LinkedIn shadowban: what is documented, what is folklore, and how to tell which you have

LinkedIn has never acknowledged a shadowban, and the term does not appear anywhere in the material the company publishes. We read the Professional Community Policies, the help pages on restricted accounts and removed content, the transparency community report, the pressroom announcements about the feed, and every engineering blog post tagged Feed. None of them uses the word. What LinkedIn does document is two different things that get merged into it: visible enforcement, which comes with a notification and an appeal, and published demotion criteria, which reduce a post's distribution without telling you but are not secret and not account-wide. Almost everything people call a shadowban is neither.

Has LinkedIn ever acknowledged a shadowban?

No, and it has come closer to addressing the underlying suspicion than most people realise.

In November 2025 LinkedIn's engineering team published a post specifically about members who believed their reach was being suppressed. The trigger was a wave of side-by-side tests on gender, but the reasoning applies to every suppression theory. LinkedIn's statement: “Our algorithm and AI systems do not use demographic information (such as age, race, or gender) as a signal to determine the visibility of content, profile, or posts in the Feed.” And on the tests themselves: “A side-by-side snapshot of your own feed updates that are not perfectly representative, or equal in reach, doesn't automatically imply unfair treatment or bias.”

The same post describes what LinkedIn checks before shipping a ranking model. It measures creator allocation, which is “how posts from different creators are ranked when they compete for the same real estate,” specifically testing whether “given similar content engagement and quality, any one group is being systematically ranked lower relative to another.” The March 2026 engineering post on the new Feed repeats the commitment, saying LinkedIn audits “that posts from different creators compete on equal footing.”

You do not have to take those statements at face value to notice what they are: a company going out of its way to describe systematic ranking suppression as something it tests against. That is the opposite of a company quietly operating one.

What do people actually experience when they say shadowban?

The symptom lists on shadowban pages are remarkably consistent, which is worth taking seriously as a description even where the diagnosis is wrong. Here they are with the mundane explanation next to each, and a test that distinguishes the two.

Only the last row describes something LinkedIn documents. The others are describing a ranking system that a member cannot see inside of, which is a genuinely uncomfortable situation and not the same as being punished.

Where do the shadowban percentages come from?

We traced the numbers on pages ranking for this term. The results were consistent enough to state flatly.

One widely linked glossary page describes a shadowbanned account as one where “content that previously reached 500 to 2,000 people suddenly gets 30 to 50 impressions,” identifies a weekly impressions drop of “70 percent or more” as the most reliable indicator, and describes a normal connection acceptance rate as “30 to 40 percent” falling to single digits under suppression. None of these figures carries a citation. LinkedIn does not publish acceptance-rate benchmarks, does not publish impression-drop thresholds, and does not publish anything from which those ranges could be derived. We could not find a primary source for any of them.

This matters more than it might seem, because the numbers are what make the pages feel authoritative. A reader who is told that a 70% drop over two to three weeks “almost certainly points to a shadowban” now has a diagnostic threshold, and a 70% drop over three weeks is an ordinary experience for anyone whose last three posts were link shares. The precision is doing work that the evidence cannot support.

The word itself arrived on LinkedIn from other social platforms, where it described specific product behaviours. Imported to LinkedIn, it became a general-purpose explanation for any decline, which is exactly the kind of theory that cannot be falsified and therefore cannot help you.

What enforcement does LinkedIn actually document?

A great deal, and all of it visible. LinkedIn's help page on restricted or removed accounts and content lists four categories:

• Content violations.LinkedIn states that “some violations of our Professional Community Policies may result in permanent account restriction after a single violation.”
• Profile violations. Elements such as photos can be removed if they do not comply with policy, with access restricted for repeated violations.
• Identity violations.If a profile is “intentionally fraudulent or does not reflect your true identity,” the account can be restricted, with identity verification as the route back.
• Automated tools violations.“Automated inauthentic activity violates the LinkedIn User Agreement and can result in temporary or permanent restriction.”

