Twitter shadowban test: check search, replies and labels on X

On X, formerly Twitter, a “shadowban” usually means reduced visibility: an account stops appearing in search suggestions, its posts drop out of search results, or its replies sink to the bottom of a thread. You can test each of these by hand from a second logged-in account that does not follow the account you are checking. Treat a missing result as a symptom, not a verdict. Only X's own record tells you that it limited an account and why: a notice or label on a post, the Under the Hood report it is piloting, or, in the EU, a statement of reasons. A missing record does not clear an account either.

The same checks help when you are reading other people. A reply you cannot find under a thread, or a post that does not come up in search, may still exist. Before deciding that someone went quiet or deleted something, it is worth knowing how X decides what each viewer sees.

What X says it does, and what people call a shadowban

In July 2018, Twitter published Setting the record straight on shadow banning. It defined a shadowban narrowly, as “deliberately making someone's content undiscoverable to everyone except the person who posted it, unbeknownst to the original poster”, and said it did not do that. In the same post it said: “We do rank tweets and search results.” It also described a bug that had kept hundreds of thousands of accounts out of search auto-suggestions while their posts still appeared in search results.

X's current range of enforcement options is more explicit. Under “Limiting post visibility”, X lists excluding a post from search results, trends and recommended notifications; removing it from the For you and Following timelines; restricting its discoverability to the author's profile; and downranking it in replies. In April 2023 X called this practice visibility filtering, acknowledged that “we have not historically been transparent when we’ve taken this action”, and began adding labels to some posts whose visibility it limited.

So when people say “shadowban”, they usually mean one of X's visibility limits, or ordinary ranking that looks the same from the outside. The useful questions are narrower: which surface is affected, for which viewer, and whether X has recorded a reason.

The four things a shadowban test checks

Shadowban checker sites commonly use four labels. They are the checkers' vocabulary, not X's, but each points at a different surface:

• Search suggestion ban. The account does not appear in the drop-down of suggested accounts while someone types its name or handle into the search box. It is also the restriction in what has been described as the first court ruling on shadowbanning under the EU's Digital Services Act. According to the DSA Observatory's summary, Danny Mekić's account appeared to drop out of search suggestions in autumn 2023. In January 2024 X's lawyers confirmed that his posts had triggered a visibility restriction, since reversed, and on 5 July 2024 an Amsterdam court ruled that X had owed him an explanation.
• Search ban. The account's posts do not appear in search results, even in a search limited to that account.
• Reply deboosting. Replies appear only near the end of a thread or behind a button such as “Show probable spam”.
• Ghost ban, or thread ban. A reply does not appear under the parent post for other people at all.

None of these checks measures reach in the For you timeline. X's published For You algorithm handles that separately: its label catalog describes many labels whose effect is that posts are “hidden from recommendations to non-followers”, while followers can still see them. A post can be fully searchable and still not be recommended to strangers.

How to run a manual shadowban test

You need a second view of X. Use another account, yours or a friend's, that is logged in, has an adult birth date on its profile, does not follow the account you are testing, and has not muted or blocked it. X withholds some sensitive media from viewers who are under 18 or have no birth date on their profile, so a test account without a birth date can miss posts for that reason alone. A logged-out browser is not a substitute: X says advanced search is available when you're logged in. Note the test account's safe-search settings, because X's search filters can exclude potentially sensitive content and muted or blocked accounts. Write down the date and time of each check.

Search suggestions

From the second account, type the handle and then the display name into the search box, slowly, and see whether the account is suggested. Then run the search and open the People or Accounts results. X's search help says account search prefers profiles with a complete name, username and bio, and that profile changes can take a few days to appear. A new account, or one that has just changed its name, can be missing for those reasons.

Search results

Search for the account's own posts on the Latest tab, which lists results by time rather than by X's ranking of top posts:

• from:handle lists recent posts by that account.
• from:handle since:2026-11-01 until:2026-11-08 narrows the search to a period around a post you know exists. X does not include the until: date itself, and the date boundary may not match your time zone, so leave a day of margin on each side.
• "a distinctive phrase" from:handle checks one specific post.

The X advanced search builder writes these queries for you and gives you a link to the results on X.

If a date-bounded from: search on Latest, from a logged-in viewer that does not follow the account and has safe search off, keeps returning none of the account's recent public posts over several days, that matches the account-level search filtering described in X's search rules. X says it applies that filtering automatically; it does not say that it tells the account. If only one post is missing, a post-level limit or one of the ordinary causes below is more likely. Either way, the reason cannot be confirmed from outside, and the lack of a label does not show that nothing was limited: Mekić was not told when his account dropped out of search suggestions.

Replies

Open the parent post from the second account and look for the reply in the thread. X's help on replies explains why a reply may be low down: when ranking replies, X considers whether the original author replied, whether the viewer follows the person replying, and whether that person subscribes to X Premium or another paid plan. Replies that X identifies as potentially lower quality or spam sit behind “Show probable spam”. Some threads have a further layer, reported in 2024 as “Show additional replies, including those that may contain offensive content”.

Also check the hidden-replies icon on the parent post. Authors can hide replies to their own posts; anyone can still view them there, and the person whose reply was hidden is not notified. To find replies by search, use from:handle filter:replies, or from:handle to:parenthandle for replies to one account.

If the reply is not under the parent post, open it from the author's Replies tab or by its own link, still from the second account. If it opens there but not in the thread, it exists and is being ranked, filtered or hidden in that conversation. If it does not open at all, it may have been deleted, or the account may be protected or withheld in your country.

