Guide

YouTube comment moderation metrics, and how to read them

YouTube Studio gives you a comment count. That single number tells you almost nothing about whether the comment section needs your attention. Here are the metrics that do, and how to interpret them.

The short version

  1. Load a video or channelMetrics are computed from the comments already loaded, so the first step is loading them.
  2. Open StatisticsVolume over time, authors, likes, frequent words, top commenters, questions and flagged counts, all from the same set.
  3. Change the granularityGroup the timeline by hour, day, week or month. Hour is for a launch; month is for a channel over a year.
  4. Act on the outliersA spike, a name appearing far too often, or a flagged share above the usual is the thing worth opening.
Open the tool

The metrics worth having

Volume over time

Comments and replies grouped by hour, day, week or month. The shape matters more than the total: a steady curve is an audience, and a vertical spike is either a video being shared somewhere new or a coordinated arrival. Both are worth knowing about; only one is good news.

Unique authors, and comments per author

A thousand comments from nine hundred people is a conversation. A thousand from forty is something else — a small group talking among themselves, or a handful of accounts posting repeatedly. The ratio is the fastest signal of which one you have.

Likes, total and average

Average likes per comment says whether the audience is reading each other or only posting. A high average with low volume usually means a small, engaged group. It is the closest thing here to a vanity metric, but the outliers are useful: the most-liked comment is often the question everyone wanted answered.

Frequent words and top commenters

Word frequency surfaces the subject the audience actually latched onto, which is rarely the one the video was about. Top commenters name the regulars — and, when a name appears far more than the rest, sometimes something that is not a regular.

Flagged counts by signal

Spam, scam, insult, criticism and channel impersonation, counted separately rather than lumped together. The split matters: a rise in criticism is feedback and a rise in impersonation is an attack on your viewers, and a single “flagged” number hides the difference.

Reading them together

No single number means much alone. The combinations are what tell a story:

  • Volume spike + low unique authors — few people posting a lot. Brigading, or a very active thread. Open the top commenters list to see which.
  • Volume spike + high unique authors — the video reached somewhere new. Worth finding out where.
  • Impersonation flags above zero — act immediately, whatever the count. One scam reply under a popular video reaches thousands of people who trust you.
  • Questions rising, replies flat — the audience is asking and nobody is answering. The cheapest engagement win there is.

Where these numbers come from

Everything above is computed in your browser from the comments already loaded. There is no server aggregating anything and no extra API request per metric, which is why opening the statistics costs nothing and why the numbers describe exactly the set on screen — filter the comments and the metrics follow.

That also sets the honest boundary: these are metrics for the comments loaded. Load a single video and they describe that video. Load a channel and they describe the channel. They are not lifetime channel statistics, and nothing here can see comments YouTube has already held for review.

See them for your channel

Open ClearComments, load a video or channel, and open the Statistics tab. Free, no account, nothing uploaded.

Related guides