How to Fix a Shadowban on Twitter (X)
How do I fix a shadowban on X?
Stop the behavior that likely triggered it (automation, repeated replies, mass follows, links in most of your posts), then wait. Recovery runs in days rather than hours, and there is no way to speed it up beyond removing the trigger.
X never confirms a restriction and never announces a lift, so the only way to know where you stand, before and after, is to measure it rather than guess. The 4 restrictions below each have their own trigger and their own exit.
Not sure which restriction you have? Test your account first.
Fixing a search suggestion ban
A search suggestion ban means your account stops appearing in the dropdown when someone types your handle, even though your profile still works fine.
- Disconnect third-party apps and automation posting or following on your behalf (Settings, Security and account access, Apps and sessions).
- Stop bulk actions: mass following, mass liking, the same reply sent over and over.
- Confirm a real email and phone number are on the account.
- Re-check by typing your own handle into the search box every day or two; the dropdown showing you again is the only confirmation there is.
What does not help: filing an appeal. There is no appeal form for this one, since it is not a strike on your account, it is a behavioral filter that lifts once the behavior stops.
Fixing a search ban
A search ban removes your posts from X search entirely, even for people who already follow you.
- Cut automation and any third-party posting tool.
- Stop hashtag stuffing and posting the same or near-identical content repeatedly.
- Confirm a real email and phone number are on the account.
- Re-run from:yourhandle in the Latest tab from a second account every day; results reappearing is the ban lifting.
What does not help: deleting the posts that seem to be missing. The ban is not about specific content, it is about the account's current standing in search, so removing old posts just costs you your history for nothing.
Fixing a ghost ban
A ghost ban detaches your replies from the conversations they belong to: everyone else sees the thread minus you.
- Slow down reply frequency and vary the content: identical or near-identical replies are the most reported trigger.
- Cut any automation that replies or engages for you.
- Avoid putting links in most of your replies.
- Re-run the logged-out thread check from a second account every day or two until a reply shows up again.
What does not help: deleting your old replies. The masking applies to how a thread renders today, not to the age of any single reply, so clearing your history does not speed anything up, it only costs you the history.
Fixing reply deboosting
Reply deboosting does not remove your replies, it pushes them to the bottom of the conversation or behind Show more replies, where almost nobody scrolls.
- Cut reply volume and drop the copy-paste replies: high frequency plus similarity causes most of it.
- Avoid rapid-fire replying to large accounts, which draws blocks and reports.
- Cut engagement automation entirely.
- Re-check across 2 or 3 different conversations every few days, since a single low ranking can just be the algorithm and not a restriction.
What does not help: replying more to push through it. More replies in the same pattern reads as more of the exact behavior that likely caused the deboosting, not less.
What causes a shadowban: the 13 causes in X's published code
X open-sourced its recommendation algorithm in 2026, including its visibility filtering and spam labeling systems. The causes below are read from that code rather than from folklore: each one names its trigger, the timer the code actually writes (or the absence of one), and the exit that follows from it. The sections above tell you what to do per symptom; this list tells you why it likely happened.
- Reply spam flag. X's automated systems flag pasted, bursty, or generic replies, and a flagged reply typically loses its visibility beyond the people who already follow you.
- Duplicate content flag. Posts matching a cluster of near-identical text across X can be flagged as spam one by one, with no expiration written in the code.
- Links to flagged domains. A link whose destination carries a bad or unsafe verdict in X's link-safety system can hide that post from non-followers, and a flagged pinned link can label the whole account for 7 days.
- Mass mentions and follow churn. Tagging strangers, especially next to low-quality links, can flag individual posts, and aggressive follow patterns can trigger account challenges.
- Automated action-pattern flag. X runs a behavior model on the sequence of your recent likes, follows, retweets, and replies, and patterns that look automated can cost you visibility regardless of what you post.
- Report ratio flag. An account drawing a high rate of spam reports from strangers, relative to likes from strangers, gets a 7 day label in X's published code, and small accounts sit closest to that line.
- Adult content flags. Flagged adult media is blurred for viewers who have not opted in and dropped from recommendations, and 3 flagged posts among your last 5 within 60 days flag the whole account for 7 days.
