Your reach dropped. Here is what actually causes it

It goes like this. You had a good week. One post did numbers you had not seen before, replies came in, the follower count moved for once. Then the next four posts land in a room with the lights off. Same effort, same topic, a tenth of the views.

So you do what everyone does. You search your own handle logged out, you find a forum thread from 2023, and by lunchtime you are fairly sure you have been shadowbanned. That conclusion is available, emotionally satisfying, and wrong most of the time. Worse, it sends you to fix a problem you do not have.

The word is doing too much work

Visibility filtering is a real thing. Platforms have always had ways to show a post to fewer people without telling anyone, and X is no exception: replies can be pushed below a fold, accounts can be held out of search, posts can be excluded from recommendations. None of that is folklore.

What is folklore is the shape people imagine. The forum version is a switch: someone flipped it, you are invisible to everyone, and you will stay invisible until you appeal. Almost nothing works like that. Distribution is a continuous decision made per post, and most of the time nothing was flipped at all. The system simply stopped betting on you as hard, which is a different sentence with a different fix.

Start with the boring explanation, because it is usually the right one

Before anything else, check whether your reach actually dropped or whether your reference point moved. These are not the same thing and people confuse them constantly.

Take your last twenty posts. Say nineteen of them did around two thousand views and one did ninety thousand. Your average is six thousand four hundred. Your median is two thousand. The average describes a post you may never write again. The median describes your Tuesday.

Now look at what happened after your breakout. Your next posts did two thousand, which is exactly what they always did. Nothing dropped. You just spent a week comparing your normal against your best day, and normal lost. This is the single most common cause of the feeling that reach collapsed, and it is not a platform problem at all.

Always use the median. An average with one outlier in it is not a description of your account, it is a description of your luckiest afternoon.

Four ordinary reasons a good account goes quiet

You changed what you post about. The feed learns which audience to try you on. Change subject and it has to relearn, which means a stretch of smaller, wider tests while it works out who to show you to now. This looks identical to being punished and is the opposite: it is the system being careful.

Your timing moved. A post that goes out ninety minutes later than usual can land after your audience has closed the app. The first hour decides most of what follows, so a small shift at the start compounds into a big difference at the end. We wrote the whole calculation in finding your own posting window.

Early velocity was low, so distribution stopped. Reach is not granted up front, it is extended in steps. If the first sample of people does not engage, there is no second sample. One post that opens badly can look like a ban when it is really a post that failed its first audition.

You stopped replying. Replies are where small accounts borrow distribution, and the week you go quiet in other people's threads is usually the week your own numbers sag. It is easy to miss because the cause and the symptom are in different places. The mechanics are in the reply strategy.

What genuinely does get your distribution cut

Now the real list. These are patterns, not accusations, and they have one thing in common: each looks like automation from the outside, whether or not you automated anything.

• External links in the main post. The platform wants people to stay. A post whose body sends them elsewhere gets measurably less distribution than the same post without it. Put the link in your first reply instead. More on this in what the algorithm rewards.
• Replies that arrive in a pattern. Same opening words, same length, many accounts, minutes apart. A human replying fast is normal. A human replying fast with an identical shape every time is the exact signature a filter is looking for.
• Copy pasted text across accounts or threads. The same paragraph appearing in many places is one of the cheapest things in the world to detect.
• Engagement bait phrasing. "Comment X and I will send you the guide." It works until it is classified, and then it drags the account that used it.
• A volume spike out of nowhere. Going from four posts a day to forty is a behavioural change, and behavioural changes get looked at. Ramp instead of jumping.
• Follow and unfollow cycling. Old trick, well understood, still costs you.

Notice what is not on that list: posting a lot, having opinions, using a tool to draft. What gets filtered is the pattern, not the effort.

A ten minute diagnostic

Do this in order and stop as soon as one step explains it.

1. Compute your median views over the last twenty posts, and the median for the twenty before that. If the two medians are close, your reach did not drop and you can go back to work.
2. Check whether your recent posts differ. New topic, different format, links in the body, much shorter or much longer. Change is a better explanation than punishment.
3. Compare the first hour. Take a post that did fine and one that did not, and look at where each was at sixty minutes. If the weak one opened slowly, the problem is the open, not the account.
4. Check replies specifically. If your own posts are normal but your replies get nothing, that is reply visibility, which is a different and more targeted thing than an account level problem.
5. Look at a post logged out. Open a private window and check that your recent posts and your profile are visible at all. This is the only step that tests the thing people actually mean by the word.

Four of those five steps usually end the investigation, and none of them require a tool that promises to detect a ban.

Three reactions that make it worse

Deleting the posts that underperformed. It does not reset anything, it removes the only evidence you had about what your audience ignores, and a wall with holes in it is a worse first impression for the visitor described in why your posts get views but no followers.

Posting three times as much to compensate. A volume spike is on the list above. You are answering a suspected filter with the behaviour most likely to trigger one, and you are diluting the median you were trying to raise.

Buying engagement to prime the pump. Bought engagement comes from accounts that never read you. You are teaching the system to show your posts to people who will not react, which is a lesson it takes seriously and does not forget quickly.

Why we do not automate the part that carries the risk

Every pattern on the real list is something a tool can produce at scale without meaning to. Templated replies, identical text across threads, a sudden jump in volume: those are the natural output of automation left running.

Which is why MomentX drafts and never sends. Every reply and post opens X's own composer with the text filled in, and you decide. It also means each draft passes through your own voice profile rather than a shared template, so two users working the same thread do not post the same sentence. That is a distribution decision as much as a writing one, and it is the same reasoning behind the tells that make a post read as AI.

Where this fits

Reach is the output of four things, and this post is about the one people misdiagnose. The other three: what the algorithm rewards, when your audience is actually awake, and how replies borrow distribution. If your medians are flat and none of the real patterns apply to you, the answer is almost always in one of those three, not in an appeal form.