First, rule out the boring explanations
View counts move for reasons that have nothing to do with punishment. The algorithm shifts mix, a trending topic you posted about cools down, or your last few posts simply landed at quieter hours. Before assuming a shadowban, compare at least two weeks of data, not one bad day.
X's own analytics (Premium) show impressions per post. If every post dropped by a similar factor at the same time, something account-wide happened. If only one or two posts flopped, that is normal variance.
The classic restriction signals
Three patterns show up again and again in restriction reports. First: your posts stop appearing when people search your exact handle or keywords from your posts. Second: your replies become invisible to anyone who does not follow you — a friend who does not follow you cannot see your reply under someone else's post. Third: your follower count keeps growing but views and profile visits fall off a cliff.
None of these signals is individually conclusive. X does not notify anyone, and some dips are caused by plain spam filtering that lifts itself in a few days.
The two-minute DIY test
Log out (or use a private window) and search your exact @handle in the X search box. If your account does not appear in the people results, that is a strong search-suggestion restriction signal. Then, from the logged-out window, open one of your recent posts that has replies and look for your own reply. If you cannot see it while logged out but can see it logged in, reply visibility is the suspect.
The catch: manual tests are noisy. Search results differ by region, the logged-out view is not the same as what followers see, and cached pages lie. That is exactly why automated checks measure the same signals in a repeatable way.
What to do in the first 72 hours
Delete any post that could have tripped spam rules — mass hashtags, duplicated links, aggressive reposting. Stop automation tools temporarily. Then go quiet for a couple of days: fewer posts, genuine replies only. Most soft spam-filter states reset within days once the triggering behavior stops.
If the restriction persists past a week, run a proper scan to document which checks fail, then keep re-checking every few days. A documented before/after is also the only way to know whether anything you changed actually worked.