern shifts) reshaped the feed between 2024 and 2026:

The "knowledge feed" pivot (Q4 2024). LinkedIn explicitly down-weighted personal-brand posts, broetry, and engagement-bait formats in favour of "expert content" — substantive posts with specific knowledge, data, or analysis. Posts with screenshots of dashboards, code, frameworks, or numbers see materially more reach in 2026 than the daily-life posts that dominated 2022-2023. The phrase "insightful" reaction was added during this period and is weighted higher in the ranking signal than a generic "like."

Comment quality weighting (Q1 2025). The ranker started weighting comment length and substance more heavily than raw comment count. A post with 5 long substantive comments now outranks a post with 50 "great share!" comments. This change disproportionately rewarded creators with engaged audiences who actually reply in paragraphs, and disproportionately punished engagement-pod groups exchanging short comments.

Connection-tier dwell-time signal (Q2 2025-Q1 2026). The ranker started weighting how long someone in your network dwells on your post — not just whether they reacted. Posts that hold attention for 8+ seconds get distributed further; posts that get scroll-by-with-a-like get throttled. This is the change that made "first-line hooks" so important — the first 200 characters now have to earn the dwell.

The cumulative effect: shallow content posted at high frequency (the 2022 playbook) now gets less reach than substantive content posted weekly. The platform has tilted toward a smaller number of higher-quality posts with longer reads.

What the ranker actually weighs (best public knowledge)

LinkedIn's published research and reverse-engineering work by creators converge on roughly six signals. None of these are confirmed by LinkedIn directly with weights, but the rank-correlation is well-documented.

Author-recipient affinity. How often does the recipient interact with this author historically? The single largest signal. If you've engaged with someone's posts in the past, their next post is much more likely to land in your feed. This is why warming a network through repeated substantive comments compounds — you're literally training the algorithm to put you in front of those people.

Content topic relevance. LinkedIn classifies your post into topic categories and matches against the topics each viewer has engaged with. A post about machine learning shown to people who engage with ML content; a recruiting post shown to people who engage with talent-acquisition content. Mixed-topic content (your founder-story-plus-product-pitch posts) underperforms because the topic classifier can't cleanly route it.

Engagement velocity in the first hour. Posts that get strong comment + reaction velocity in the first 60-90 minutes get pushed wider. Posts that don't get traction in the first hour rarely recover. This is why "post-and-comment-on-your-own-post" is a real tactic — you're seeding the velocity signal manually.

Dwell time per impression. As above, dwell time is now a first-class signal. A 200-impression post where average dwell is 12 seconds outranks a 200-impression post where average dwell is 3 seconds.

Comment substance and reciprocity. The ranker weights long substantive comments higher than short ones, and weights comment threads (back-and-forth) higher than one-shot comments. This is why "asking a question in the post body" works — it nudges the comment pattern toward threads.

Author Trust Score / reputation. LinkedIn's internal Trust Score (which determines connection-request capacity, InMail capacity, and automation-classifier sensitivity) also factors into post distribution. Higher-trust authors get wider initial distribution; lower-trust authors get throttled before the post even reaches the velocity-test phase. See Trust Score deep-dive for the signals that move it.

What the ranker actively penalises

The negative signals matter as much as the positive ones in 2026.

External links in the post body. LinkedIn does not want you sending traffic off-platform. A post with a link in the body gets significantly less distribution than the same post with the link in the first comment. This is documented and stable since 2022.

Engagement-bait phrasing. "Comment YES below" / "tag someone who needs this" / "agree?" closings now get explicitly down-ranked. The classifier was retrained in 2024-2025 to recognise these patterns.

Repetitive posting on the same theme within 48 hours. Posting two posts on the same topic within 48 hours triggers a same-author-same-topic suppression. The second post gets ~30-40% of the first post's reach.

AI-generated content signatures. As of mid-2025, LinkedIn started detecting and down-ranking content that pattern-matches as raw GPT-4-class output — characteristic phrasings, em-dash overuse in certain patterns, listicle structures with sentences of similar length. The detection isn't perfect but it's real, and posts that read as obviously AI-generated get less reach than the same content rewritten in a human voice.

High follower-to-engagement-ratio asymmetry. Accounts with 50K followers and 10 engagements per post get systematically throttled because the ratio signals a purchased or stale audience. Accounts with 2K followers and 80 engagements per post are favoured.

What this means by audience

The algorithm changes affect different LinkedIn users in different ways depending on their goal.

Recruiters posting roles. Job-posting content is in a separate algorithmic lane from feed posts and gets steady (but capped) distribution to job-seekers in the relevant skill cluster. Posting about recruiting (insights, market data, candidate experience) is in the regular feed and benefits from the substantive-content tilt. The recruiter who posts a hot take on the talent market once a week, with data, gets more reach than the recruiter who posts "we're hiring!" daily.

Sales reps and SDRs. The post-broadcast model (sales rep posts about their product, hopes prospects see it) was always weak; the 2026 algorithm makes it weaker. The motion that works is commenting substantively on prospects' posts, which lands in their notifications and is visible to their network. The algorithm now actively rewards this — comments are the ranker's favorite engagement form, and your name appearing under thoughtful comments on the right posts is far better distribution than your own posts ever achieve.

Founders building inbound. Substantive posts (frameworks, data, behind-the-scenes operational details) get reach; promotional posts about funding rounds and product launches get less reach than they used to (the algorithm explicitly down-weighted "company news" in the knowledge-feed pivot). The pattern that works for founders in 2026 is one substantive weekly post, plus daily commenting on people in your TAM.

Job seekers networking. Posting your own job-search posts has limited reach (LinkedIn knows your account is in job-seeker mode and the ranker doesn't push that content widely). Commenting substantively on hiring managers' posts in target companies builds the visibility that posts can't.

Consultants and operators building presence. Same pattern as founders — one substantive weekly post on your domain expertise, plus daily commenting in your client persona's feed. The ranker rewards expertise-content authors with consistent topic focus more than it rewards general thought-leaders covering everything.

What still works in 2026 — the short version

The pattern that the algorithm now structurally favours is consistent across audiences:

One substantive post per week on a tightly-defined topic, formatted for dwell (a sharp first line, a paragraph break early, ideally a screenshot or specific data point), published when your audience is on the platform (Tuesday-Thursday 8-10am US Eastern is still the strongest window for B2B), with no external link in the body.

Daily substantive commenting on 3-5 posts from people in your target network — comments that extend the post's argument, add a related data point, or ask a sharper follow-up question. 30-80 words per comment. This is where the algorithmic compounding actually happens — your name appears repeatedly in the right people's notifications, the affinity signal builds, and your future posts get pushed wider to that audience.

The combined effect of these two motions is roughly 5-10× the reach of someone posting daily without commenting on others' content. The 2026 LinkedIn algorithm is not a content-volume game; it's a focused-engagement game.

For more on the engagement-first motion that the algorithm rewards, see how to grow your LinkedIn network in 2026. For the warming sequence specifically (which uses the same algorithmic mechanics that comments unlock), see InMail reply rate collapse and what's still working. For the Trust Score signals that gate distribution capacity, see LinkedIn Trust Score explained.