Something broke on LinkedIn in 2026, and almost everyone noticed at the same time.
Posts that would reliably pull thousands of impressions started pulling hundreds. Accounts that had grown steadily for years flattened out. The advice that worked in 2024 stopped working, and in a few specific cases it started actively hurting.
The reason is that LinkedIn changed how ranking works, and the change was structural rather than cosmetic. This is not a tweak to how much a comment is worth. It is a different way of deciding what your post is about and who should see it.
This guide covers what is actually known about how LinkedIn ranks content in 2026, what it rewards, what it demotes, and what to do about it. Where something is reported rather than confirmed, it says so, because LinkedIn does not publish its ranking logic and anyone who tells you otherwise is guessing with confidence.
Key Takeaways
- Ranking now runs through a single AI model rather than the separate filtering and scoring stages that older explanations describe.
- Dwell time, saves, and comment depth carry the weight. Likes carry very little.
- Topic authority is the genuinely new concept: alignment between your headline, your experience, and what you post about affects how far a post travels.
- Outbound links cost roughly 60% of reach, and putting the link in the first comment no longer avoids that.
- Engagement pods are detected and neutralized, and posts caught in one get cut.
- Native vertical video is the one format whose distribution to non-followers is expanding.
- Reach fell platform-wide, so check whether your decline is steeper than everyone else's before assuming you caused it.
What Actually Changed
For most of LinkedIn's history, the feed was explained as a pipeline. A post was classified, filtered for quality, shown to a small test audience, scored on the response, and then either promoted or left to die. Every guide written between 2020 and 2024 described some version of that sequence, and it was close enough to be useful.
In 2026 that description no longer matches what people observe. Industry coverage attributes the shift to LinkedIn consolidating ranking into a single foundation model. LinkedIn's own published research calls it 360Brew. It replaces the stack of separate specialized systems that came before it.
The distinction sounds academic. It is not, and here is why it matters in practice.
A pipeline of separate systems evaluates a post as an object. Does it contain a link? How many reactions did it get in the first hour? Each check is independent.
A single model evaluates the same post in context. It weighs four things at once:
- Who wrote it.
- What that person usually writes about.
- Whether this reader has engaged with that subject before.
- How similar readers responded to similar posts.
So the post is not scored on its own. It is scored as something you wrote.
That is why the two biggest changes in 2026 are both contextual rather than mechanical. Topic authority is contextual. Dwell time is contextual. Neither can be gamed by adding or removing a feature of the post itself.
The Signals That Decide Reach Now
- Dwell time
- Saves
- Comment depth
- Topic authority
Four things determine how far a post travels. Notice what is not on the list.
Dwell time
Dwell time is how long someone stays with your post before scrolling past. It is the closest available proxy for whether the post was worth reading, and it is close to unfakeable, which is exactly why it now carries so much weight.
Forbes reported in July 2026 that generic machine-written posts are not flagged so much as ignored. Readers scroll past in under three seconds, and that dead dwell time is what tells LinkedIn to stop showing the post.
This is the mechanism behind the widespread belief that LinkedIn penalizes AI writing. It does not. It penalizes writing nobody finishes, and a lot of unedited AI output falls into that category on its own merits.
The practical implication is that length is not the enemy. A long post people finish beats a short post people skip. What kills dwell time is a hook that does not earn the second line, and a middle that does not earn the end.
Saves
Forbes reported in the same July 2026 piece that a save drives roughly five times the reach of a like. That single ratio should change what you write.
A like means someone approved. A save means someone expects to need this again. Those are completely different signals about value, and only one of them is hard to earn. Posts that get saved tend to share a shape: a framework, a checklist, a set of numbers, a process someone will run later. Posts that get liked tend to be observations people agree with.
Both are legitimate. Only one of them compounds.
Comment depth
Comments still matter, but not all comments count the same. A one-word "Great post" and a four-sentence disagreement are not equivalent signals, and the ranking system appears to treat them accordingly. Replies that generate further replies are worth the most, because a thread is the strongest possible evidence that a post was worth someone's attention.
This is also why comment bait backfires now. Asking people to type "YES" produces exactly the kind of comment that carries the least weight, while marking the post as manipulative.
Topic authority
This is the new one, and it gets its own section below.
What is missing from the list
Likes. Forbes reported in July 2026 that likes now carry minimal algorithmic weight. They remain a useful signal for you, in that they tell you a topic resonated, but they no longer move distribution. If you have been optimizing for reactions, you have been optimizing for the metric that got demoted.
If you want the underlying definitions, our explainer on what LinkedIn impressions actually measure covers how these numbers relate to each other.
Topic Authority, the Concept Nobody Had to Think About Before
Here is the change most people have not adjusted to.
Multiple 2026 analyses report that if your headline and experience do not align with the topics you post about, your reach is suppressed.
The reasoning is intuitive once a single model is doing the ranking. To decide whether to show your post to someone who has never heard of you, the system needs some basis for believing you are worth reading on this subject. Your track record is the most available evidence it has.
