Your Reels were doing fine. Then one week they were not — views down to a fraction of the usual number, and nothing about how you work looks different. The first instinct is to assume you have been penalised. That is occasionally true and usually not, and the difference matters because the fixes have nothing in common.
What follows is an order of operations: the cheapest checks first, the dramatic explanations last, where they belong.
Step zero: confirm the drop is real
Reel performance is heavily skewed. A small number of videos collect most of the views, and the rest sit near a much lower baseline. So a run of ordinary videos following one outlier looks exactly like a collapse, even though it is a return to normal.
Before you diagnose anything, work out your actual baseline. Take your last thirty Reels, drop the single best and single worst, and look at the median rather than the average. One video with ten times your usual reach drags the average up far enough that everything afterwards feels like failure.
Then compare like with like. If your last four videos are a format you rarely post, or half the length you normally publish, or on a topic your audience has not seen from you before, you are not looking at a distribution problem. You are looking at a different product.
The single most useful number in Insights
Instagram splits reach between followers and non-followers. That split is the closest thing you have to a diagnostic, because the two numbers fail for different reasons.
- Non-follower reach collapsed, follower reach normal. Your videos are still being served to the people who chose you, but the recommendation surfaces have stopped picking them up. That points at the content itself or at recommendation eligibility, not at your account being broken.
- Both collapsed. Something account-level or format-level changed. Check posting behaviour, account status, and whether you switched something in the app.
- Follower reach collapsed, non-follower reach normal. Usually an audience problem rather than a platform one — your existing followers are scrolling past, which eventually drags the rest down too.
Most people skip this split and go straight to guessing. It answers half the question in about a minute.
The likely causes, roughly in order
- Normal variance. Covered above. It explains more "collapses" than everything below combined.
- You changed the content without noticing. New topic, new length, new opening style, a different person on camera, a switch from demonstration to talking head. Any of these can move performance substantially.
- Visibly recycled content. Instagram has said publicly that it deprioritises reposted content in recommendations, including videos carrying another platform's watermark. If your workflow is export-from-elsewhere-and-upload, that is the first thing to fix.
- Recommendation eligibility. Instagram publishes guidelines for what it will and will not recommend to people who do not follow you. Content can be fully allowed to stay up and still be ineligible for recommendation. That produces precisely the pattern above: follower reach fine, non-follower reach gone.
- Audio. Music rights are regional and can change. A track that is available to you may not be available in a large share of your audience's countries, which quietly caps distribution.
- Timing and season. Audience behaviour shifts around holidays, school terms, and major events. This is real but small, and it is the explanation people reach for far too quickly.
- An actual restriction. Rare, and knowable rather than guessable. See below.
- Metric changes. Instagram has changed what it counts and displays more than once. When a definition changes, a number can move without anything about your reach changing at all.
An hour-long diagnostic you can run today
Do this in a spreadsheet. It is dull and it works better than any amount of theorising.
- List your last thirty Reels, one per row. Columns: date, length in seconds, topic, format, hook type, views, average watch time, shares, saves, follows, and the follower/non-follower split.
- Mark the row where the drop started. Not the week — the specific video.
- Compare the five videos before that line with the five after, column by column. You are looking for a variable that changed at the same time. Usually one is obvious once it is in a grid.
- Open your account status page in the app settings. It tells you directly whether your account or specific posts have been found ineligible for recommendation. Do not guess at this; it is displayed.
- Calculate shares per thousand views and saves per thousand views for both groups. If those rates held steady and only views fell, the content is still landing with the people who see it and the problem is upstream in distribution. If those rates fell too, the content is the problem.
- Post one video in your most reliable old format, made as carefully as you can. This is your control. If it performs at your old baseline, nothing systemic is wrong.
The control video matters most. It converts an argument into an observation.
About the shadowban question
The word covers several different real mechanisms plus a lot of folklore. The real mechanisms are documented: content can be removed, an account can be restricted, and content can be kept up but excluded from recommendation. The folklore is everything else — posting at the wrong minute, using too many hashtags, editing a caption too soon after publishing, and the rest of it.
Check account status instead of speculating. If it says something, act on it. If it says nothing, take that seriously rather than assuming a hidden penalty you cannot see.
Two things worth stating plainly. Nobody sells a working "shadowban removal" service, and paying someone to lift a restriction that a page in your own settings would show you is a straightforward waste. And buying engagement will not restore distribution for a video the system has decided not to recommend — that is not the lever it pulls.
The explanation people skip
Before blaming distribution, check the openings of the videos in the slump. Hooks decay quietly. You settle into a format, the opening gets slightly slower each time, and three months later you are spending the first two seconds on a greeting you did not used to include.
The tell is in average watch time. If it fell at the same time views did, people are seeing the videos and leaving early, which is a content signal rather than a distribution one. If average watch time held steady and only views fell, fewer people were shown the video in the first place, and the openings are fine.
This single comparison sorts a large share of cases, and it costs nothing.
What to do while you work it out
The most common mistake is panic. People change six variables at once, lose the ability to attribute anything, and then conclude the platform is broken.
- Change one thing at a time and give it several posts before judging.
- Keep posting at your normal rate. Going quiet removes the only data you have.
- Do not mass-delete old posts hoping to reset something. There is no reset, and you lose whatever was still bringing in search and profile traffic.
- Do not repeatedly switch account type or username. It solves nothing and it breaks your existing links.
- Do not rewrite your whole content strategy in week one of a dip. Two weeks of ordinary results is not a signal.
Most dips resolve themselves, which is unsatisfying but true. The ones that do not usually turn out to have a specific, boring cause sitting in the spreadsheet: a format change, a watermark, a topic your audience did not want, or a run of videos with weaker openings than you thought.
Run the split, run the grid, post the control. That is the whole method, and it takes an evening rather than a month of anxiety.