The scale is published too. LinkedIn's community report for July to December 2025 says automated defences blocked 97.8% of the fake accounts stopped in that period, and that 99.7% of fake accounts were stopped proactively, before any member reported them. For spam and scams, which the report calls “by far the most common type of inappropriate content we take action on,” automated defences accounted for 98.6% of removals. On copyright, LinkedIn received 2,955 requests covering 3,723 reported infringements, removed 3,530 and rejected 193, an acceptance rate of 95%.

Two things stand out. First, the enforcement machine is overwhelmingly automated and overwhelmingly aimed at fake accounts and spam, not at ordinary members posting about their work. Second, every one of these actions is a removal or a restriction: an event with a record, a notification and, in the report's own words, the ability for members to “ask us to take a second look.”

Does LinkedIn reduce reach without telling you?

Yes. This is the strongest version of the shadowban intuition and it deserves a straight answer rather than a denial.

LinkedIn's March 2026 pressroom post lists things it is actively reducing in the feed: engagement pods (“working to make engagement pods ineffective”), comment automation and unauthorized third party tools, engagement bait such as “comment to agree” prompts and videos that do not match the text, and “repetitive, low-substance posts.” Nobody gets an email when a post is ranked lower for any of these. So demotion without notification is real.

The difference from a shadowban, and it is the whole difference, is that the criteria are published and the effect is per-post rather than per-account. Here is the same distinction as a table.

If your posts stopped working and you have never used a pod, an automation tool or a “comment YES below” call to action, then nothing on rows one to three applies to you and row four is where to look. That is the subject of our step-by-step diagnosis of why a LinkedIn post gets no views.

How do you check whether you are actually restricted?

In this order, because each step is cheaper than the last and rules out more.

1. Check your notifications and the email on your account. LinkedIn communicates restrictions. If nothing was sent, nothing was actioned. This single step resolves the great majority of cases.
2. Try to post, comment and message. Restrictions that affect distribution almost always affect capability too. If everything still works, you are not restricted.
3. Ask one person to check a comment. Leave a comment on a public post, then have a colleague open that post in their own account and look for it. This is the only reliable test for the hidden-comments claim, and it usually comes back fine because comment ordering, not visibility, was the issue.
4. Check your public profile settings before trusting a logged-out search. The incognito search test that shadowban guides recommend is a poor test: public profile visibility and guest controls change the result, and search engines index on their own schedule. Verify the setting first or the test tells you nothing.
5. Compare surfaces. Genuine account-level problems hit posts, comments, search and messaging together. A problem confined to post impressions is a content and ranking problem.
6. Look at your last 20 posts, not your last 2.Median impressions, median comments, grouped by format. Almost every “shadowban” resolves into a format shift or a topic shift once you tabulate it.

One thing not to do: pay anyone for shadowban removal or account warming. There is no documented state for those services to remove, and the tools involved fall squarely into the automated activity category that produces the restrictions people are worried about.

Does LinkedIn shadowban you for using AI to write posts?

This question now shows up in every discussion of the topic, and the honest answer is narrower than either camp claims. We found no LinkedIn policy, help page or announcement that prohibits or penalises AI-assisted writing. The Professional Community Policies do not mention it. The March 2026 feed announcement does not mention it. What those documents do target is a different thing that sounds similar.

Read LinkedIn's language closely and the line it draws is about operation, not authorship. The pressroom post names “comment automation, engagement pods, and unauthorized third-party tools.” The help page names “automated inauthentic activity” as a User Agreement violation carrying temporary or permanent restriction. Those describe software acting as you: firing connection requests, spraying comments, operating accounts at machine speed. None of them describes a person using a model to help draft a post they then read, edit and publish.

Where AI genuinely does interact with reach is through the quality criteria LinkedIn has published, and it is not flattering. “Repetitive, low-substance posts” is exactly what unedited generated text tends to be, and generic advice is the thing semantic retrieval has the least reason to surface to anyone in particular. The risk is not detection. The risk is producing the sort of post the system was rebuilt to show less of. Our page on AI LinkedIn post generators covers what that means in practice for anyone using one.

What are the mundane causes people mistake for a shadowban?

Ordered roughly by how often they turn out to be the answer.

