Skip to main content
Sentiment analysis is the core of OpenBat’s user intelligence. Every user message is scored on a positive-to-negative scale, giving you a real-time view of how your users feel about their chatbot interactions.

​
How scoring works

OpenBat analyzes each user message and assigns a sentiment score:
  • Positive (green badge) — User is satisfied, grateful, or expressing approval
  • Neutral (yellow badge) — User is asking a straightforward question with no emotional charge
  • Negative (red badge) — User is frustrated, confused, disappointed, or expressing disapproval
Scores are numeric values that roll up into averages at the conversation, user, and organization level.

​
Where sentiment appears

Sentiment data surfaces in several places across the platform:
  • Dashboard overview strip — Average sentiment for all conversations in the selected period with a trend indicator
  • Dashboard sentiment trend chart — Time-series visualization of sentiment over days, weeks, or months
  • Dashboard “Most dissatisfied users” — Ranked list of users with the lowest sentiment, with “Investigate →” links
  • Dashboard “Sentiment by segment” — Sentiment broken down by customer plan tier
  • Conversations table — Colored badge and score on each conversation row
  • Conversation detail — Per-message sentiment badge on each user message
  • Users table — Aggregate sentiment score and badge for each user
  • Organizations table — Aggregate sentiment score and badge for each organization
The sentiment trend chart on the dashboard shows how user satisfaction changes over time. Use this to:
  • Spot sudden drops in sentiment (possible incident or bad deployment)
  • Track the impact of prompt improvements
  • Compare sentiment across different time periods
  • Identify seasonal patterns in user satisfaction

​
Sentiment by segment

Filter sentiment data by customer plan (e.g., Free, Pro, Enterprise) to understand:
  • Whether paid users are more or less satisfied than free users
  • Which plan tier has the most frustrated users
  • How sentiment differs across customer segments
If your Enterprise users have worse sentiment than Free users, it may indicate that your chatbot doesn’t handle complex use cases well — a signal to review capability gaps.

​
Acting on sentiment data

  • Investigate dissatisfied users — Click “Investigate →” on the dashboard to see their conversations and understand what went wrong
  • Set up workflow alerts — Create a workflow that triggers when sentiment drops below a threshold
  • Track improvements — After making prompt changes, watch the sentiment trend to measure impact
  • Segment analysis — Compare sentiment across plans to identify which customer groups need attention
Sentiment analysis only runs on user messages (not assistant messages). Enable it in analysis configuration under the User Analysis tab.