If your LinkedIn posts reach a fraction of what they did two years ago, you are reading the platform correctly. The decline is real, it is measurable, and it is structural rather than a bad month. Studies through early 2026 put company-page organic reach down roughly 60% since 2024 — the average company post now surfaces to a low-single-digit share of a feed — while the average professional's views are down about half from their 2024 peak; a post that reached 10,000 people then now struggles past 4,000 on the same follower count. But the headline number hides the more useful story, which is that LinkedIn did not simply turn reach down. It rebuilt the machine that assigns reach. In late 2024 it replaced a patchwork of ranking models with 360Brew, a single in-house foundation model, and over 2025–2026 it shifted the feed from a relationship graph — content from people you know, so reach scaled with follower count — to an interest graph, where a post is shown to whoever the model thinks cares about the topic, connection or not. That one change decoupled follower count from reach, which is why big accounts posting the same way they always did watched their numbers fall while smaller, sharply-focused accounts sometimes climbed. Layered on top are three deliberate throttles: company pages are held down to push brands toward ads, generic and template-shaped AI content is actively deprioritized, and the June 2026 ranking update cut distribution for engagement-bait, recycled posts, and inconsistent posting. This guide separates the structural cause from the tactical symptoms, tells you which of your reach loss you can recover and which you can't, and lays out the content profile the interest graph actually rewards — because the fix is not posting more, it is posting the thing the new machine was built to distribute.
Start by ruling out the two comforting explanations, because neither holds. It is not a bad month, and it is not just you. Analyses through early 2026 converge on a large, sustained drop: company-page organic reach is down roughly 60% since 2024, and the average company post now surfaces to only a low-single-digit percentage of the feed — several trackers put company-page posts at just 1–2% of the content in a typical feed, down from a share that was multiples of that a few years ago. For individual professionals the fall is gentler but still severe: average views are down on the order of half from their 2024 peak, and the most-quoted illustration is a post that reached 10,000 people in 2024 now landing around 4,000 on an identical follower count. Those are different magnitudes, and the difference is the first clue: the decline is not one thing happening uniformly. It is several things, and they hit account types differently.
That unevenness is the whole reason a generic 'reach is down' post is useless to you. Some of your lost reach is a deliberate throttle you cannot optimize away (company-page suppression). Some is a quality filter you can pass by changing what you publish (AI-generic deprioritization). And a large share is a structural change in how distribution is decided at all — the shift most people never see, which quietly rewrote the relationship between follower count and reach. Separating those causes is the difference between chasing tactics that can't work and changing the one thing that can. So before the fixes, the machine.
For most of LinkedIn's history the feed ran on a relationship graph. It showed you content from the people and pages you were connected to, ranked by a stack of separate models scoring engagement likelihood. The defining property of that system, for a creator, was that reach scaled with your network: more followers mechanically meant a larger first audience, and the game was to grow the number and post often enough to stay in front of it. That is the model almost every 'LinkedIn growth' habit was built for, and it is the model that is gone.
In late 2024 LinkedIn began rolling out 360Brew, an in-house foundation model that replaces the old patchwork of ranking systems with a single unified model that evaluates a post's relevance to each individual viewer. LinkedIn has described 360Brew as a large decoder-only foundation model built for personalized ranking and recommendation across its surfaces. The technical detail matters less than the consequence: instead of asking 'is the viewer connected to the author, and will connected people engage,' the system increasingly asks 'does this specific viewer care about this specific topic, based on everything they engage with.' That is the shift from a relationship graph to an interest graph, and it is the root cause underneath most of the reach loss.
The interest graph decouples follower count from reach. A post is now eligible to be shown to interested strangers who never followed you, and — the flip side that stings — it can be withheld from your own followers if the model doesn't read it as relevant to them. This is why the decline looks so different across accounts. A large account that kept posting broad, unfocused content the old way watched reach fall, because raw follower count stopped doing the work it used to. A smaller account posting sharply about one domain sometimes saw reach hold or climb, because the interest graph could confidently match its posts to an interested audience. Reach stopped being a function of how many people follow you and became a function of how legibly the model can classify what you are about and who wants it.