Check what X itself records

A search test can only tell you what one viewer saw. These are the places where X itself says that it limited something:

• Notices and labels on posts. Since April 2023, X has added labels to some posts whose visibility it limited. Authors can appeal from the label, according to its enforcement options. Account-level actions, such as read-only mode or a request to verify a phone number or email address, show up when the owner tries to use the account.
• Under the Hood. In August 2026 X began piloting a report showing “aggregate stats on labels that can affect post visibility in the X algorithm, for your posts and account last month”. Each label comes with what it means, its effect on visibility and, where one exists, how to appeal. The published code sets two requirements: an account at least one year old with 10 or more posts in the prior month. Accounts outside the pilot group see a message that the report is not available to them yet. In September 2026 the report was extended to show limits applied to comply with local laws, including posts withheld in a country. It covers only your own account, for the previous month. Many of its effects concern recommendations to non-followers rather than search, but its label catalog also includes post labels that restrict a post's discoverability to the author's profile.
• A statement of reasons in the EU. Article 17 of the Digital Services Act requires a hosting service to give affected users “a clear and specific statement of reasons” for restrictions on the visibility of their content, “including … demoting content”, when the restriction is based on illegality or the service's terms and the provider has the user's contact details. This duty does not apply to deceptive high-volume commercial content. The Mekić ruling applied this duty to search delisting.

Innocent explanations to rule out

Many missing results have ordinary causes. Check these before you conclude that X limited an account:

• Protected posts. Protected posts are searchable only by the author and approved followers. Replies from a protected account to someone who does not follow it are not visible to that person.
• Sensitive content and safe search. X's safe search filters potentially sensitive content from search. If an account marks its media as sensitive, or X has applied that setting, its posts can disappear for viewers who keep safe search on.
• A new or incomplete account. X asks users to confirm an email address and complete the profile. In 2018 Twitter listed a confirmed email, account age and a profile image among the signals it used to decide whose posts to rank lower in search results and conversations.
• The viewer's own filters. Muted and blocked accounts are excluded from search results, and muted words can hide posts or place them behind a notice. The notifications Quality filter removes some lower-quality content from notifications only, not from search or threads.
• Ordinary reply ranking. Replies from paid subscribers, from accounts the viewer follows and replies the author answered are ranked higher, according to X's help on conversations. A reply further down is not necessarily penalized.
• Hidden replies and reply settings. The author may have hidden the reply, or limited who can reply. A reply that was never allowed cannot appear.
• Limits and loading failures. X sets daily limits on posting and says it shows an error message telling you which limit you hit. Viewers can be limited too: in July 2023 Twitter, now X, temporarily capped how many posts accounts could read in a day, the BBC reported. A timeline that stops loading is not a decision about the author.
• Deleted, deactivated or withheld. A notice may replace a post that was deleted, comes from a deactivated account, comes from an account that blocked you, or is withheld in your country.

Some causes are not innocent but are documented. X's search rules say it may remove accounts from search for repeated duplicate posts or links, hashtag abuse, automated posts or replies, similar messages across several accounts, or aggressive following and unfollowing. They add that followers still receive the account's updates, and that replies and mentions are still delivered.

What third-party shadowban checkers can and cannot tell you

Most shadowban checkers act as one more viewer: they request search suggestions, run searches or open threads, then report what they did not find. Browser extensions may instead read flags from the data X sends to your browser. Either way there are three limits:

• They see what their own access returned at one moment. They cannot see another viewer's ranking, filters or follows.
• They depend on how X serves search and threads to them. When X changes those interfaces, checks can fail without saying so.
• They cannot see X's internal labels or the reasons for them. At most they read a flag that X sends to the browser, or a label shown on a post. A “ban” result means the tool did not find something; it does not say why.

Some tools say this themselves. The open-source Shadowban Scanner extension explains that it estimates account-level shadowbans from account sensitivity flags, and that this method “can produce false positives or false negatives”. Use a checker's result as a reason to run the manual test, not as evidence. Do not give any checker your X password.

When the missing post belongs to someone else

When you are reading another person, the same distinctions keep you from mistaking low visibility for silence or deletion:

1. Open the source directly. Go to the profile and its Posts and Replies tabs. If you have a link to the post, read the post's time from its ID and open the original, then use that date in step 2.
2. Search by account and date. A from: search on the Latest tab with a date range, built with the advanced search builder, can find posts that general searches and timelines skip.
3. Look through the whole thread. Check “Show probable spam” and the hidden-replies icon before concluding that a reply was deleted.
4. Consider why you might not see it. The account may protect its posts, the content may be withheld in your country, or you may have muted the person or a word in the post.

The same caution applies to someone who says they are shadowbanned. A screenshot from a checker shows that tool's result at one time. A label on a post, an Under the Hood report or a statement of reasons is stronger evidence. A drop in likes or views can also come from ordinary ranking.

If you follow people through an X List, remember that it is not a full archive of their conversations either. The guide to Crypto Twitter explains why a List can miss a member's replies to people outside it.

If X has limited your account

Read the label or notice first: it names the policy. Appeal from the label, or, for a post label, follow “How to appeal” where Under the Hood offers one. In the EU, a visibility restriction based on X's terms or on illegality requires a statement of reasons, except for deceptive high-volume commercial content; a Dutch court found that X had breached that duty in the Mekić case.

If there is no label but your posts stay missing from search, the cause cannot be confirmed from outside. Stop any behaviour on X's search-rules list, such as near-duplicate posts, automated replies or posting the same message from several accounts, check again after a few days and, in the EU, ask X for a statement of reasons. X does not publish how long a restriction lasts. It documents reply prioritization for paid subscribers, but it does not say that a subscription removes a visibility label.