Viewing AI Agent analytics for Zoom Virtual Agent
Zoom Virtual Agent for voice and chat provides comprehensive reporting capabilities that help you monitor, analyze, and optimize your conversational AI performance. These advanced analytics tools allow you to measure containment rates, understand knowledge base effectiveness, evaluate agent performance, and analyze individual customer engagements in depth. With these insights, you can continually refine your AI agent implementation, identify areas for improvement, and quantify cost savings from AI-driven automation.
Requirements for accessing AI agent analytics for Zoom Virtual Agent
- Account owner or admin privileges; or relevant role with Virtual Agent permission
- Voice/Chat SKU purchased for Zoom Virtual Agent
How to access AI Agent analytics for Zoom Virtual Agent
- Sign in to the Zoom web portal.
- In the navigation menu, under Personal, click Analytics & Reports.
- Click the Virtual Agents tab.
- Select AI agent performance to view performance metrics and use the channel filter within reports to separate chat/voice.
Reporting capabilities
Overview tab
The overview report provides a high-level summary of your Zoom AI Agent performance with key metrics that help you evaluate overall effectiveness and business impact.
Total engagements/Total interaction duration
This section displays key metrics on which you are billed (duration for voice/ engagements for chat):
- Historical duration/engagement count trends
Containment rate
This critical metric shows the percentage of voice/chat engagements successfully handled by your virtual agent without requiring transfer to a human agent. A high containment rate indicates effective AI resolution capabilities, similar to the self-service rate (SSR) metric in chatbot analytics.
Topic analytics
Topic analytics uses AI to identify:
- Popular topics: Frequently asked topics in customer inquiries.
- Solved topics: Topics that are contained effectively by AI agents.
- Escalated topics: Topics that are frequently escalated to human agents. They can help identify gaps and guide the development of new skills, tools, or flows to handle them more effectively.
This feature helps you discover where to focus your attention to fine tune your AI agents. Clicking on individual topics helps you see the queries, and you can even download those for further analysis.
Cost savings analysis
This section provides estimated cost savings generated by your virtual agent by automatically calculating:
- Total engagements for chat/total duration for voice calls handled by AI.
- Average cost of human-handled interactions.
- Projected savings based on successful containment.
This helps quantify the ROI of your virtual agent implementation for your organization by adopting AI agents.
AI Agent tab
The AI Agent report details individual agent performance metrics to help you identify top performing agents, sharing best practices and improvement opportunities.
Tools
This section shows which AI agent tools are being most effectively utilized:
- Unique tools invoked: Total tools invoked to understand the breadth of usage of additional support set up in addition to conversational experience.
- Total tools invocation: Unique tools invoked to understand the depth of usage of setup tools.
- Tools usage leaderboard: Leaderboard of top tools used across all agents.
Latency analytics
This section shows how long AI agents take to respond to consumer queries. Key metrics include:
- P50 latency (Median): Reflects the typical response time experienced by users.
- P90 latency: Highlights latency after removing extreme outliers.
- Historical trends: Track latency over time to identify patterns or performance issues.
Individual agent containment rates
This section shows containment performance across your deployed AI agents, helping you identify:
- Agents with the highest containment rates
- Agents with the lowest latency offering good customer service
- gents with the highest engagement volume contributing to the greatest cost savings
Using agent performance report for optimization
- Identify top-performing agents and analyze their interaction patterns
- Use latency data to optimize response workflows
- Target skills build for underutilized but high-value tools
- Create performance benchmarks based on successful agent configurations
Engagements tab
The Engagement report provides detailed analysis of specific customer interactions, allowing you to review and optimize on a case-by-case basis.
Aggregated engagement metrics
This section shows compiled statistics across all engagements, including:
- Total engagements: Overall number of engagements, along with historical trends.
- Multi-turn engagements: Analysis of multi-turn interactions to understand how many high-value or complex queries your virtual agent successfully handled.
- Total Interaction duration: The cumulative time users spent interacting with your virtual agents, useful for measuring engagement depth and agent workload.
Detailed metrics per engagement
Understand metrics such as:
- Engagement outcome
- Engagement duration
- Interaction duration (billable unit for voice engagements)
- Tools used
- Associated topics
Engagement recordings and transcripts
Access voice recordings and chat/voice transcripts for in-depth analysis:
- From the list of engagements, go to the last column and click Play Recording for voice engagements or View Transcript for chat engagements.
- Click Download and select either Recording or Transcript:
- Selecting Transcript will download a .vtt file.
- Selecting Recording will download an .mp4 file.
Using engagement performance report for optimization
- Identify patterns among engagements to understand how duration and turns may result in better outcomes.
- Listen / view transcripts of individual engagements to further optimize your flows for escalated engagements.
- Understand how tools might have resulted in better containment.
Knowledge base tab
The Knowledge Base report helps you understand which knowledge base articles are effectively addressing customer queries and identify content gaps through topic analytics for escalated topics that didn't have an article match.
Aggregated knowledge base metrics
This section shows compiled statistics across all interactions, including:
- Knowledge-assisted engagements: The total number of engagements that contained at least one knowledge retrieval.
- Knowledge-assisted engagements rate: The number of engagements that contained at least one knowledge retrieval divided by total number of engagements.
- Article referenced: The number of times an article was referenced during engagements.
- Average knowledge referenced per engagement: The average number of knowledge bases accessed during each engagement.
- Average article referenced per engagement: The average number of articles referenced during each engagement.
Topic analytics
Topic analytics uses AI to identify key patterns in customer interactions, including:
- Article matched topics: Common topics in customer queries that successfully matched relevant knowledge base articles.
- Article escalated topics: Topics frequently escalated due to lack of a suitable knowledge base match, highlighting potential knowledge gaps.
This feature helps you discover alternative methods to connect your knowledge base and optimize its structure
Using the knowledge base report for optimization
- Identify your highest-performing knowledge base articles and knowledge base sources.
- Address content gaps revealed through topic analytics.
- Understand how AI combines several articles in an engagement to improve conversational experience.
How to customize the AI Agent Analytics reports
- Adjust the date range using the selector to view data for your desired timeframe.
- Filter the report by channel, agent, or other available criteria based on the report’s context.
- Click the settings icon
to customize visible columns within any table view. - In the upper-right corner, click the Export button to export the data in CSV format.
How to download AI Agent Analytics reports
- Access the AI Agent analytics report.
- From any tabular data view in the dashboard, click the Export button in the upper-right corner.