The interest-graph shift is structural and neutral — it redistributes reach rather than uniformly cutting it. But LinkedIn also made three intentional choices that pull reach down for specific kinds of posting, and it is worth naming them separately because your response to each is different.
The steepest declines land on company pages, and that is by design. LinkedIn is a business with an ads product, and holding down organic page reach pushes brands toward sponsored distribution to reach their own followers. This is the throttle you cannot out-post: no cadence or format trick restores organic page reach to 2021 levels, because the suppression is a business decision, not a quality judgment. The rational response is not to fight the page algorithm but to move your organic ambition to where the interest graph actually rewards it — individual people — and treat the page as a credible home base and a paid channel rather than your primary organic engine.
360Brew is also used to identify and reduce distribution of low-value, generic content, and the era of AI-drafted, template-shaped LinkedIn posts made that a priority. Content that reads as formulaic — the recognizable AI cadence, the hook-line-list-CTA skeleton everyone copied, the recycled 'insight' with no specific point of view — gets suppressed. This is a quality filter, which means it is passable: the deprioritization targets the generic shape and empty substance, not the use of AI to help produce a post. LinkedIn itself has published guidance on how professional content earns visibility, and the throughline is originality and genuine expertise, not abstinence from tools. The related crackdown on outright automated posting is a separate, sharper line covered in LinkedIn's AI-content detection and automation crackdown; this throttle is the softer, everyday one that quietly costs reach on bland posts.
In a June 2026 ranking update, LinkedIn reprioritized content-relevancy signals, conversational context, and creator-reliability metrics — and in doing so cut distribution for three specific behaviors: engagement-bait ('comment YES below'), recycled or duplicated content, and inconsistent posting. The reliability signal is the one creators underrate. The model rewards accounts that show up recognizably and regularly on a topic, and penalizes stop-start posting, because consistency is part of how it builds confidence about what you are and who to show you to. Two other well-documented penalties belong in the same bucket: reaction-bait that farms likes without conversation, and external links in the post body — studies of the 2026 feed find a link in the body cuts median reach meaningfully (one widely-cited analysis put it near 19%), because LinkedIn suppresses posts that try to send people off-platform.
Sorting the causes gives you an honest recovery map. The company-page throttle is not recoverable organically; that reach moved behind the ads product, and the mitigation is to shift organic effort to people and reserve the page for what it is still good at. The interest-graph shift is not reversible either, but it is not really a 'loss' to recover — it is a new distribution logic to align with, and accounts that align with it can reach more people than the old relationship graph ever sent them. The genuinely recoverable losses are the quality-and-behavior throttles: generic AI content, engagement-bait, recycling, inconsistency, and in-body links are all things you control, and fixing them lifts reach on a timescale of weeks, not never.
The strategic error is to spend your energy on the irreversible bucket — grinding to grow follower count, or trying to hack company-page reach — while ignoring the recoverable one. Under the interest graph, follower count is the lever that matters least of the ones you can pull. Topical focus, native format, conversation, and consistency are the levers that matter most, and they are the ones you fully control.
Pull the causes together and a clear content profile falls out — not a list of hacks, but the shape of a post the 2026 machine was built to distribute.
Because reach now flows through topic-to-person matching, the single highest-leverage move is to be legibly about something. Post recognizably about the same domain, and make your profile — headline, experience, past posts — align with what you publish, so 360Brew can classify you and route your posts to the right interested audience. Scattershot posting across unrelated subjects starves the model of the signal it needs, and diffuse accounts are exactly the ones that lost the most reach. Choosing and holding a lane is now a distribution decision, not just a branding one.
The feed rewards content consumed in-feed and penalizes attempts to route people away. That favors native text posts, document carousels, and — most of all — video, which LinkedIn has pushed aggressively and which continues to see strong year-over-year view growth. The in-body external link penalty is the mirror image of the same preference. The practical rule: deliver the value inside the post, and if you must link, put it in the first comment rather than the body.
