The “Business Performance” dashboard is the core data window of the MarsMind backend. It is not just a data display board, but also your command center for optimizing AI strategies. Through integrated core metrics, trend charts, and user insights, you can grasp the AI’s working status and service effectiveness at a glance.
Core Metric System
The metric cards at the top of the dashboard present the “health status” of the AI service. We divide them into two categories:
1. Volume Metrics
Reflect business scale and AI load pressure.
- Received: Total number of all user messages received.
- Inquiries: Number of messages identified as valid inquiries and triggering the AI processing flow.
- Replies: Number of messages to which the AI actually sent a reply.
- Suggestions: Number of reply suggestions generated by AI for manual review (applicable to human-machine collaboration mode).
- Learning: Number of times AI triggered the self-learning mechanism.
2. Quality Metrics
Reflect the service capability and effect of AI, key to optimization.
| Metric | Meaning | Optimization Suggestion |
|---|
| Inquiry Rate | Inquiries / Received | If too low, you may need to optimize the welcome message to guide users to ask questions. |
| Reply Rate | Replies / Inquiries | Most core metric. If low, it indicates insufficient knowledge base coverage, and knowledge needs to be supplemented in time. |
| Adoption Rate | Actual Sent / Generated Suggestions | Key metric for human-machine collaboration. Reflects the accuracy and usability of AI answers. |
Pro Tip: Focus on the change in “Reply Rate”. A mature AI Agent’s reply rate should rise steadily with the improvement of the knowledge base.
Trend Analysis
Through the timeline view, you can discover business patterns and abnormal fluctuations.
- Volume Trend: View fluctuations in absolute values such as received volume and inquiry volume. For example, finding a surge in consultation volume on Friday nights may require arranging additional manual duty.
- Ratio Trend: Monitor the stability of service quality. For example, after a new product launch, check if the reply rate drops in the short term (indicating the need to supplement new product knowledge).
User and Question Insights
In addition to macro data, the dashboard also provides a micro perspective:
Active Users/Groups
- Function: List of users or groups ranked by message volume.
- Value: Helps you quickly lock in potential VIP customers with high-frequency interactions, or discover users who are spamming abnormally.
Question Distribution
- Function: AI automatically classifies user inquiries (supports primary and secondary classification).
- Value: Visually displays the problem areas customers care about most (such as “price”, “after-sales”, “functions”), providing data support for product iteration and marketing strategies.
