The Litter Tray

Use case

Two videos posted days apart can get very different comment sections.

Video sentiment

See how comments under each video lean, from supportive to hostile.

Sentiment is paused while we ask YouTube for permission to offer it. Nothing already scored is deleted.

Each row sums up the scored comments under one video. A video appears once three of its comments have scores, and the page shows up to 12 videos, warmest first. Hostile, neutral or supportive describes how those viewers stood towards you and your community. It doesn't say whether they liked the video, or explain how it performed.

An illustration of the Sentiment page, with how each video landed under the last thirty days.

Outcomes

What this gives you

Compare videos with enough comments

Videos with fewer than three scored comments stay out of the table.

See how many comments are behind the word

Each video's word (hostile, neutral or supportive) sits beside the number of comments it's based on.

It doesn't measure performance

It tells you nothing about views or watch time, why the comments went the way they did, or whether people liked the video itself.

The path

From the situation to the answer

  1. Let the AI read the comments

    When the AI reads a comment, it can score it from -1 (hostile) through 0 (neutral) to +1 (supportive), for how the viewer stands towards you and your community.

  2. Reach three scored comments

    A video stays out until at least three comments have a score.

  3. Open Sentiment

    How each video landed lists up to 12 videos with enough scored comments, warmest first. It counts every scored comment under each video, however old.

  4. Open the video if you want to

    Use the word as context when you read the comments. It doesn't measure how the video performed.