✏️Prompts

AI Agent KPI Dashboard Spec Prompt

Prompt

You are a support analytics architect. Define the complete specification for an AI agent performance monitoring dashboard for the support operations team.

[PASTE: AI agent platform and available data exports]
[PASTE: Primary stakeholders — ops manager, head of support, CX director]
[PASTE: Current reporting cadence — daily standup, weekly leadership review, monthly QBR]
[PASTE: Any existing metrics already tracked]

YOUR TASK:
1. Define 8–10 core KPIs with exact calculation logic: containment rate, CSAT, escalation rate, avg session length, intent recognition accuracy, resolution time, cost-per-contact deflected
2. For each KPI, specify the data source, update frequency, and alert threshold
3. Design the dashboard layout: which KPIs go on the executive view vs. the operational view
4. Define drill-down requirements: from aggregate to topic-level to individual session
5. Specify 3 automated alerts the system should send and the trigger conditions

OUTPUT: {kpi_definitions_with_calculations, data_sources_and_update_cadence, dashboard_layout_spec, drill_down_requirements, automated_alert_specs}

Why it works

Separate executive and operational views prevent leaders from micro-managing and ops teams from missing strategic signals. Pre-defined alert thresholds enable real-time response rather than weekly review.

Watch out for

Dashboards with 20+ metrics get ignored. Enforce a maximum of 5 metrics on the executive view — everything else lives in the operational layer.

Used by

Customer Success ManagersIT & Ops Teams