Engagement is dropping: four things to check before blaming the algorithm

When a channel dips, "the algorithm changed" is the most used explanation and the least checkable. The four causes below are readable from your own channel's public data, and they explain most cases.

First step: confirm there is a dip

Run channel report — 5 credits, 60 most recent videos — for your own channel. Compare the median views of the 20 most recent videos against the 20 before.

Use the median, not the mean. If you had a viral video three months ago, the mean then was unusually high and you are comparing against a piece of luck, not against your normal.

If the median has not fallen, your channel is not dipping — you are remembering your best video.

Cause 1 — frequency fell

Check the posting dates across the 45 videos. Posting three times a week for three months and then dropping to once is the most common cause, and the one people least often admit.

See posting frequency.

Cause 2 — the audience got diluted

If you recently had an off-topic trend video that brought many new followers, those people do not care about your usual content. The engagement rate falls because the denominator grew and the numerator did not.

The sign: follower count stepped up, then later videos' engagement fell. See catching trends without losing your brand.

Cause 3 — the content is repeating itself

Check the topics of the 20 most recent videos. If they are variations on one thing, the audience has already heard it.

The fix: take ideas from comments rather than from memory. Run comments analysis — 12 credits, 300 comments — and pull the five most repeated questions. See content ideas from comments.

Cause 4 — the whole niche is falling

The only cause that is not your fault, and the one to check last because it costs the most credits.

Run channel report for three or four channels in the same niche at 5 credits each. If all of them fell at the same time, it is the niche or the platform, not you. If only you fell, go back to the three causes above.

When the algorithm is a fair suspect

When all four causes above are ruled out: frequency held, the audience was not diluted, the content stayed varied, and niche peers are fine. Then it may be the algorithm, or your videos may have had their reach limited.

Public data does not show whether a video was limited. The indirect sign is views dropping to an unusually low and unusually consistent level.

Frequently asked questions

How long until I know a fix worked? At least ten videos, because individual videos vary a lot. Compare the median of ten new videos against ten before the fix.

Does deleting weak videos help? There is no public evidence for that, and you lose the data to compare against.

Are 45 videos enough to see a trend? For a channel posting consistently, yes. For longer history, run it periodically and keep the results.