Analysis Metrics

Analyze messages successfully responded to by Solvea in all tickets to check AI performance in conversations:

▶︎ Message Dimensions: Total messages, messages processed by Solvea, messages processed by human customer service.

▶︎ Ticket Dimensions: Total tickets, AI replies, AI processed, transferred to human agent.

▶︎ Message Data Performance (Last 4 Weeks): Weekly data and week-over-week (WoW) comparison.

▶︎ Ticket Data Performance (Last 4 Weeks): Weekly data and week-over-week (WoW) comparison.

Business Insights

I. Basic Operational Status Diagnosis

– Business Activity Analysis: Determine the “peak/off-peak seasons” or operational rhythm of the business within the selected time period by looking at total messages and total tickets.

– AI and Human Division of Labor Efficiency: Compare the proportion of messages processed by Solvea and human customer service to the total messages, as well as the proportion of AI replies and AI processed tickets to the total tickets. Analyze the extent to which AI replaces manual labor and evaluate the optimization effect of AI application on labor costs.

II. Trend and Fluctuation Insights (Combined with Weekly Data and WoW Comparison)

– Message and Ticket Trends: Observe weekly changes in message and ticket data over the last four weeks. For example, if total messages suddenly increase in a particular week, analyze the reasons in conjunction with business scenarios (e.g., promotional activities, high incidence of failures). Use the week-over-week comparison (change rate of current week’s data compared to the previous week) to determine if the business volume is continuously growing, declining, or fluctuating, and provide early warnings for business anomalies.

– AI Service Efficiency Trends: Look at the weekly data and WoW comparison of AI replies and AI processed tickets over the last four weeks to understand the stability of AI service capabilities. If the AI processing volume shows a continuous WoW decline, investigate whether the complexity of business scenarios has increased beyond AI’s coping ability, and thereby adjust the AI processing flow.

– Manual Intervention Trends: Analyze weekly data and WoW comparison of transfers to human agents. If the proportion of transfers to human agents continues to rise, it may indicate that AI service is not effectively resolving issues, requiring optimization of AI training data or rules. It could also mean that business needs are more complex, necessitating additional human service resources or adjustments to the service flow.

III. Problem Identification and Optimization Directions

– Service Breakpoint Investigation: If total messages are high but AI processing conversion is low and transfers to human agents are frequent, identify logical loopholes in AI replies (e.g., inaccurate answers to frequently asked questions). Examine the ticket routing mechanism to see if unclear categorization leads to AI misjudgment or repetitive manual processing.

– Resource Allocation Optimization: Adjust human and AI computing resources based on the peaks and troughs in the data from the last four weeks. For example, increase customer service staffing and adjust AI service response settings in advance for peak business weeks.

– User Experience Analysis: Combine message and ticket processing paths (e.g., AI to human transfer situations) to simulate user interaction flows and identify experience pain points.

Operational Steps

① After logging in, click “Report” in the left sidebar, select “Total Messages Processed by Solvea”. The redirected page will display ticket and message data successfully responded to by Solvea.

② Here, you can filter by Solvea’s response time.

③ Here, you can filter by the source channel of user messages.