Documentation Index
Fetch the complete documentation index at: https://semantiks.mintlify.app/llms.txt
Use this file to discover all available pages before exploring further.
Escalation Insights is an analytics dashboard inside the Contact Center > Live Chat section that gives you full visibility into conversations that were escalated from your AI agent to a human operator. Use it to understand escalation patterns, identify problematic topics, monitor team workload, and track quality metrics for escalated conversations.
Escalation Insights requires an active agent with escalation-capable integrations (e.g., Freshchat or Zendesk). If no escalations have occurred in the selected period, the charts will display an empty state.
Accessing Escalation Insights
Navigate to Contact Center > Live Chat > Escalation Insights in the left sidebar.
Date range and granularity
At the top of the page you’ll find two controls that apply to all sections:
| Control | Options | Description |
|---|
| Date range | Last 7 days, Last 30 days, Last 90 days | Filters all data to the selected period |
| Granularity | Day, Week, Month | Controls how the over-time chart buckets data points |
KPI tiles
Five headline metrics appear at the top of the dashboard:
| KPI | What it shows |
|---|
| Total Escalations | Number of conversations escalated to a human agent in the selected period |
| Escalation Rate | Percentage of total conversations that were escalated. A colored arrow indicates whether the rate is high (red, above 25%) or healthy (green) |
| Total Conversations | Total number of conversations (escalated + non-escalated) for context |
| Time to Escalate (P50) | Median time from conversation start to escalation |
| Time to Escalate (P90) | 90th-percentile time to escalation—useful for spotting outliers |
Keep an eye on the Escalation Rate KPI. A rate consistently above 25% may indicate gaps in your agent’s knowledge base or missing FAQ entries.
Escalations over time
A line chart plots two series across the selected date range:
- Escalations (left axis) — absolute count of escalated conversations per bucket
- Escalation Rate (right axis, dashed line) — percentage of conversations escalated per bucket
Use this chart to spot trends: are escalations increasing after a product launch? Did a knowledge update reduce the rate?
Escalations by topic
A horizontal bar chart shows the top topics that trigger escalations, ranked by count. Topics are auto-classified from conversation content.
If a single topic dominates escalations, consider adding or improving FAQ entries and agent instructions for that topic in your Knowledge base.
Escalations by channel
A vertical bar chart breaks down escalation counts by integration channel. This helps you understand whether certain channels (e.g., WhatsApp vs. Web) produce more escalations than others.
Escalations by assignee
A list view shows the human agents who handle escalated conversations. Each row displays:
- Agent name (or truncated ID if the name isn’t available)
- Average resolution time — how long it takes the agent to resolve escalated conversations
- Escalation count — how many escalated conversations are assigned to this agent
Use this to balance workload and identify agents who may need additional support or training.
Quality distributions
Three stacked bar visualizations show how escalated conversations score across quality dimensions:
| Dimension | What it measures |
|---|
| Task Success | Whether the user’s request was ultimately resolved |
| CSAT | Customer satisfaction score distribution |
| Sentiment | User sentiment distribution (positive, neutral, negative) |
Escalated conversations table
Below the charts, a paginated table lists individual escalated conversations with the following columns:
| Column | Description |
|---|
| Created at | When the conversation started |
| Topic | The auto-classified topic (shown as an uppercase badge) |
| Summary | A short AI-generated summary of the conversation |
| Sentiment | The detected user sentiment |
| Task Success | Whether the task was resolved |
| CSAT | The satisfaction score (if collected) |
- Click any row to navigate to the Conversation Analytics page with that conversation pre-selected.
- Use the Previous / Next buttons at the bottom to navigate through pages.
Sort through escalated conversations to find patterns. If many escalated conversations share the same topic and negative sentiment, that’s a high-priority area to improve.
Common workflows
- Reduce escalation rate — Filter by the last 30 days, check the “By Topic” chart, and update your agent’s knowledge for the top escalation topics.
- Monitor team performance — Review the “By Assignee” section to see resolution times and redistribute workload if needed.
- Track improvement — After updating your agent’s knowledge or instructions, compare the 7-day escalation rate to the previous period to measure impact.
- Investigate quality — Use the distributions section to see if escalated conversations tend to have low CSAT or negative sentiment, and drill into specific conversations from the table.