- Engagement bait flag. Posts that ask people to reply, tag, follow, or share to unlock something can be flagged and removed from recommendations to non-followers, while your followers still see them.
- Policy strike or suspension. A rules violation can place a reach-limiting label or a suspension on the account, with 30 day timers on some labels and none on others.
- Account security hold. An account flagged as compromised, locked read-only, or marked as possible impersonation has its posts blocked from reaching non-followers until the underlying status changes.
- Small or new account, a built-in limit rather than a sanction. An account under roughly 400 active followers has no community cluster yet in X's published code, so it gets little reach beyond its followers. That is a threshold, not a penalty.
- Predicted audience rejection, a built-in limit rather than a sanction. For each viewer, X predicts the chance they would mute, block, report, or dismiss your post, and strong negative predictions push the post to the bottom for that viewer, with no memory between requests.
- Built-in limits, not a ban, a built-in limit rather than a sanction. Several reach ceilings in X's published code apply to everyone by design, and they explain most scares where every check comes back clean.
What is a reply spam flag on X?
X's automated systems flag pasted, bursty, or generic replies, and a flagged reply typically loses its visibility beyond the people who already follow you.
What triggers it:
- Pasting reply text instead of typing it. The composer reports pasted text to the reply judge, according to X's published code.
- Posting many replies in a short window.
- Generic, templated, or AI-sounding replies that add no specific context.
- Getting blocked by several people within 24 hours, which feeds the same judge.
How long it lasts: X's published code writes two timers on flagged replies: 14 days on one labeling path and a fixed 30 days on another. A third path applies the label with no visible expiration, so no return date is certain.
What does not help: Deleting the flagged reply. Nothing in the published code shows deletion clearing this label. Appealing. No appeal path exists for this label in the published code. Switching to the official app to look cleaner. The signals this judge reads do not change with the app.
The exit: stop the trigger above, then let the timer run. The step-by-step exit protocol for this exact cause, matched to your account's measured signals and re-checked daily, comes with monitoring.
What is the duplicate content flag on X?
Posts matching a cluster of near-identical text across X can be flagged as spam one by one, with no expiration written in the code.
What triggers it:
- Posting the same message repeatedly, copy-paste replies included.
- Many posts or replies with only minor variations between them.
- Template posts built around the same link or hashtag.
- Text that matches what many other accounts are posting. The check clusters text across accounts, not just yours.
How long it lasts: No expiration timer and no removal rule exist for this flag in X's published code. It typically stays until the flagged post is deleted.
What does not help: Editing a flagged post. Nothing in the published code shows an edit triggering a re-check. Deleting one post from a duplicate batch and keeping the rest up.
The exit: stop the trigger above, then let the timer run. The step-by-step exit protocol for this exact cause, matched to your account's measured signals and re-checked daily, comes with monitoring.
Do links to flagged domains cause a shadowban?
A link whose destination carries a bad or unsafe verdict in X's link-safety system can hide that post from non-followers, and a flagged pinned link can label the whole account for 7 days.
What triggers it:
- Posting a link whose destination or redirect chain carries a bad or unsafe verdict.
- A shortener or intermediate redirect that looks low quality, even when the destination is fine.
- Pinning a tweet that contains a flagged or low-quality link.
- Following new accounts while a flagged link is pinned. That action re-triggers the pinned-tweet review, according to X's published code.
- Linking to the same flagged domain repeatedly.
How long it lasts: Most link labels in X's published code carry no expiration and persist until the domain's reputation record changes upstream, which you cannot trigger. The account-level pinned-link label is the exception, with a written 7 day timer.
What does not help: Removing the link after the fact. It stops new labels, but nothing in the published code removes a label already applied. Swapping URLs on the same domain. The verdict is usually about the domain, not one link. Reposting the same link after deleting the old post. Each new post is checked against the domain's current verdict.
The exit: stop the trigger above, then let the timer run. The step-by-step exit protocol for this exact cause, matched to your account's measured signals and re-checked daily, comes with monitoring.
Can mass mentions or follow churn cause a shadowban?
Tagging strangers, especially next to low-quality links, can flag individual posts, and aggressive follow patterns can trigger account challenges.
What triggers it:
- Mentioning accounts that do not follow you while linking to a low-quality domain.
