Template 5 — People and Change KPI Tree

FlowRidge

COMPEL Specialization — AITE-WCT: AI Workforce Transformation Expert Artifact Template 5 of 5

How to use this template

Populate one KPI tree per programme. The tree is reviewed weekly at programme level (Level 3 metrics), monthly at steering level (Level 2 drivers), quarterly at board level (Level 1 outcomes), and fully redesigned annually.

The tree is a living document. Version it on each annual redesign; mark material in-cycle changes (e.g., new drivers added mid-year) with a revision date.

People and Change KPI Tree

Tree header

Field	Value
Programme name
Tree version	1.0
Tree date	YYYY-MM-DD
Tree author
Tree owner (standing role)
Next full redesign	YYYY-MM-DD (annual)
Standing review rhythm	weekly programme / monthly steering / quarterly board / annual redesign

Level 1 — Outcomes

The aggregate states the transformation is designed to produce. Organisation-level, multi-year, board-meaningful.

Standard outcomes

Outcome	Definition	Year-1 target	Year-2 target	Year-3 target	Source of truth
Productivity	specific definition — output per unit of input, domain-adjusted
Retention	voluntary exit rate, filtered and segmented
Engagement	engagement score from specific instrument
Capability (optional)	composite: literacy coverage + skills adjacency + proficiency

Outcome choice rationale

Why these three (or four). If only three, why not four; if four, why capability is an outcome rather than a driver.

Level 2 — Drivers

Intermediate states that produce the outcomes. Programme-level, shorter-horizon.

Drivers for productivity

Driver	Definition	Feeds outcome
AI-tool adoption depth	fraction of target workflow using AI tool	Productivity
AI-tool quality-of-use	proxy-measure quality of AI use	Productivity
Manager coaching cadence	fraction of manager-report pairs with weekly 1-to-1 in past 4 weeks	Productivity + Engagement
Workflow-friction reduction	completion of named upstream/downstream process changes	Productivity

Drivers for retention

Driver	Definition	Feeds outcome
Psychological-safety score	team-level safety measure (Article 30)	Retention + Engagement
Role clarity	survey-measured clarity of AI-augmented role	Retention
Growth-path visibility	survey-measured career-path clarity	Retention
Recognition alignment	recognition pattern alignment with stated values	Retention + Engagement
Compensation competitiveness	pay-band competitiveness for key talent segments	Retention

Drivers for engagement

Driver	Definition	Feeds outcome
Meaningfulness	survey-measured meaning of AI-integrated work	Engagement
Autonomy	survey-measured appropriate latitude	Engagement
Mastery	survey-measured capability development	Engagement
Inclusion	survey + behavioural-indicator composite	Engagement + Retention

Drivers for capability (if Level 1)

Driver	Definition	Feeds outcome
Literacy coverage	role-to-level map fulfilment	Capability
Skills-adjacency coverage	skills-graph coverage vs roadmap	Capability
Observed proficiency in AI-integrated tasks	sampled proficiency assessment	Capability

Level 3 — Metrics

Specific measurements that let the organisation know whether the drivers are moving. Programme-operational, weekly to monthly.

Discipline: 3–5 metrics per driver. More is clutter.

Metrics for “AI-tool adoption depth”

Metric	Source system	Computation method	Refresh cadence
e.g., Fraction of target workflow invocations using AI tool	AI-tool telemetry	numerator/denominator per named query	daily
e.g., Distribution of use-depth across user population
e.g., Month-on-month trajectory of active users

Metrics for “Manager coaching cadence”

Metric	Source system	Refresh cadence
e.g., Fraction of manager-direct-report pairs with weekly 1-to-1 in past 4 weeks	calendar data	weekly
e.g., Fraction of 1-to-1s including AI-coaching content	sampled review + direct-report survey	monthly
e.g., Direct-report-reported coaching usefulness	engagement survey	quarterly

Metrics for “Psychological-safety score”

Metric	Specific definition	Source system	Computation method	Owner	Refresh cadence	Current value	Trajectory
e.g., Team-level safety score (Edmondson instrument)		engagement platform			annual with pulse cycles
e.g., Behavioural indicator: escalated concerns per quarter		governance-escalation log			quarterly

Metrics for “Retention risk segment”

Metric	Source system	Refresh cadence
e.g., Voluntary exit rate for high-performer segment	HRIS	quarterly
e.g., Intent-to-stay survey item (high-performer segment)	engagement platform	semi-annual
e.g., External hiring signal (job-board benchmarks for AI-adjacent roles)	external labour-market data	monthly

Repeat for each driver. 3–5 metrics per driver.

Metric quality checks

For each metric, confirm:

Specific — measured in a defined way, from a defined source.
Reliable — measurement consistent across time and observers.
Valid — actually measures what the driver claims.
Actionable — a change in the metric provides information the programme can act on.

Bias check

No vanity metrics at Level 1 or Level 2 (per Article 33).
Proxy metrics at Level 3 are combined with direct measurement where possible.
Goodhart’s Law risk reviewed: if the programme optimises the metric, does the underlying construct drift?

Wiring documentation

For each data source used:

Source system	Metric(s) populated	Wiring pattern (API / extract / manual)	Automation status	Owner	Refresh latency
HRIS (specific system)	retention metrics, role data, tenure
LMS (specific system)	literacy coverage metrics
Engagement platform	survey-based metrics
AI-tool telemetry	adoption metrics
Calendar system	coaching cadence proxy
Governance log	escalation metrics

Wiring completeness check

All Level 3 metrics have a documented source.
All automated metrics have been validated against source-system meaning.
Manual-computation metrics have documented computation and owner.

Review rhythm

Weekly (programme)

Review Level 3 metrics.
Investigate anomalies.
Decide short-term interventions.
No tree modifications.

Monthly (steering)

Review Level 2 drivers.
Review aggregate Level 3 view.
Decide intervention adjustments.
Modify tree only on explicit decision.

Quarterly (board)

Review Level 1 outcomes with Level 2 context.
Strategic adjustments.
Board engagement with forward commitments.

Annual (full redesign)

Full tree re-examination.
New tree version.
Recalibration against observed outcome trajectory.

Change-log

Version	Date	Changes	Approver
1.0		initial
1.1

Quality rubric — self-assessment of template

Dimension	Self-score (of 10)
Three-level structure (Articles 33)	10
Metric discipline (3–5 per driver)	10
Wiring rigour (documented source, computation, owner, refresh)	10
Quality checks (specific / reliable / valid / actionable; bias; Goodhart)	10
Review-rhythm integration	10
Weighted total	50 / 50