← All features
📊

Sentiment Insights

Understand how your audience feels about every video, automatically.

Go to Insights

Sentiment Insights automatically scores every synced comment on a scale from negative to positive, tags it by type, and surfaces trends over time — so you can understand your audience's mood without reading thousands of comments manually.

How sentiment scoring works

Each comment is processed through an AI embedding model that converts the text into a vector representation. That vector is then compared against pre-trained anchor vectors for positive, neutral, and negative sentiment to produce a score between -1 and +1.

Individual comments can occasionally be scored incorrectly — sarcasm is notoriously hard for any AI system. But aggregate trends across a video or time period are highly reliable. Use the per-video score as a signal, not a verdict on any single comment.

  • Scores run from -1 (very negative) to +1 (very positive)
  • The per-video score is the average across all scored comments on that video
  • Scores are computed when comments are synced — no extra action required
  • Available on all paid plans (Plus, Pro, Premium)

AI comment tags

In addition to sentiment, each comment is assigned one or more tags describing its content type. Tags include categories like:

  • Question — the commenter is asking something
  • Praise — positive feedback or compliment
  • Complaint — expressing dissatisfaction
  • Suggestion — requesting a feature or change
  • Spam — promotional or off-topic content

Tags let you filter your comment inbox by intent. For example, filter to Questions to quickly find all unanswered questions across a video, or filter to Complaints to triage issues before they snowball.

Sentiment alerts

You can configure an alert that emails you when the negative sentiment on a video crosses a threshold you set — for example, when more than 40% of comments are negative.

  • Alerts are configured per-channel from the Insights page
  • They fire when sentiment is getting worse — not when it's already bad and improving
  • There is a 6-hour cooldown between alerts on the same video to prevent email floods
  • Useful for catching controversial uploads or PR situations early

The Insights dashboard

The Insights dashboard shows sentiment trends over a selected time range across all your videos. You can drill into any video to see a comment-by-comment breakdown, filter by tag, and export data. The time range picker lets you compare periods — useful for tracking audience mood before and after a format change.

If you're on the Free plan, you'll see a preview of the Insights page with sample data. Upgrade to any paid plan to unlock scoring for your real comments.

Frequently asked questions

What is comment sentiment analysis?+
Sentiment analysis uses AI to score each comment on a scale from negative to positive. CommentShark automatically classifies every synced comment so you can see at a glance how your audience feels about each video, without reading thousands of comments manually.
Which plans include sentiment insights?+
Sentiment analysis, AI comment tagging, and the Video Insights dashboard are available on all paid plans (Plus, Pro, and Premium). Free-tier users can upgrade from the Insights page to unlock these features.
What are AI comment tags?+
Comment tags are automatic labels applied to each comment based on its content. Tags include categories like questions, praise, complaints, suggestions, and spam. They let you quickly filter and focus on the comments that matter most.
Can I get alerts when negative sentiment spikes?+
Yes. You can configure sentiment anomaly alerts from the Insights page. Set a threshold (e.g. 40% negative) and CommentShark will email you when a video crosses it. Alerts have a 6-hour cooldown and only fire when sentiment is getting worse, not when it improves.
How accurate is the sentiment scoring?+
Sentiment scores are computed using AI embeddings and pre-trained anchor vectors. The system works well for general YouTube comment language. Individual comments can occasionally be misjudged (sarcasm is hard for any AI), but aggregate trends across a video are highly reliable.