Use Agent analytics to monitor, understand, and improve the effectiveness of AI agents across your employee support experiences.
Agent analytics provides insight into AI conversations, resolution outcomes, feedback, channel usage, tools, and topic performance.
Unlike Global analytics, which provides a platform-wide overview, Agent analytics focuses on AI agent performance. Unlike Knowledge analytics, which focuses on knowledge content performance, Agent analytics shows how effectively AI-powered experiences support employees and resolve requests.
Why use Agent analytics?
Conversation volume shows how often employees interact with AI. Agent analytics helps you understand whether those interactions are effective.
Use Agent analytics to answer questions such as:
- How much AI conversation activity is occurring?
- How often do AI agents resolve requests without human escalation?
- How helpful do employees find AI responses?
- Which agents, channels, tools, or topics are performing well or poorly?
- Where should we improve AI configuration, knowledge support, or escalation handling?
Reviewing activity, effectiveness, and outcome metrics together helps you identify opportunities to improve AI-powered service delivery.
What you can learn
Depending on your organization's configuration, Agent analytics can provide insight into:
- AI conversations and messages
- Average helpfulness
- Tier zero support resolution
- Agent-level performance and health
- Channel usage
- Most-used tools
- Most and least effective topics.
Who uses Agent analytics?
Different roles use Agent analytics to understand AI performance and identify opportunities for improvement.
| Role | Typical focus |
| AI administrators | Monitor AI effectiveness, feedback, and opportunities to improve agent configuration. |
| HR leaders | Understand whether AI is resolving requests and supporting self-service outcomes. |
| Platform administrators | Compare agent, channel, tool, and topic performance. |
Best practices
- Review usage metrics, such as conversations and messages, separately from effectiveness metrics, such as helpfulness and resolution.
- Don't treat conversation volume alone as evidence of AI effectiveness.
- Compare agent health, topic performance, and feedback before changing configuration.
- Use Knowledge analytics when weak AI outcomes appear related to missing or underperforming knowledge content.