The consequence is uncomfortable for a lot of people. You can write an excellent post and still get limited distribution, purely because it sits outside the lane your profile and history established.
Two things follow from that. Your headline is now a ranking input rather than a business card, so it should name the subject you actually write about. And range has a cost it did not have before: posting across eight unrelated topics does not make you interesting to the system, it makes you unclassifiable.
This deserves more than a section, so we wrote a full guide to LinkedIn topic authority and how to build it, including what drift looks like and how to recover from it. FeedBoss's Authority Map exists for this specific problem, tracking which topics you have actually established and where the gaps are.
What Gets Demoted
- Outbound links in the post
- Link promised in the comments
- Engagement pods
- Comment bait
- Writing that reads as generic
Forbes published a piece in July 2026 on the content moves LinkedIn started punishing. Five stand out, and four of them are things people were explicitly taught to do.
Outbound links, including in the first comment
Sharing an external link reduces visibility by approximately 60%, per that reporting. The part that surprised people is that putting the link in the first comment now carries the same penalty, closing the workaround that had been standard practice for years.
The logic is not hidden. LinkedIn ranks for time spent on LinkedIn, and a link is an exit. We cover the alternatives in detail in our guide to links and LinkedIn reach.
Engagement pods
LinkedIn's 2026 system is built to detect pods and render them entirely ineffective, and it cuts the reach of every post caught in one. This turns a tactic that used to be merely inefficient into one that is actively harmful.
If you are in a pod, leaving is the highest-return action available to you this week, and it costs nothing.
Comment bait
"Comment YES if you agree" and its variants are suppressed. The tell is the prescribed response: you are not asking a question, you are requesting a mechanical action. An open question at the end of a post is a completely different thing and remains fine.
Writing that reads as generic
Covered above under dwell time. It is worth repeating here because people keep filing it under "AI penalty" when the actual mechanism is that nobody finishes reading.
If you use AI to draft, the fix is not to stop. It is to ground the draft in something only you have:
- A number from your own work.
- A conversation you actually had.
- A decision you got wrong.
Our post on what AI slop actually is goes deeper on the difference.
Format by Format
Distribution now varies enormously by format, more than at any previous point.
| Format | Reported distribution in 2026 | Best for |
|---|---|---|
| Native vertical video | Widest reach to non-followers of any format | Growing beyond your existing network |
| Documents and carousels | Strong dwell time, high save rate | Frameworks and reference material |
| Text posts | Steady, mostly within your network | Opinion, stories, discussion |
| Newsletters | Notification-driven rather than feed-driven | Compounding a subscribed audience |
| Polls | Diminished from their 2022 peak | Little, at this point |
| Posts with outbound links | Roughly 60% less reach | Only when the click matters more than the audience |
Native vertical video is the headline here. 2026 coverage of the format shift describes distribution to non-followers at a rate no other format is matching. That makes it the single biggest new opportunity in the current ranking system.
It is also the format most people on LinkedIn are least comfortable with, which is precisely why the opportunity is still open. FeedBoss's AI Video Studio exists to lower that barrier, on the Pro plan and above.
One caveat worth stating plainly: format advantages close. Video is wide open now because supply is low relative to the distribution on offer. That will not be permanent.
The First Hour
2026 coverage consistently points at the first 60 minutes as the window that decides a post's ceiling. The early response appears to determine whether the post gets tested on a wider audience at all.
Three practical consequences.
Post when your people are actually around, not when a generic best-times chart says to. Your own analytics beat any published average. Our guide on the best time to post on LinkedIn covers how to find yours rather than borrowing someone else's.
Be available afterwards. Replying to early comments in the first hour extends threads at exactly the moment threads count for most. Posting and walking away wastes the window.
Do not confuse this with urgency theater. The first hour matters because it is when the response happens, not because posting faster helps. A worse post published at the perfect minute is still a worse post.
Company Pages Versus People
The gap between personal profiles and company pages widened into a chasm in 2026.
Coverage this year puts company page organic reach down 60% to 66% against 2024. Posts reshared by employees are reported to travel dramatically further than the same content published by the page itself.
Some analyses put that multiple in the hundreds of percent. That is a large enough claim to deserve a hedge. The direction is well supported. The precise multiple is not.
Either way, the strategic read is the same. If your company page is your primary LinkedIn channel, you are pushing against the platform's strongest current. The people at your company are the distribution. The page is a reference surface.
For teams, that means the work is enabling employees to post in their own voices rather than producing more page content. It is slower, it requires trust, and it is the only thing that works.
What Did Not Change
Enough changed in 2026 that it is worth being clear about what did not, because a lot of panic advice is currently throwing out things that still work.
These all still hold:
- Consistency still beats volume.
- Specificity still beats abstraction.
- Comments on other people's posts are still one of the most efficient ways to be discovered.