- Two or more @mentions in one post, which opens a dedicated review path in X's published code.
- Rapid follow and unfollow cycles, especially through tools.
- Following someone while a flagged link is pinned to your profile.
How long it lasts: The post label on the mention-plus-link path has no expiration in X's published code and goes away only when the flagged post is deleted. The pinned-link account label expires after 7 days.
What does not help: Mass unfollowing to compensate. It does not undo a label or a challenge. Posting more to look human. The mention and link rules read each post on its own.
The exit: stop the trigger above, then let the timer run. The step-by-step exit protocol for this exact cause, matched to your account's measured signals and re-checked daily, comes with monitoring.
Does automation trigger a shadowban on X?
X runs a behavior model on the sequence of your recent likes, follows, retweets, and replies, and patterns that look automated can cost you visibility regardless of what you post.
What triggers it:
- Bursts of actions less than a second apart.
- Actions spaced at nearly identical intervals, a mechanical rhythm instead of natural browsing.
- Liking, replying to, or following things you never actually saw in your own feed, a classic sign of API-driven activity in X's published code.
- Third-party tools acting on your account.
How long it lasts: The behavior model writes no fixed sanction length in the published code. It re-scores accounts on a rolling basis, so recovery follows the behavior change rather than a timer. The related labels it can feed carry 7 to 14 day timers.
What does not help: Expecting an instant reset after turning tools off. Re-scoring runs on a delay of up to about a day. Asking followers to like your posts to compensate. The relevant spam ratio counts likes from outside your network only.
The exit: stop the trigger above, then let the timer run. The step-by-step exit protocol for this exact cause, matched to your account's measured signals and re-checked daily, comes with monitoring.
What is the report ratio flag on X?
An account drawing a high rate of spam reports from strangers, relative to likes from strangers, gets a 7 day label in X's published code, and small accounts sit closest to that line.
What triggers it:
- A burst of spam reports from people outside your follower circle.
- Very few likes from strangers, which makes even one or two reports weigh heavily on a small account.
- A suspension for spam within the last 30 days.
- Repeating the reported behavior before the 30 day report window clears.
How long it lasts: A fixed 7 day timer, written in the code, for both the post and the account version. Reports also age out of a rolling 30 day window on their own. No early removal path exists.
What does not help: Deleting reported posts. The ratio is tracked per account over a rolling window, so past reports keep counting. Asking your community to like your posts. In-network likes are excluded from the ratio entirely.
The exit: stop the trigger above, then let the timer run. The step-by-step exit protocol for this exact cause, matched to your account's measured signals and re-checked daily, comes with monitoring.
Do adult content flags cause a shadowban?
Flagged adult media is blurred for viewers who have not opted in and dropped from recommendations, and 3 flagged posts among your last 5 within 60 days flag the whole account for 7 days.
What triggers it:
- Posting sexual or suggestive media that X's automated review flags.
- Quote-tweeting someone else's adult media. The review reads it as part of your own post.
- An adult-flagged post gaining traction. Past 128 likes, a stricter model re-reviews it.
- A sensitive-media setting left on, which blurs your media even for followers. That one is a switch you control, separate from any review.
How long it lasts: The account-level flag lasts 7 days from the last matching post, and a new match inside the 60 day window restarts it. Post-level flags carry no written expiration in the published code.
What does not help: Counting on account size. Adult content rules carry no big-account exemption in the published code. Expecting deletion to clear an active account flag. It only stops future counting.
The exit: stop the trigger above, then let the timer run. The step-by-step exit protocol for this exact cause, matched to your account's measured signals and re-checked daily, comes with monitoring.
What is the engagement bait flag on X?
Posts that ask people to reply, tag, follow, or share to unlock something can be flagged and removed from recommendations to non-followers, while your followers still see them.
What triggers it:
- Asking for a reply with a specific word to reveal something.
- Tag-a-friend, follow-to-see, or share-to-unlock mechanics.
- The same call-to-action template across many posts.
- Engagement trading with other accounts.
How long it lasts: The label application in X's published code passes no expiration value, and the removal function is not public, so treat it as not expiring on its own.
