Why LinkedIn company page reach collapses after 90 days

LinkedIn company pages are losing reach at a rate most marketing teams have not accounted for. Between 2024 and early 2026, organic reach on company pages dropped 60 to 66%, and the average post now reaches roughly 1.6% of a page's own followers. For many teams, the decline feels sudden because it follows a predictable pattern: broad initial distribution that contracts sharply around the 90-day mark. That contraction is structural, driven by how LinkedIn's 360Brew recommendation system calibrates a topic-authority profile for each page. Understanding the mechanism is the first step to responding to it.

Why does LinkedIn company page reach drop after the first 90 days?

The 90-day drop is not a LinkedIn penalty in any conventional sense. It is the end of a temporary distribution window that LinkedIn's 360Brew recommendation system runs on every page during its first roughly 90 days of consistent posting. During that window, the algorithm grants broader-than-normal reach while it collects enough post data to build a topic-authority profile. When calibration ends, distribution contracts to whatever level the page's engagement history justifies.

What 360Brew does during those 90 days is evaluate semantic consistency. It reads the content of each post and tracks how tightly the page stays within a defined subject domain. LinkedIn recommends that creators stick to 2-3 core professional topics for at least 90 days before the system assigns stable distribution. That recommendation is not a general best practice. It is a description of how the algorithm works.

Pages that stay within 2-3 topic clusters receive stable, predictable distribution once calibration ends. Pages that range across 4 or more clusters get classified as semantically ambiguous, which translates directly into contracted reach. The collapse is not a drop from a stable baseline. It is the termination of a novelty period that precedes stable scoring.

Cross-account data from SocialNexis confirms this. Pages posting across 4 or more topic clusters in their first 90 days land at a reach floor 3-4x lower than pages that hold to 2 clusters throughout the same period. This pattern holds consistently enough across account types and industries that we treat it as a calibration outcome rather than an account-level flag.

The practical implication is counterintuitive for most content teams. The first 90 days on a LinkedIn company page are not an opportunity to test content variety and learn what resonates. They are the window that determines a page's baseline distribution for the next phase. Experimenting with off-topic content during calibration is not free. It costs reach for months afterward.

360Brew and the LinkedIn company page growth stall of 2026

360Brew is a 150-billion-parameter generative recommender model. LinkedIn built it on Meta's LLaMA 3 architecture, fine-tuned it on proprietary LinkedIn data, and announced it on March 12, 2026, in a LinkedIn Engineering Blog post by Hristo Danchev. It did not incrementally improve the existing feed ranking system. It replaced five separate retrieval pipelines with a single generative model that makes unified distribution decisions.

What distinguishes 360Brew from prior LinkedIn ranking systems is the input scale. Before serving a post, the model processes more than 1,000 of each viewer's past interactions as a chronological sequence. That sequence includes which posts they paused on, which they skipped, which they commented on, and how long they spent in each content type. The model then predicts what that specific viewer wants next, making that prediction separately for every candidate post it evaluates.

Company page content enters this system at a disadvantage that has nothing to do with post quality. A personal profile carries a professional expertise signal derived from work history, endorsements, and connection patterns. A company page entity carries none of that. It has a logo, a description, and a follower count. That difference matters because 360Brew weights the expertise signal when deciding how much credibility to assign to content before it even measures engagement.

The result is that a company page post and an identical personal profile post are not evaluated on equal terms. The page entity must earn every unit of distribution through engagement quality alone. The personal profile gets a head start from the platform's expertise modeling. This entity-level scoring gap, compounded by the 90-day calibration timeline, is the structural cause of the company page growth stall that became widespread in early 2026 as 360Brew rolled out fully.

Company pages now reach 1.6% of their own followers per post

Company page organic reach fell 60-66% between 2024 and early 2026, measured against the same pages' prior performance on the same platform. The average company page post now reaches roughly 1.6% of the page's own followers. That figure is not a worst-case outcome for underperforming pages. It is the median.

Platform-wide data from the 360Brew rollout period tells a similar story at scale: a 47% year-over-year visibility decline, a 39% engagement decline, and a 42% follower growth decline across LinkedIn accounts of all types. Company pages sit below the cross-account average on all three metrics.

The feed allocation explains part of the mechanism. Company page content occupies approximately 1-2% of a typical LinkedIn user's feed. Personal profiles account for roughly 65% of that same feed. These proportions reflect how 360Brew classifies content entity types before any engagement signal is applied, not a consequence of engagement differences between the two entity types.