1. More competition.LinkedIn stated in November 2025 that daily content volume “has grown rapidly over the past year, which means more competition for attention.” A constant post quality against a rising supply produces falling reach with no change on your side.
2. A quiet format shift. Drifting into link shares and stock images is the single most common invisible change. In our data those two formats have the lowest median engagement rates in the cohort.
3. A topic shift. Retrieval is semantic. Moving to a new subject means being matched against a new audience that has no history with you.
4. A dormant network.If you stopped commenting on other people's posts, you stopped appearing in the places that put you back in front of them.
5. Engagement bait phrasing.Named explicitly by LinkedIn. “Comment AGREE if you...” is on the published reduce list.
6. Reposting the same material. Also on the published list, as recycled and repetitive content. Our page on whether reposting on LinkedIn hurts reach covers what is and is not established there.
7. Comparing yourself to the wrong baseline. A post that would have been your median a year ago feels like a failure after one outlier.

What does the data say separates weak posts from strong ones?

We scored 12,988 English LinkedIn posts by engagement rate and compared the top decile against the bottom half. If your account has gone quiet, these are the differences worth auditing before you reach for an enforcement theory.

Format tells the same story. Native video accounted for 26.3% of top-decile posts and only 10.1% of bottom-half posts, over-indexing about 2.6 times. Shared articles and links ran the other way: 22.1% of the top decile against 32.7% of the bottom half. Hashtag stuffing was about 1.6 times more common at the bottom, which is why we recommend cutting them in our piece on whether LinkedIn hashtags still work.

And the widest gap of all is comments. In our study of 34,000 LinkedIn posts, the median top-decile post drew 72 comments against 5 for the median bottom-half post, a far bigger separation than reactions (235 against 80). An account that used to get conversations and now gets scroll-past reactions has a content signal to work with, not an enforcement mystery.

What should you do if you really are restricted?

1. Read what LinkedIn actually sent you. The category matters: content, profile, identity or automated tools each have a different route back.
2. Use the appeal.LinkedIn's help page tells restricted members to log in and follow the onscreen prompts to ask the company to revisit the decision, and says it will review and report the outcome.
3. Verify your identity if that is the category. Identity restrictions have a specific verification path and no amount of posting will resolve them.
4. Disconnect every automation tool immediately. This is the category most likely to recur, and the User Agreement language is explicit about temporary or permanent restriction.
5. Do not open a second account.Identity and fake account enforcement is the most automated part of LinkedIn's stack, at 99.7% proactive detection. A second profile is the fastest way to turn a temporary problem into a permanent one.

What should you do if you are not restricted, which is likely?

Treat it as a content and distribution problem and give it four weeks:

• Pick one topic and stay on it. Semantic retrieval rewards an account that is easy to characterise.
• Change the format mix. Fewer link shares, more text and native video.
• Rewrite your openings. First person, direct address, no hashtag block. Our guide to LinkedIn hooks covers what the top decile does differently.
• Comment properly on ten posts a day. This rebuilds the relationships that put you back in feeds, and it is the fastest of these levers.
• Measure medians, not moments. Twelve posts, then compare.

Four weeks is the right unit because anything shorter cannot distinguish a real change from normal variance. In our cohort the gap between a median post and a top-decile post is roughly fifteenfold on engagement rate, which means a run of three disappointing posts is statistically unremarkable even for accounts that are doing everything right. Most people who conclude they have been shadowbanned reached that conclusion inside a week, from a sample of two or three posts, against a baseline they never wrote down.

If the decline is real, sustained and account-wide, the diagnostic sequence in our piece on why LinkedIn reach drops goes through the causes in order, and impressions versus reach explains which counter you should be watching before you decide anything moved at all.

So is there such a thing as a LinkedIn shadowban?

Not as a documented product behaviour. LinkedIn publishes its enforcement categories, its appeal routes, its removal volumes and its demotion criteria, and none of that material describes a silent account-level throttle. What it does describe is a ranking system that picks a handful of posts per viewer out of millions, a content supply that is growing faster than attention, and a short list of tactics whose distribution the company is deliberately reducing. The percentages that dominate search results for the term LinkedIn shadowban come from nowhere we could trace. Check your notifications, check your comments with a colleague, tabulate your last 20 posts by format, and in nearly every case you will find something you can actually fix.