Skip to content

Methodology

How a governance judgment becomes a defensible score.

Twelve competencies, ten governance failure patterns, a 1.00–5.00 scale, and an evidence ceiling on every single score. No percentages, no personality typing, no pass mark.

The 1.00–5.00 judgment scale

EXHIBIT 1

The 1.00–5.00 judgment scale

Material deficiencyBelow expected standardAdequate: expected minimumStrong1.00Material deficiency2.00Below expected standard3.00Adequate: expected minimum4.00Strong5.00Exceptional

The scale measures demonstrated governance judgment, not confidence and not seniority.

Evidence caps

Evidence quality sets the highest score a competency can reach. A strong answer with missing evidence is capped, and the report says so explicitly.

Evidencecap
Strong5.00
Good4.50
Moderate4.00
Weak3.25
Missing2.75

Twelve competencies

Assessed through open-ended scenarios. Every response is scored against the same rubric in both languages.

  • C1Strategic Judgment
  • C2Financial Acumen
  • C3Risk Oversight
  • C4Technology, Cyber & AI Literacy
  • C5Regulatory & Legal Awareness
  • C6Human Capital & Talent Oversight
  • C7ESG & Stakeholder Judgment
  • C8Crisis & Reputation Management
  • C9Behavioral Governance
  • C10Independence & Conflict Resilience
  • C11Boardroom Contribution & Follow-Through
  • C12Board System Contribution

Ten governance failure patterns

Patterns are detected across responses, not from a single answer. A critical pattern triggers a red flag.

  • FP1Weak Risk Oversight
  • FP2Skills Mismatch
  • FP3Filtered Information
  • FP4Passivity / Deference
  • FP5Groupthink
  • FP6Incentive Blindness
  • FP7Conflict Fragility
  • FP8Crisis Unreadiness
  • FP9Compliance Formalism
  • FP10Strategic Myopia

Human review and reproducibility

  • Every AI evaluation records its prompt version, model and output, so a score can be replayed and reproduced.
  • Critical red flags lock the report until a human reviewer clears or escalates them.
  • Reviewer edits are versioned; the original AI draft is never silently overwritten.
  • Reports are produced in English and French from the same scored evidence.