What does not help: Counting on account size or verification. This category carries no high-credibility exemption in the published code. Growing followers to route around it. The restriction only ever applied to non-followers.
The exit: stop the trigger above, then let the timer run. The step-by-step exit protocol for this exact cause, matched to your account's measured signals and re-checked daily, comes with monitoring.
Does a policy strike cause a shadowban on X?
A rules violation can place a reach-limiting label or a suspension on the account, with 30 day timers on some labels and none on others.
What triggers it:
- Content flagged as hateful conduct, violent speech, harassment, or a civic-integrity violation. These hide a post even from your followers, according to X's published code.
- Behavior classified as a coordinated or inauthentic campaign.
- Repeated posts flagged as spam, low quality, or AI-generated filler.
- Severe violations that route straight to permanent suspension.
How long it lasts: Several labels carry an exact 30 day timer in the published code, and re-triggering the same violation typically restarts it. Coordinated-campaign labels and suspensions carry no timer at all in the published code.
What does not help: Deleting the flagged post to clear an account-level penalty. The published code shows no such path. Waiting out a suspension that has no timer. Those resolve through an appeal or not at all.
The exit: stop the trigger above, then let the timer run. The step-by-step exit protocol for this exact cause, matched to your account's measured signals and re-checked daily, comes with monitoring.
What is an account security hold on X?
An account flagged as compromised, locked read-only, or marked as possible impersonation has its posts blocked from reaching non-followers until the underlying status changes.
What triggers it:
- Signs of a compromised account, like unusual sessions or a leaked password.
- A reported post left up after a notice, which holds the account read-only.
- An automated impersonation flag on the profile.
How long it lasts: No timer exists. The published code reads the account's security status live on every request, so the effect ends the moment that status changes, and only the matching fix changes it.
What does not help: Waiting a set number of days. There is no timer on this hold in the published code. Deleting unrelated posts. Only the post named in a read-only notice is documented as relevant.
The exit: stop the trigger above, then let the timer run. The step-by-step exit protocol for this exact cause, matched to your account's measured signals and re-checked daily, comes with monitoring.
Is a small or new X account shadowbanned by default?
An account under roughly 400 active followers has no community cluster yet in X's published code, so it gets little reach beyond its followers. That is a threshold, not a penalty.
What triggers it:
- Fewer than roughly 400 active followers, the written threshold for community clustering.
- A young account still building engagement history.
- Most engagement coming from the same small circle.
- Starting a fresh account, which resets accumulated credibility to zero.
How long it lasts: Nothing to expire. Reach grows as the thresholds are crossed. The clustering job runs weekly and the credibility score recalculates daily, so structural changes typically show within a week.
What does not help: Restarting the account. The new-account penalty is a myth in the published code, and the real cost of restarting is losing every accumulated signal. Buying followers. The threshold counts active followers, and purchased ones are not that.
The exit: stop the trigger above, then let the timer run. The step-by-step exit protocol for this exact cause, matched to your account's measured signals and re-checked daily, comes with monitoring.
What is predicted audience rejection on X?
For each viewer, X predicts the chance they would mute, block, report, or dismiss your post, and strong negative predictions push the post to the bottom for that viewer, with no memory between requests.
What triggers it:
- Content resembling what your reachable audience previously muted, blocked, or reported.
- A mismatch between what you post and who currently receives it.
- Divisive posts served to audiences that historically dismiss them.
How long it lasts: Nothing is stored on your account on this path. The prediction is recalculated from scratch on every request, so there is no penalty to wait out. It fades as soon as the predictions do.
What does not help: Deleting the poorly received posts. Scores are recalculated per request from predictions, never from a stored history of one post. Fighting a mass-block brigade with counter-engagement. A comment in the ranking code states organized reports do not move this signal, since only served impressions count.
The exit: stop the trigger above, then let the timer run. The step-by-step exit protocol for this exact cause, matched to your account's measured signals and re-checked daily, comes with monitoring.
What X design limits look like a shadowban?
Several reach ceilings in X's published code apply to everyone by design, and they explain most scares where every check comes back clean.
What triggers it:
- Replies and retweets to people who do not follow you. They are never recommended to strangers, by design.
- A Protected account setting, invisible to non-followers by definition.
