Dashboard and Analytics
The analytics layer is split into two surfaces: the dashboard is the executive overview and screenshot-ready command center, while the reports page is the filterable drilldown for managers and operators.
1) Dashboard vs Reports
| Screen | Primary purpose |
|---|---|
| Dashboard | Quick KPI overview, alerts, chart summaries, and screenshot-friendly widgets. |
| Chat > Reports | Drilldown analytics with date, widget, department, agent, sentiment, confidence, and lead filters. |
2) Metric Definitions
The reporting service keeps metric definitions explicit so dashboard totals and report totals do not drift apart over time.
| Metric | Definition in this build |
|---|---|
| AI-handled | Conversation contains at least one AI reply and no public agent reply. |
| Agent-handled | Conversation contains at least one public agent reply. |
| Escalation rate | Conversations with at least one escalation divided by total conversations in range. |
| Lead capture rate | Conversations with an attached lead divided by total conversations in range. |
| Lead conversion rate | Converted leads divided by total leads in range. |
| Most asked questions | Normalized opening visitor question clusters. |
| Most used sources | Knowledge source references stored in conversation message source usage. |
3) Available KPI and Chart Widgets
- Total conversations
- Open and pending counts
- Breached SLA count
- Average first response time
- Average resolution time
- Confidence distribution
- Sentiment distribution
- Frustration trend
- Open escalations
- Knowledge gap trend
- Lead funnel
- Lead conversion rate
- FAQ suggestion count
- Agent snapshot
- Unresolved chat list
4) Filters and Cache Behavior
Reports can be filtered by date range, widget, department, assigned agent, sentiment, confidence, and lead status. Results are cached per workspace and filter set for five minutes.
5) How to Read the Numbers
- Compare AI-handled vs agent-handled before enabling AI on more widgets.
- Watch escalation rate and low-confidence topics together, not separately.
- Use lead capture rate to judge whether pre-chat or post-chat forms are too aggressive.
- Track most-used sources to identify stale or missing training content.
- Use unresolved chats and agent snapshot together to spot staffing bottlenecks.
6) Demo and CodeCanyon Tips
The dashboard is one of the most valuable screenshot surfaces in the product. Import the demo workspace from Platform Ops before taking preview screenshots so cards, charts, leads, tickets, and knowledge widgets are populated with believable data.