The profile-versus-page gap shows up most clearly in comparative reach data. Personal profiles sharing identical content generate 561% more reach than company pages, 2.75x more impressions, and 5x more engagement, even when the personal profile has fewer followers than the company page it is being compared to. The content is the same. The entity type changes the outcome.

Most company page benchmarks in use today predate the 360Brew rollout and overstate realistic performance. A post reaching 1.6% of followers is performing at approximately the current platform median for company page content, not underperforming relative to peers. Teams that plan around pre-2026 benchmarks are building models on numbers that no longer describe the platform.

Inside the 60-to-90-minute window that decides 70% of your reach

Every company page post enters a two-stage distribution test the moment it publishes. LinkedIn exposes it to 2-5% of the page's followers in the first 60-90 minutes and watches what happens. It measures engagement velocity: likes per minute, comment response speed, and saves. Based on those measurements, it decides whether to expand or permanently suppress the post.

Approximately 70% of a post's total reach is determined in that initial window. A post that clears the velocity threshold expands to a broader audience. One that does not is permanently suppressed. The algorithm does not revisit that decision when engagement arrives hours later. The window closes, and the outcome is fixed.

External links embedded in a post body reduce median reach by 18.8%. The common workaround of placing the link in the first comment is also penalized as of early 2026. LinkedIn's link penalty applies to the full post context, not just the body text, so both placements carry a distribution cost. The answer is to remove the link from the post entirely and direct traffic through a non-penalized path.

Posting time carries more weight than most content calendars account for. The golden window captures the behavior of whoever is online in that first 60-90 minutes. If a post goes live when the page's most engaged followers are not active, the initial sample is weak and the velocity signal is lower, even when the content itself is strong.

The follower-base problem compounds all of this. A page with a small follower count faces a structural ceiling: even well-crafted content can fail the velocity threshold purely because the pool of potential early engagers is too small to generate the required signal. This is why content quality improvements alone often fail to move reach numbers for newer pages. The bottleneck is not the content.

What the standard company page advice gets wrong about the invitation cap

LinkedIn reduced free company page invitation credits from approximately 250 to 50 per month in a staggered 2026 rollout. Premium Company Pages receive 300 credits per month at $1,199.88 per year. Credits are shared across all page admins, apply only to first-degree connections of those admins, and rejected or withdrawn invitations are permanently lost. There is no refund mechanism.

Most advice treats this as an inconvenience: fewer invitations means slower follower growth, so plan accordingly. That framing misses the compounding effect. With only 50 monthly credits, a new page can add at most 10-30 net followers per month after accounting for rejections and withdrawals.

At that growth rate, the engagement pool never reaches the size needed to generate the volume of comments and saves required to clear the golden window velocity threshold. A page can produce good content and still fail the distribution test consistently, not because the content is weak, but because the pool of followers available to engage in the first 60-90 minutes is too small to produce sufficient velocity signal.

SocialNexis operator data shows this as a page that "feels" stuck even when content quality objectively improves. Post after post earns low reach. The team improves the creative. Reach stays low. The team tries different topics. Reach stays low. The bottleneck is follower base size, not content strategy, but nothing about the platform interface makes that diagnosis visible.

The invitation cap and the reach collapse interact as a compounding trap rather than two separate problems. Fewer credits produce a smaller follower base, which generates smaller engagement pools in the golden window, which produces lower velocity signals, which produces suppressed reach, which slows organic discovery, which further slows follower growth. Each constraint tightens the others.

Depth Score, dwell time, and why likes no longer drive LinkedIn company page reach

LinkedIn's Depth Score is a 2026 ranking signal that changed which engagement types move distribution. The score measures dwell time, comment depth, saves, shares to private messages, and content completion rate. Raw reaction counts carry far less algorithmic weight than before. The platform shifted from measuring how many people clicked something to measuring how much attention they gave it.

The weight differences are significant. A substantive multi-sentence comment carries approximately 15x the algorithmic weight of a simple reaction. A save carries roughly 5x the weight of a like. Shares to private messages also feed into the score. A post generating 200 reactions but no substantive comments or saves is algorithmically weaker than a post generating 20 reactions, 8 substantive comments, and 5 saves.

Company page posts must generate higher dwell time and comment depth than equivalent personal profile posts to achieve the same distribution expansion. This is the Depth Score manifestation of the entity-level gap. The page entity type does not receive the expertise-signal boost that personal profiles get from 360Brew, so the page must compensate for that missing signal entirely through measured viewer behavior.