- An edited tweet. Each edit permanently retires the earlier version, with no exemption for the author.
- Very high posting volume. Past roughly 50 posts or 30 replies in 48 hours, older ones drop out of one internal feed pool.
- A country-specific legal takedown.
How long it lasts: These carry no timers because they are not sanctions. Each one lifts the moment its condition changes: the reader follows you, the setting flips, the volume window rolls over.
What does not help: Reading normal design limits as a shadowban. If every check passes, low reply reach to strangers is the system working as documented. Deleting posts to escape the volume window. The window rolls on time, not on deletions.
The exit: stop the trigger above, then let the timer run. The step-by-step exit protocol for this exact cause, matched to your account's measured signals and re-checked daily, comes with monitoring.
Common mistakes
- Deleting your posts or your whole account. You lose your history, and a brand-new account starts with less trust than an established one, not more. It skips nothing.
- Creating several backup accounts "just in case". It does not touch the restriction on your main account, and a batch of new accounts behaving like an established one is itself one of the patterns that gets flagged.
- Paying for a "shadowban removal" service. There is nothing to buy. X has no such lever, and nobody selling one has special access to your account.
- Appealing through X support. Suspensions have a strike you can appeal. A shadowban is not a strike, so there is nothing on file to contest.
- Tweeting "am I shadowbanned" and asking your followers to check. It proves nothing. People who already follow you see a different view than a stranger does, which is exactly why every manual check on this site starts from a second account that does not follow you.
How long it takes
Duration follows behavior: a restriction whose cause keeps going does not run out a clock on its own. Once the trigger stops, reported windows run from a few days to a few weeks depending on which restriction it is. X's published code writes exact timers on several of the underlying flags: 7 days for report ratio, pinned link, and adult content account flags, 14 or 30 days for a flagged reply, 30 days for several policy-violation labels, and no timer at all for duplicate content flags. See how long each one tends to last for the breakdown.
How you'll know it's over
X will not tell you. Either re-run the manual checks above by hand every day or two, or let the agent do it for you: it keeps checking your visibility throughout the day and emails you when it starts improving, and when the restriction fully lifts. Nothing otherwise, silence means nothing changed yet. See pricing.
Frequently asked
Which restriction do I actually have?
You cannot tell from the outside without testing, since the 4 restrictions have different symptoms and none of them show up in your own logged-in session. Test all 4 by hand from a second account, or run the free automatic check, before picking a fix.
Does deleting my account and starting a new one fix it faster?
No. You lose your posting history, and a brand-new account starts with less trust than an established one, not more, so it does not skip the line. It also does nothing about the behavior that likely triggered the restriction in the first place.
Can I pay someone to remove a shadowban?
No. X has no removal service and no paid lever that lifts a restriction faster. Anyone selling a guaranteed fix is selling you nothing.
Will X notify me when it's fixed?
No, in either direction. X never announces a restriction landing or lifting. The only way to know is to re-test, by hand or with a tool that re-checks for you.
I stopped the behavior that caused it. Why is it still there?
Restrictions clear on their own timeline once the trigger stops, typically days to a few weeks depending on which one it is, and X gives no countdown. See how long each restriction tends to take, then keep re-testing rather than assuming a fix landed instantly.
What causes a shadowban on X?
In X's published recommendation code, reduced visibility comes from specific flags with specific triggers: pasted or bursty replies, near-duplicate posts, links to flagged domains, mass mentions of strangers, automated action patterns, a high report-to-like ratio from strangers, adult content flags, engagement bait, policy strikes, and account security holds. Three more patterns look like a shadowban but are built-in limits: small or new accounts below the clustering threshold, predicted audience rejection, and design ceilings that apply to everyone.
Does deleting posts fix a shadowban?
Rarely, and X's published code shows why: most flags either expire on their own timer or persist regardless of the post. The exceptions are narrow. A duplicate-content flag sits on the post itself, so deleting the post removes what carries it. Unpinning a tweet with a flagged link stops the account label from being applied again. Deleting flagged adult posts plausibly stops them counting toward the account-level window, though the code does not spell that one out. Outside those narrow cases, deleting your history costs you the history and changes nothing.
Test it yourself
Run the 4 checks on any handle
More on each restriction
Last reviewed