Analyze user sentiment through customer satisfaction (inferred CSAT) score and CSAT factors such as empathy, frustration and confusion, transfers and escalations from conversations with assistants to improve the quality of user
interactions.
The Sentiment dashboard page aggregates metrics related to user satisfaction, emotional feedback, empathy levels, and conversation outcomes. These metrics enable you to monitor inferred CSAT, track transfers and escalations, analyze
empathy distribution, and review negative emotion trends. The insights from these metrics support targeted improvements to assistant behavior, response quality, and overall user experience.Figure 1. Sentiment dashboard page in Assistant analytics
The visualizations on the Sentiment page help you with the following.
Monitor user satisfaction and sentiment trends to identify strengths and areas for improvement in assistant interactions.
Track emotional feedback and empathy levels, enabling you to address frustration, confusion, and other negative emotions.
Analyze conversation outcomes and recommended next steps to guide assistant optimization and enhance resolution rates.
Overall Sentiment
This area of the dashboard shows the overall average inferred CSAT score for analyzed conversations in the selected date range. The CSAT score is measured on a scale from 0 to 5, where 0 represents the lowest satisfaction and 5
represents the highest. Use this metric to track changes in sentiment over time and evaluate the impact of assistant updates.Figure 2. Overall Sentiment
Conversations Analyzed
This area of the dashboard shows the total number of conversations analyzed for sentiment in the selected date range. This number indicates the breadth of data supporting CSAT scores.Figure 3. Conversations Analyzed
High Empathy Rate
This area of the dashboard shows the percentage of conversations where high empathy was detected in assistant responses. It's calculated as ((Number of conversations with high empathy)/(Total number of conversations analyzed)) x
100. High empathy rate is an indication of the assistant's ability to respond with empathy to users queries.Figure 4. High Empathy Rate
Conversations with Negative Emotions
This area of the dashboard shows the percentage of conversations where negative emotional feedback in terms of confusion or frustration was detected. It's calculated as ((Number of conversations with Frustration or
Confusion)/(Total number of conversations analyzed)) x 100. This metric highlights the prevalence of negative experiences in assistant interactions. Use the metric to monitor and reduce negative emotion rates through targeted
assistant improvements.Figure 5. Conversations with Negative Emotions
Average Inferred CSAT Over Time
This area of the dashboard shows daily average of inferred CSAT scores in the selected data range. The CSAT scores are measured on a
scale from 0 to 5, where 0 represents the lowest satisfaction and 5 represents the highest. This chart highlights periods of improvement or decline in user sentiment.Figure 6. Average Inferred CSAT Over Time
Transfers and Escalations Over Time
This area of the dashboard tracks the number of conversations transferred or escalated to live agent. Hover over the trend line to view the number of conversations transferred or escalated to live agent on a given day. This chart
helps you with how often assistants require human intervention.Figure 7. Transfers and Escalations Over Time
Average Inferred CSAT (Virtual Agent)
This area of the dashboard shows the average Inferred CSAT score for Virtual Agent interactions in the selected period. For conversations involving both Virtual Agent and live agent, this score reflects only
the Virtual Agent CSAT. Scored on a 5-point scale, 0 = least satisfied and 5 = most satisfied. Use this metric to benchmark assistant performance and prioritize improvements where satisfaction is lowest.Figure 8. Average Inferred CSAT (Virtual Agent)
Average Inferred CSAT (Live Agent)
This area of the dashboard shows the average Inferred CSAT score for live agent interactions in the selected period. For conversations involving both Virtual Agent and live agent, this score reflects only the live agent CSAT.
Scored on a 5-point scale, 0 = least satisfied and 5 = most satisfied. Use this metric to benchmark assistant performance and prioritize improvements where satisfaction is lowest.Figure 9. Average Inferred CSAT (Live Agent)
Average Inferred CSAT (Session)
This area of the dashboard shows the average Inferred CSAT score for all interactions handled by Virtual Agent or a combination of Virtual Agent and live agent in the selected period. Scored on a 5-point scale, 0 = least satisfied
and 5 = most satisfied.Figure 10. Average Inferred CSAT (Session)
Assistant Recommended Next Steps
This area of the dashboard shows how clearly the assistant explained what happens next or what the user should do. Low: Conversations where no clear guidance was provided. Medium: Conversations where some guidance was provided.
High: Conversations where clear and complete guidance was provided. Figure 11. Assistant Recommended Next Steps
Conversation Insight Inferred Resolution State
This area of the dashboard shows the conversations where the user's issue was resolved. Yes: conversations where the assistant met the user's needs. No: conversations where the assistant didn't meet the user's needs. Unknown:
conversations where the resolution state couldn’t be determined.Figure 12. Conversation Insight Inferred Resolution State
Empathy Levels Distribution Over Time
This area of the dashboard shows the distribution of empathy levels (High, Medium, Low) in assistant responses in the selected date range. Use this chart to assess the emotional intelligence of assistant interactions and target improvements.Figure 13. Empathy Levels Distribution Over Time
Negative Emotion Feedback Over Time
This area of the dashboard tracks the feedback related to negative emotions: frustration and confusion in conversations with assistants. This chart helps you identify trends in negative user experiences and take necessary steps to
reduce the negative feedback.Figure 14. Negative Emotion Feedback Over Time