Practitioners running both a company page and a personal profile from the same dashboard can observe this gap directly in side-by-side post analytics. SocialNexis cross-account analysis shows that identical carousels posted on a company page require meaningfully more comment depth than the same carousel on a personal profile to achieve equivalent reach expansion. The content is the same. The entity type absorbs a base discount before format effects are applied.

The shift this requires in content design is specific. Calls to action that prompt one-click engagement do not move the Depth Score. Calls to action that prompt readers to describe their own experience, name a challenge they face, or explain their position generate the multi-sentence responses the algorithm rewards. That is not a subtle difference in tone. It is a different content objective.

When employee reshares compound your page reach and when they actively shrink it

Despite only about 3% of employees typically sharing company content, those shares generate roughly 30% of total company engagement. Employee advocacy is structurally more efficient per post than the company page itself, and the gap is large enough that managing it well or poorly materially changes overall reach.

360Brew complicates this. The system detects and suppresses near-duplicate content: employees sharing word-for-word identical company page copy trigger the platform's deduplication layer, which reduces reach for both the employee post and the originating company page's distribution signal. The content gets penalized twice. Templated advocacy programs that distribute the same text to multiple employees actively harm both entities.

LinkedIn has not publicly disclosed exactly how much copy variation clears the deduplication threshold. What is clear from the platform's behavior is that employees need to write their own framing around the core message rather than copying the company page text verbatim. A reshare that preserves the facts but uses the employee's own phrasing is treated as distinct content. A direct copy-paste is not.

Timing is the variable that standard advocacy advice omits entirely. Employee amplification within the first 30 minutes of a company page post produces the strongest combined Depth Score signal. SocialNexis automation data shows this clearly: reshares that arrive in the golden window contribute to the velocity measurement and help the original post clear its distribution threshold. Reshares that arrive 4 or more hours after posting arrive after the golden window has closed and do not meaningfully expand the original post's distribution, regardless of how strong the engagement on the reshare itself is.

An effective advocacy approach treats the company page post and each employee post as distinct signals that together build velocity. The company page publishes first. Within 30 minutes, employees who have drafted their own framing reshare to their networks. The combined engagement from both entities, measured separately by 360Brew, produces a stronger aggregate depth signal than either entity achieves alone. This requires coordination infrastructure. A 'please share this' Slack message is not that infrastructure.

Rebuilding LinkedIn company page growth after a plateau

Recovery begins with a content audit focused on topic scope, not content quality. The first question is how many distinct topic clusters the page has posted across in the last 90 days. If the answer is more than 3, the page is likely classified as semantically ambiguous by 360Brew, and format optimization will not fix a misclassified entity. Narrow the scope to 2-3 core subjects and hold that focus for at least 60-90 days to re-enter the topic-authority calibration cycle on stable ground.

Format selection matters during recovery because Depth Score rewards content that generates dwell time and comment depth, and some formats generate those signals more reliably than others. Educational carousels of 8-12 slides generate a 6.6% engagement rate compared to 4.85% for single static images. Native PDF documents reach 5-10x more viewers than typical text posts. These are not marginal differences.

Portrait-orientation video under 30 seconds has 200% higher completion rates than longer video formats. Content completion rate feeds directly into the Depth Score, making short portrait video one of the most efficient formats for building the signals the algorithm uses to expand distribution. Landscape video and longer formats both underperform on the completion metric.

Cadence matters independently of format. Posting daily on LinkedIn causes a 26% drop in average reach per post and a cumulative 45% negative impact over time. The optimal frequency for company pages is 2-4 posts per week. At that cadence, each post generates approximately 1,234 more impressions than a once-per-week schedule would produce. Daily posting is not a sign of commitment. It is a self-imposed reach penalty.

Embedding a named human face in the post creative is one of the more consistent recovery levers in SocialNexis cross-account analysis. Posts featuring an executive photo, an attributed quote, or an employee spotlight consistently outperform logo-only content on Depth Score signals. Users dwell longer on posts that show a recognizable person and comment with more substantive responses. That behavior feeds back into the Depth Score regardless of whether the post originates from a company page or a personal profile. The page cannot replicate the expertise signal that a personal profile receives from 360Brew, but it can design content that generates the behavioral signals the algorithm measures in its place.

Frequently asked questions