CSAT Prediction Reports
Updated
CSAT Prediction Reports help you analyze predicted customer satisfaction across customer conversations at the message, case, agent, and AI Agent levels.
Predicted CSAT scores are calculated for interactions handled by both human agents and AI Agents, enabling consistent measurement of customer satisfaction regardless of who handled the conversation. Agent-level reporting also tracks the customer's predicted satisfaction at the beginning and end of an agent's ownership of a case, helping you understand the impact of individual agents on the customer experience.
Available Dimensions
Use the following dimensions to analyze predicted customer satisfaction trends and changes over time.
Dimension | Description |
Predicted CSAT Score for Message | The predicted CSAT score calculated for an individual message. |
Predicted CSAT Score (Case) | The predicted CSAT score calculated for an entire case. |
Previous Predicted CSAT | The predicted CSAT score calculated from the preceding customer (fan) message. |
Difference Between Current and Previous Predicted CSAT Ratings | The difference between the current predicted CSAT score and the score calculated from the preceding customer message. |
Predicted CSAT Delta (Per Agent) | The change in predicted CSAT during the portion of the conversation handled by a specific agent. This dimension helps measure the impact of an agent's interactions on customer satisfaction. |
Understanding Predicted CSAT Delta (Per Agent)
Predicted CSAT Delta (Per Agent) compares customer satisfaction at the start and end of an agent's ownership of a case.
Examples:
- A value of 15 indicates that predicted customer satisfaction increased by 15 points while the agent handled the case.
- A value of 0 indicates no change in predicted customer satisfaction.
- A negative value indicates that predicted customer satisfaction declined during the agent's ownership.
Use this dimension to:
- Evaluate agent effectiveness.
- Identify coaching opportunities.
- Measure the impact of agent interactions on customer satisfaction.
- Compare performance across agents and teams.
Note: The Predicted CSAT Delta (Per Agent) dimension is controlled by the CASE_CSAT_AGENT_DELTA_ENABLED dynamic property. Enable this property to make the dimension available in reporting.
Available Metrics
Use the following metrics to measure customer satisfaction trends across conversations and agent interactions.
Metric | Description |
Predicted CSAT Score | The predicted CSAT score calculated from the last customer (fan) message in the case. |
Difference Between Current and Previous Predicted CSAT Ratings | The difference between the predicted CSAT scores of the latest and preceding customer messages. |
Previous Predicted CSAT | The predicted CSAT score calculated for the preceding customer message. |
Initial CSAT Score | The most recent predicted CSAT score available before the brand's first response. |
Net Sentiment Change | The difference between the current predicted CSAT score and the Initial CSAT Score. |
Predicted CSAT Score – Agent Initial | The predicted CSAT score when an agent is assigned to a case, representing customer satisfaction at the beginning of the agent's ownership. |
Predicted CSAT Score – Agent Final | The predicted CSAT score at the end of an agent's ownership of a case, representing customer satisfaction after the agent's interaction. |
Assigned To User | The user assigned to handle the case. |
Unique Case Count | The total number of unique cases on which the user has taken action. |
Understanding Agent-Level CSAT Metrics
Agent-level CSAT metrics help you measure how customer satisfaction changes during an agent's ownership of a case.
Predicted CSAT Score – Agent Initial
Captures the predicted CSAT score at the time the agent is assigned to the case.
This score serves as the baseline for evaluating the customer's predicted satisfaction before the agent begins working on the case.
Predicted CSAT Score – Agent Final
Captures the predicted CSAT score at the end of the agent's ownership of the case.
This score reflects the customer's predicted satisfaction after the agent's interaction with the case.
Example
Consider the following scenario:
- Predicted CSAT score when the agent is assigned: 50
- Predicted CSAT score when the agent's ownership ends: 65
In this case:
- Predicted CSAT Score – Agent Initial = 50
- Predicted CSAT Score – Agent Final = 65
- Predicted CSAT Delta (Per Agent) = +15
This indicates that the customer's predicted satisfaction improved by 15 points during the agent's ownership of the case.
AI Agent CSAT Analysis
Predicted CSAT is calculated using the same scoring model for messages generated by AI Agents and human agents.
This approach enables you to:
- Analyze customer satisfaction consistently across all conversations.
- Compare outcomes between human-assisted and AI-assisted interactions.
- Include AI Agent interactions in customer experience reporting.
- Measure the impact of AI-powered engagements on customer satisfaction.
Because the same prediction methodology is used across both interaction types, you can evaluate customer satisfaction trends using a single reporting framework.
Common Reporting Use Cases
CSAT Prediction Reports help you answer questions such as:
- Which agents consistently improve customer satisfaction during case ownership?
- Which interactions show the largest decline in predicted satisfaction?
- How does customer satisfaction evolve throughout a case lifecycle?
- How do AI Agent interactions compare with human agent interactions?
- Which teams or channels deliver the best predicted customer experiences?
By combining message-level, case-level, and agent-level dimensions and metrics, you can gain deeper insight into customer satisfaction trends and identify opportunities to improve service quality and operational performance.