KPI/KRI Definitions & Dashboard Spec

Spec MONITOR

Purpose

Defines key performance and risk indicators for AI governance with targets, thresholds, and dashboard specifications.

Related Controls

ISO Clause 9 NIST MS-1

1. Purpose

Explain why KPIs and KRIs are needed and how they drive decisions.

This document defines the Key Performance Indicators (KPIs) and Key Risk Indicators (KRIs) that measure the health, effectiveness, and risk posture of [ORGANIZATION NAME]'s AI governance program. These metrics drive data-informed decisions, surface emerging risks, and demonstrate program maturity to stakeholders.

Document Owner: [ROLE TITLE], [DEPARTMENT]

Last Updated: [DATE]

KPIs measure how well the AI governance program is performing against its objectives.

KRIs provide early warning signals of increasing risk that require attention or intervention.

2. KPI Definitions

Define each KPI with targets, data sources, and measurement frequency.

KPI IDNameDescriptionTargetData SourceFrequency
KPI-01Policy Training Completion% of employees who completed AI policy training> 95%LMS / HR SystemMonthly
KPI-02Control Implementation Rate% of required controls fully implemented> 85%GRC PlatformQuarterly
KPI-03AI Asset Inventory Accuracy% of AI systems registered vs. discovered> 95%Asset Register + Discovery ScanQuarterly
KPI-04Mean Time to Resolve AI IncidentsAverage time from detection to resolution< 4 hours (High), < 24 hours (Medium)Incident Management SystemMonthly
KPI-05Vendor Review Completion% of AI vendors with current evaluation100%Vendor ManagementQuarterly
KPI-06Deployment Gate Pass Rate% of deployments passing all gates on first attempt> 80%CI/CD PipelineMonthly

3. KRI Definitions

Define each KRI with warning and critical thresholds.

KRI IDNameDescriptionWarningCriticalData SourceFrequency
KRI-01Prompt Injection AttemptsCount of detected injection attempts> 50/week> 200/weekWAF / AI Gateway LogsDaily
KRI-02Output Quality Decline% decrease in output accuracy from baseline> 5% decline> 15% declineEvaluation PipelineWeekly
KRI-03Unregistered AI SystemsCount of AI systems found outside inventory> 1> 3Discovery ScanQuarterly
KRI-04Overdue Risk RemediationsCount of overdue remediation items> 3> 10Risk RegisterWeekly
KRI-05AI Incident FrequencyNumber of AI-related incidents per month> 3/month> 8/monthIncident ManagementMonthly
KRI-06Policy Exceptions ActiveNumber of active policy exceptions> 5> 10Exception RegisterMonthly

4. Dashboard Specification

Define the layout and components of the governance dashboard.

Dashboard Layout

Section 1 — Executive Summary (Top Row)

  • Overall governance health score (composite RAG status)
  • KPI trend sparklines (6-month view)
  • Active incident count
  • Days since last Severity 1 incident

Section 2 — KPI Detail (Middle Row)

  • 6 KPI cards with current value, target, trend arrow, and RAG indicator
  • Click-through to detailed metric history and drill-down

Section 3 — KRI Heatmap (Middle Row)

  • 6 KRI tiles with current status (Green/Amber/Red)
  • Warning and critical threshold indicators
  • Trend indicator (improving/stable/degrading)

Section 4 — Operational (Bottom Row)

  • Recent incidents list (last 30 days)
  • Upcoming reviews and deadlines
  • Open remediation items by severity

Access Control

  • Executive view: Summary metrics only (Section 1)
  • Management view: Full dashboard (Sections 1-4)
  • Operational view: Sections 2-4 with drill-down capability

5. Reporting Cadence

Define how and when KPI/KRI data is reported to stakeholders.

Reporting Schedule

  • Daily: KRI monitoring (automated, alerts on threshold breach)
  • Weekly: Operational team reviews KPI/KRI dashboard
  • Monthly: AI Program Lead presents KPI summary to management
  • Quarterly: Full KPI/KRI report to AI Governance Committee including trends, analysis, and recommendations
  • Annually: Year-over-year maturity assessment incorporating KPI/KRI trends

Report Distribution

  • AI Governance Committee: Quarterly report + dashboard access
  • Executive Leadership: Monthly executive summary
  • System Owners: Monthly metrics for their systems
  • Security Team: Weekly KRI alerts and trends

Target Review Process

KPI targets and KRI thresholds are reviewed quarterly. Adjustments require AI Governance Committee approval and must be documented with rationale.

← Back to all templates