The AI Agent Analytics interface gives you a clear view of how your AI agents are performing — so you can optimise their behaviour, maximise their impact, and continuously improve the experience for your customers.
Whether you are running an AI FAQ Agent (SuePort) or an AI Product Finder (Recco), Charles provides dedicated analytics tailored to each agent type.
What Can You Measure?
Each agent type surfaces a distinct set of metrics aligned to its core purpose:
Agent | Reporting Focus |
AI FAQ Agent — SuePort | Customer Support Efficiency · Cost Savings & Operational Efficiency · Customer Experience & Satisfaction |
AI Product Finder — Recco | Engagement & Adoption · Recommendation Effectiveness · Conversion & Revenue |
How to Access Your Analytics
Navigate to Charles AI Squad → Agents and select your active agent.
Open the Analytics tab at the agent level.
Use the date range selector to filter metrics for your desired timeframe.
AI FAQ Agent Analytics — SuePort
The FAQ Agent Analytics help you understand how effectively SuePort is handling customer inquiries, reducing the need for human intervention, and delivering a satisfying support experience.
Metrics Explained
Metric | Definition |
Interactions | Total number of times this agent was triggered by a customer |
Timeouts | Total number of times a customer stopped responding and the timeout was triggered |
Fallbacks | Total number of times the agent could not answer a question and sent the fallback message |
Total Escalations | Total number of times a customer pressed the escalation button to reach a human agent |
How to Interpret These Metrics
High Fallback rate? Review and expand your Knowledge Base to cover more customer queries.
High Escalation rate? Investigate whether your agent's responses are meeting customer expectations, and refine your agent's instructions accordingly.
High Timeout rate? Consider adjusting your Timeout Limit or improving the clarity of your agent's messages to encourage continued engagement.
AI Product Finder Analytics — Recco
The Product Finder Analytics help you understand how effectively Recco is engaging customers, recommending relevant products, and driving revenue.
Conversion Metrics
Metric | Definition |
Total Revenue | The amount of net revenue this agent has generated |
Total Clicks | Total number of clicks on product links |
Total Orders Placed | Total number of times a customer made a purchase following an agent conversation |
Average Order Value | Average order value of all orders placed within the selected timeframe |
Total Products Recommended | Total number of products recommended by the agent |
Average Recommended Products | Average number of products the agent recommends per conversation |
Interaction Metrics
Metric | Definition |
Interactions | Total number of times this agent was triggered by a customer |
Timeouts | Total number of times a customer stopped responding and the timeout was triggered |
Fallbacks | Total number of times the agent could not answer a question and sent the fallback message |
Total Escalations | Total number of times a customer pressed the escalation button to reach a human agent |
How to Interpret These Metrics
Low Click rate? Review your Carousel message format and CTA button copy to make product cards more compelling.
Low Conversion rate? Assess whether the products being recommended are well-matched to user queries — consider refining your attribute mapping or boosting rules.
High Fallback or Escalation rate? Revisit your product catalogue completeness and agent instructions to ensure the agent can handle a wider range of queries.
Key Performance Indicators at a Glance
Across both agent types, keep a close eye on these overarching KPIs to gauge overall agent health:
KPI | What It Tells You |
Resolution Rate | How often the agent successfully resolves a customer inquiry without human escalation |
Escalation Rate | How often the agent hands off the conversation to a human agent |
Customer Satisfaction | How customers rate their experience with the AI agent |
💡 Tip: Use these metrics to iteratively refine your agent's instructions, Knowledge Base, and attribute configuration for better results over time. Small, regular optimisations will have a compounding positive effect on agent performance.