The June 2026 update elevated conversational context, which means comments — especially substantive back-and-forth — count for more than likes, and reaction-bait counts for less. Posts that pose a real question, take a defensible position, or invite genuine disagreement travel further than posts engineered for a quick tap. And because creator-reliability is now a ranking input, a steady cadence beats sporadic bursts: the account that posts thoughtfully and consistently accrues the reliability signal that a stop-start account never builds. The deeper shift toward reply-driven distribution is worked through in LinkedIn's feed shift toward replies and comments, and the broader out-of-network reach dynamics of the interest graph in LinkedIn's AI-driven content playbook for out-of-network reach.
Every signal above compounds on personal profiles and works against company pages. The durable organic pattern for a brand is founder-and-employee-led content: real people posting in their own voice about the company's domain, each one a topically-focused node the interest graph can route. That is also where genuine expertise reads as expertise — the specific, opinionated, experience-backed post the model and the audience both reward, and the generic corporate broadcast neither does. The full modern approach to using the feed this way is in the new LinkedIn content playbook, and the wider saturation context — why generic content is losing reach across professional feeds — in how saturated LinkedIn and X really are.
Look at that content profile and notice what it actually demands. Not one polished post a week to broadcast to a follower base, but a steady stream of topically-focused, native, on-brand posts — text, carousels, and especially video — from real people, on a reliable cadence, each specific and opinionated enough to pass the quality filter and spark conversation. The old relationship-graph game rewarded a large network and occasional posting; the interest-graph game rewards presence and focus produced consistently. That is a production problem as much as a strategy one, and it is where most creators and teams stall — they understand the new rules and simply cannot sustain the output the rules require by hand, especially the video and the per-person volume that a founder-and-employee-led approach implies.
Best-time-and-cadence discipline helps you spend that output well (see best time to post on LinkedIn), but timing only matters if you have the content to schedule. The bottleneck almost always turns out to be throughput: the strategy is clear, and the volume of native, focused, on-brand assets it needs is more than a small team can produce manually week after week.
Kompozy is a full AI content generation and multi-platform publishing engine, and its relevance to declining LinkedIn reach is specific: it makes the content profile the interest graph rewards producible at the volume and consistency the interest graph requires. It does not buy you reach or trick the algorithm — nothing does — and it does not restore company-page organic reach, which LinkedIn moved behind ads on purpose. What it removes is the throughput wall that stops most people from executing the strategy this guide lays out.
The trap to avoid is 'AI content,' which is exactly what LinkedIn deprioritizes — so the way you use the engine matters. Brief it once on your actual expertise and point of view through a Persona Brief that governs voice, and generate a focused, native spread from that single input: text posts and document-style Carousel Posts on your one topic, and Persona Shorts and other avatar video for the video format LinkedIn is pushing hardest — 18 output formats in total, so the same expert angle ships as text, carousel, and video without three separate manual efforts. The Persona Brief holds voice and specificity steady so the output reads as a real person with a point of view rather than the generic template the model penalizes, and HyperFrames keeps every asset visually on-brand. The engine is a way to produce a focused expert's real output at volume — the opposite of the bland, recycled posting that lost reach in the first place.
Then Autopilot publishes and schedules that spread on a reliable cadence — the creator-reliability signal the June 2026 update rewards — across LinkedIn and the seven other social platforms plus blog and email from one queue, behind a per-post review gate so a human signs off before anything ships. Read the boundary honestly: Kompozy is the production-and-distribution layer for the content the interest graph rewards, not a reach cheat and not a fix for the company-page throttle. Where it earns its place is the constraint this guide ends on — the new LinkedIn rewards focused, native, consistent output from real people, and that is a volume problem before it is a strategy one. An engine that lets a person or a small team actually sustain that output is what turns understanding the reach decline into reversing your share of it.