- Your first three lines still carry the post. LinkedIn still truncates at roughly 140 characters on mobile and 220 on desktop, and you can still check yours with a post preview tool before publishing.
And writing that sounds like a person still outperforms writing that sounds like a brand. That has been true through every ranking change LinkedIn has made and there is no sign of it reversing.
If you are short on things to write about, our list of LinkedIn post ideas is organized by the kind of post rather than by industry, which makes it more useful when you are stuck.
Your Next 30 Days
- Headline names the subject you actually post about
- Current role and experience support that subject
- Featured section carries the link you would otherwise post
- Two or three adjacent topics, not eight unrelated ones
- First 140 characters carry the hook
- At least one post a month worth saving, not just reading
- One native vertical video to see how your audience responds
- No engagement pods, at all
- No comment bait phrasing
- Available to reply for the first hour after posting
- Judge on a rolling four-week average, never post to post
- Track saves and comments, not likes
- Give any change six to eight weeks before reading it
Work these in order. The profile items are one-time fixes that take an afternoon. The content and behavior items are habits. The measurement items exist to stop you from drawing conclusions from noise, which at current volatility is the most common mistake people make when they try to fix their reach.
If your reach has already dropped and you want a diagnostic rather than a checklist, our six-step guide to diagnosing a reach drop works through the causes in order of how much they typically explain.
Where Tools Fit
None of this requires software. It requires knowing what the system rewards and being consistent for longer than feels comfortable.
Where tools help is the parts that are pure friction:
- Keeping a posting cadence you can actually hold.
- Knowing which of your own topics have traction.
- Getting from a blank page to a draft that sounds like you.
FeedBoss covers those in one place, starting at $29 per month. The AI post generator is trained on your own top-performing posts. The analytics dashboard reports reactions, comments, shares and follower growth, broken down by format. A content calendar handles the consistency part.
One honest limit worth knowing before you shop for any tool: no third-party tool can show you impressions, reach, or saves for your posts. LinkedIn does not expose those to them. Only LinkedIn's own analytics has them.
What no tool can do is give you something worth saying. That part is still yours.
Conclusion
The 2026 LinkedIn algorithm rewards a narrower and more demanding thing than its predecessors did. It wants writing on a subject you have established, that people finish reading, and that some of them keep.
That is harder than what worked before. It is also more durable, because no tactic reproduces it. The people who lost the most reach this year are the ones whose growth depended on mechanics: pods, bait, link workarounds, posting volume. The people who barely noticed are the ones who were already writing well about one thing.
If you are rebuilding, start with your headline, pick two or three adjacent topics, and give it eight weeks before you judge anything. That is a slower answer than most algorithm guides offer, and it is the one the current system is built to reward.
Consistency is what the ranking rewards. Holding it is the hard part.
Start growing freeNo credit card required.
Frequently Asked Questions
What is the LinkedIn algorithm in 2026?
It is the ranking system that decides which posts appear in each person's feed and in what order. In 2026 it works as one unified AI model rather than the separate stages of filtering and scoring that earlier explanations described. The practical difference is that it judges a post in the context of who you are and what you usually write about, not just the post on its own.
Why did my LinkedIn reach drop so much?
Probably not because of anything you did. Independent analyses published through 2026 report platform-wide declines ranging from around 25% on engagement to as much as 66% on reach, and they do not agree with each other. Check whether your own drop is steeper than that band before you assume you caused it.
Do likes still matter on LinkedIn?
Barely, as a ranking signal. Forbes reported in July 2026 that likes now carry minimal algorithmic weight, while a save drives roughly five times the reach of a like. Likes are still useful as a sign that your topic resonates, but they are no longer what moves distribution.
What is dwell time and why does it matter?
Dwell time is how long someone stays with your post before scrolling on. It is the clearest available proxy for whether a post was worth reading, so the ranking system leans on it heavily. This is why a post that people finish reading beats a post that collects fast reactions and gets scrolled past.
Does LinkedIn penalize AI-generated posts?
Not directly, and this is the most common misunderstanding. There is no reported AI penalty. What happens is that generic writing gets scrolled past in about three seconds, and that dead dwell time is what suppresses the post. AI-assisted writing grounded in a real number or moment from your own work does not have that problem.
Which format gets the most reach on LinkedIn in 2026?
Native vertical video is currently getting distribution to non-followers at a rate no other format matches, according to 2026 coverage of the format shift. Documents and carousels remain strong for saves. Text posts still work, but they mostly reach people already connected to you.
How important are the first 60 minutes after posting?
Very. The early response to a post appears to determine how widely it is tested afterwards, which is why the first hour gets so much attention in 2026 algorithm coverage. Post when your audience is actually around, and be available to reply to early comments.
Do hashtags still help on LinkedIn?
Much less than they used to. Hashtag following was deprecated and the ranking system infers topic from the text itself, so a wall of hashtags mostly signals reach-chasing. Two or three relevant ones are harmless. Ten are not helping.