Was your reach cut, or was it the content?

A view count is the product of two things: how many people were shown the video, and how many of those kept watching. A drop in views is consistent with either falling, and the two call for opposite responses.

Engagement rate separates them, because it measures how the people who did see it responded. It is a rate rather than a total, so it is not dragged down by the fall in views itself.

Recent window: videos posted in the last 14 days that are at least 72 hours old. A video still inside its testing window has not finished collecting views and will distort the comparison.

Baseline window: videos posted between roughly 15 and 90 days ago. You want at least ten of them.

For each video, add likes, comments, shares and saves, and divide by views. Take the middle value in each window.

Divide the recent middle value by the baseline one. That single ratio is the answer.

At or above 1.15 — the people who saw it engaged more than usual. Strong evidence that distribution was cut and the content is fine.

Between 0.95 and 1.15 — engagement held steady while views fell. Consistent with reduced distribution.

Between 0.70 and 0.95 — ambiguous. Something moved in both, and this range does not settle it.

Below 0.70 — engagement collapsed alongside views. The recommendation system stopped promoting content that people were not responding to. This is the system working, and it is the most common result.

One caveat that matters: only run this comparison when views have genuinely fallen. Engagement rate wobbles between videos for a dozen ordinary reasons, and reading it in isolation will find a crisis in a healthy account.

"Your reach was not cut, the content did not land" is unwelcome and it is the answer most creators need. It is the only one of the two you can do something about this week.

It also protects you from the standard failure mode: reading an ordinary underperforming run as enforcement, then reacting to an imaginary penalty by changing everything at once, which destroys whatever consistency the system had to work with.