LinkedIn reach declined in 2026 because the platform rebuilt its distribution engine, not because it simply dialed reach down. The 360Brew foundation model and the move from a relationship graph to an interest graph decoupled follower count from reach, so accounts that kept posting the old way fell while focused accounts held or grew. Layered on top are three deliberate throttles: company pages suppressed toward ads (irreversible organically), generic AI content deprioritized (fixable by being specific), and the June 2026 penalty on bait, recycling, inconsistency, and in-body links (fixable by changing behavior). Spend your effort on the recoverable losses, not the follower count that no longer drives reach. The content that still travels is topically focused, native, conversation-driving, and consistent — from real people, not company broadcasts. That profile is clear; producing it at the volume and cadence the interest graph rewards is the part most teams still have to solve, and it is a production problem, not a mystery about the algorithm.
Mostly because LinkedIn rebuilt how it distributes content. In late 2024 it replaced separate ranking models with 360Brew, a single in-house foundation model, and over 2025–2026 shifted the feed from a relationship graph (content from people you know, so reach scaled with follower count) to an interest graph (a post is shown to whoever the model predicts cares about the topic, connection or not). That decoupled follower count from reach, so accounts that kept posting the same way saw declines. On top of the structural change, LinkedIn deliberately throttles company-page posts to push brands toward ads, deprioritizes generic and template-shaped AI content, and — in a June 2026 ranking update — cut distribution for engagement-bait, recycled posts, and inconsistent posting.
It depends heavily on account type. Analyses through early 2026 put company-page organic reach down roughly 60% since 2024, with the average company post now reaching only a low-single-digit share of the feed (several trackers put company pages at just 1–2% of the content in a typical feed). For the average professional, views are down about half from their 2024 peak — a post that reached 10,000 people in 2024 now commonly reaches around 4,000 on the same follower count. Personal profiles substantially outperform company pages, which is why the reach gap between the two has widened, not narrowed.
360Brew is LinkedIn's in-house foundation model for ranking and recommendation, rolled out from late 2024 to replace a patchwork of separate ranking systems with one unified model. It evaluates a post's relevance to each viewer directly rather than leaning on who follows whom, which is the mechanism behind the shift to an interest graph. The practical effect is that reach now tracks topical relevance and demonstrated expertise more than follower count — so a focused post can reach interested strangers, and an unfocused one can underperform even to your own followers.
It means follower count and reach are largely decoupled, not that followers are worthless. Under the old relationship graph, a bigger network mechanically meant more reach. Under the interest graph, 360Brew shows a post to whoever it predicts is interested in the topic, so a tightly-focused account with a few thousand engaged followers can outperform a large but unfocused one. Followers still matter as a warm base and a trust signal, but consistent topical focus — posting recognizably about the same expertise so the model can classify you — is now the bigger lever than raw follower volume.
Content the interest graph can confidently classify and that people genuinely engage with. In practice: posts with a clear topical focus that matches your profile and history so 360Brew knows what you are about; native formats that keep people on-platform (text, document carousels, and especially video, which LinkedIn has pushed hard); posts that spark substantive comments rather than reaction-bait; and a consistent cadence so the model treats you as a reliable creator. Generic AI-sounding content, recycled posts, engagement-bait, and posts with an external link in the body (studies find a link in the body cuts median reach meaningfully) all get suppressed.
For organic reach, mostly yes — but as a shift in emphasis, not an abandonment. Personal profiles consistently outperform company pages in the 2026 feed because LinkedIn throttles page reach to push brands toward paid distribution, and because the interest graph rewards individual expertise and voice. The durable pattern is employee-and-founder-led content: real people posting in their own voice about the company's domain, with the company page used for the things it is still good at (a credible home base, hiring, proof, and paid amplification) rather than as the primary organic reach engine.
LinkedIn reach declined in 2026 mainly because the platform rebuilt its distribution engine, not because it simply turned reach down. From late 2024 it replaced separate ranking models with 360Brew, a single in-house foundation model, and shifted the feed from a relationship graph — where reach scaled with follower count — to an interest graph, where a post is shown to whoever the model predicts cares about the topic. That decoupled followers from reach. On top of it, LinkedIn throttles company pages toward ads, deprioritizes generic AI content, and (June 2026) cut reach for engagement-bait, recycled posts, and inconsistent posting. The content that still travels is topically focused, native, consistent, and conversation-driving.
Get started → · ← All guides · Compare Kompozy vs other tools