METHOD / MFG AI Governance
Assessment methodology
How the self-assessments score, weight and map your answers.
Design principles
- Observable answers. Every question offers five levels written as things you could verify on a site walk or in a document, not as opinions such as "good" or "excellent".
- Today, not the binder. Answer for what is true at most of your plants today. If practice differs a lot between plants, run the assessment once per plant.
- Proportionate targets. Targets are set at a level that is achievable for a mid-market manufacturer within 12 to 18 months. Level 5 is not the goal for every dimension.
- Private. All scoring happens in your browser. No answers are transmitted or stored on a server.
Scoring
- Each answer maps to a level from 1 to 5. "Not sure" is scored as level 1 and flagged, because in governance an unknown control is an absent control.
- A dimension score is the average of its questions.
- The overall index converts each dimension average to a 0 to 100 scale,
(average − 1) ÷ 4 × 100, and takes the weighted average across dimensions. Weights are shown in each assessment's data and are equal unless noted. - The index maps to five maturity stages. Each stage description says what that stage typically looks like in a plant.
Action plans
Each level of each question carries a concrete next action. The results page takes the three weakest dimensions, relative to their targets, and assembles a 30/60/90-day plan from the actions attached to the levels you chose. Treat it as a starting draft for your own plan, not a prescription.
Framework alignment
| Assessment | Main references |
|---|---|
| AI governance readiness | NIST AI RMF 1.0 functions, ISO/IEC 42001 management system themes, EU AI Act roles and risk tiers |
| OT security maturity | NIST CSF 2.0 functions (Govern, Identify, Protect, Detect, Respond, Recover), ISA/IEC 62443 programme and system requirements |
| Industrial data readiness | ISA-95 / IEC 62264 context models, data quality and stewardship practice |
| EU AI Act risk triage | Regulation (EU) 2024/1689 as amended by the Digital Omnibus on AI, Regulation (EU) 2026/1744 |
References are at the level of themes and functions. The assessments do not reproduce standard text and are not a certification or conformity assessment.
Limits
Self-assessments are only as good as the answers. Use them to start a conversation between operations, engineering, IT and quality, then validate the important findings with evidence. The EU AI Act triage is orientation only and not legal advice.