How long do X shadowbans last?

Search the settings screens and you will not find 'shadowban expires in 7 days.' X's public descriptions of restricted reach talk about what the limit does: drop a post from search, pull it out of recommendations, downrank it in replies. They do not attach a standard duration to each action. When a temporary-label notice shows up in the app, people consistently report that it says reach may be affected and does not say when that stops.

That absence is why blogs and checkers publish numbers anyway. Someone's suggestion list started working again after a weekend, and the weekend became a rule. Someone else's replies stayed hidden for much longer, and that length became a different rule. Both stories can be true as personal outcomes. Neither is a policy you can hold X to. If a page gives you a precise schedule by ban type and presents it as fact, it is filling a blank X left empty.

This site used to repeat ranges like that because they are what people expect to read. They are not something we can verify, and they are not something you should plan a launch around. Use this page to decide what to retest and what behavior keeps a limit in place. Use your own notes for the clock.

A suggestion miss, a Latest-tab search miss, a ghosted reply, a deboosted reply, and a sensitive image are not one switch. They can start on the same bad afternoon and end on different days. Clearing autocomplete while from:handle is still incomplete is a partial change, not a contradiction. The names are in what is a shadowban.

• Autocomplete and People suggestions depend on a ranking of handles. They can lag a rename, ignore a quiet account, or drop you after a spam score, and they can return before post search looks normal.
• Post search (from:handle on Latest) tracks whether posts are in the index and eligible to be shown. A single labeled post can be excluded while the rest come back.
• Reply ghosting and reply deboost are about thread placement. They can persist after your original posts are searchable again, because you are still replying in the pattern that pushed you down.
• A sensitive media flag is attached to a post or to your media setting. Waiting does not peel a label off a post that is still live. Changing the setting affects what you publish next. Deleting or successfully appealing affects that post.

Account-level labels, when Under the Hood shows them, are summarized for a prior month. That report is a look backward. It will not tick down in real time as a limit expires. A label that no longer appears in next month's report is a slow confirmation, not a live all-clear for today's reply.

Repeating the trigger is the reliable way to stay limited. If duplicate replies and follow churn are what the spam systems scored, another day of the same activity is a new input, not a neutral wait. People do this by accident when they 'test' twenty times or post more aggressively to prove the account still works.

• Bursts of follows, unfollows, likes, or replies after you have already seen a visibility drop.
• Reposting the same link that was in the posts search will not show.
• Leaving a sensitive setting on, then uploading more images and wondering why the warnings continue.
• New accounts created to route around the limit. They start with less history, and the same behavior on the new account is a fresh score, not a reset of the old one.
• Bought engagement. It is a spam signal of its own, and it does not expire the underlying label.

Support tickets do not function as a timer either. A ticket can be appropriate for a specific post you believe was mislabeled. Sending the same ticket every morning does not shorten an automated limit that X has not offered an appeal clock for. Keep one report if you need it. Put your effort into the retest.

Write down the date and the exact failed check. Not 'reach feels bad.' The check: logged-out from:handle Latest missed three original posts; autocomplete missed a unique prefix; a logged-out thread hid a reply even after Show more; or a post showed a sensitive warning. The procedure is the manual test.

Change the matching behavior, then repeat that same check later. A same-day repeat tells you whether you were looking at indexing lag. A later repeat tells you whether the symptom is still there. When the check passes, that is the duration for that symptom on this account. It is not a promise about the next time.

The checker is useful on those retest days because it runs several probes in one pass and gives you a link you can keep. Do not retest every few minutes. Search results and suggestion lists move around, and hammering them does not lift a label. The homepage asks for a short wait between runs so the button is not a refresh loop. Give the account a real gap, on the order of a day or more after you change behavior, unless you are only checking whether a brand-new post has been indexed.

If you are eligible for Under the Hood, download or view the report once when you first notice the problem and again in a later month. You are comparing labels X named, not watching a live countdown. If the report says no labels and your logged-out search is still empty, trust the search for the live symptom and treat the report as 'X did not list a label in that monthly aggregate,' which is not the same sentence as 'nothing is wrong.'

Plan as if you do not know the end date, because you do not. Do not schedule a product launch, a campaign, or an argument on the assumption that suggestions return on a particular morning. Keep posting at a human pace if the account is otherwise healthy, so you still have normal posts to test with. Avoid the burst behaviors above.

Use a second, established account only if you genuinely have one and you are not using it to evade a suspension. A suspension is a different state: X tells you, and creating accounts to get around it can extend the mess. A visibility limit on a working account is not a license to spin up replacements. The recovery guide is the sequence to follow while the clock is unknown.

When someone quotes a duration at you, ask which test they repeated and whether they stopped the triggering behavior. If they cannot answer, they have a story, not a measurement. Your notes from the logged-out checks are the measurement.

It will not give you a number of days for a search ban, a ghost ban, or a deboost. Publishing one would be a fabricated statistic. It will not tell you that waiting alone fixes a sensitive label that is still attached to a live post. It will not tell you that a checker's cached 'fail' from last week is proof the limit is still active today. Run the test again.

If you want a single place to recheck the public signals after a gap, use the shadowban checker and keep the /check/yourhandle link for yourself. Those result pages are noindex on purpose. They are a bookmark, not a search listing, and they are still a heuristic rather than X's clock.