Why six dimensions
We chose six dimensions deliberately. Three is too coarse: a single number for "data and tooling" or "people and governance" hides exactly the structural causes of AI failure that the Index is meant to expose. Ten is too granular: the dimension scores become individually noisy and the overall score loses its interpretability in a board pack. Six is the inflection point where each dimension is independently meaningful and the overall score still fits in one slide.
The choice is not original. The MIT Sloan / BCG longitudinal study works at a similar resolution. The five framework anchors we use — the EU AI Act, the OECD AI Principles, NIST AI RMF, ISO/IEC 42001, and the MIT Sloan / BCG cluster of work — converge on roughly six independent constructs once you cross-walk them. We adopted that resolution rather than inventing one.
The six dimensions
1. Strategy & vision (18% weight)
What it measures: clarity of AI ambition, board-level alignment, prioritisation discipline, the cadence at which AI strategy is reviewed and refreshed. Anchored in ISO/IEC 42001 Clause 5 (Leadership) and the MIT Sloan / BCG 2024 Strategy cluster. Sample question — STR-01: "Has your board or executive team formally endorsed an AI strategy in the last 12 months, with named accountable owners and a documented refresh cadence?"
2. Data foundations (17% weight)
What it measures: data quality, governance, accessibility for AI use cases, the speed and cleanliness with which a use-case team can acquire the data it needs. Anchored in NIST AI RMF — Map function (Data) and EU AI Act Article 10. Sample question — DAT-01: "When you need to use customer data for an AI model, how easy is it to access, with documented provenance and quality criteria?"
3. People & capability (17% weight)
What it measures: AI literacy across the workforce, hiring posture for AI-adjacent roles, training cadence, role evolution as AI absorbs work that used to be human. Anchored in OECD AI Principle 2.4 (building human capacity for AI) and Stanford AI Index 2024 talent indicators. Sample question — PPL-01: "What proportion of your workforce has had structured AI literacy training in the last 12 months, and how is that proportion differentiated by role?"
4. Governance & ethics (17% weight)
What it measures: risk framework, EU AI Act readiness specifically, model oversight processes, ethical guardrails, the presence of an independent review function. Anchored in the EU AI Act, the OECD AI Principles, and the NIST AI RMF Govern function. Sample question — GOV-01: "Have you classified your AI use cases against the EU AI Act risk categories, with a named owner of the classification register?"
5. Tooling & infrastructure (14% weight)
What it measures: maturity of the AI/ML stack, vendor strategy, integration capability, platform discipline, the cleanliness of the path from experimentation to production. Anchored in NIST AI RMF — Map (Infrastructure) and ISO/IEC 42001 Clause 8 (Operation). Sample question — TOL-01: "How is your organisation provisioning AI capabilities today — vendor SaaS, in-house build, or a deliberate hybrid with documented decision criteria?"
6. ROI & measurement (17% weight)
What it measures: value tracking, the existence and discipline of an AI ROI scorecard, kill-switch culture, attribution rigour from outcome to intervention. Anchored in ISO/IEC 42001 Clause 9 (Performance evaluation) and the MIT Sloan / BCG Value capture cluster. Sample question — ROI-01: "Do you have a documented scorecard for AI investments — five to seven metrics, monthly cadence — that is reviewed at executive level?"
How weights are calibrated
The weights are not arbitrary, and the asymmetry between them is deliberate.
Strategy gets the highest weight (18%) because in the MIT Sloan / BCG longitudinal data — the most rigorous multi-year evidence base on AI value capture — the clarity and board-level alignment of an organisation's AI strategy is the single strongest predictor of every downstream outcome: pilot-to-production ratio, ROI per use case, revenue lift attributable to AI. Strategy clarity is the input that compounds the others.
Data foundations, People & capability, Governance & ethics, and ROI & measurement each carry 17%. These are the four operational dimensions on which AI value either compounds or breaks down. The published evidence does not give us a robust basis for ranking them against each other in mid-market settings, so they are weighted equally. The Index does not pretend to a precision the data does not support.
Tooling & infrastructure carries the lowest weight (14%). Tooling matters, but in the causal order it is downstream. A company with the right strategy, data, people and governance will procure or build adequate tooling within a couple of budget cycles. A company with the wrong strategy will procure expensive tooling and waste it. Tooling is the easiest of the six to fix once the others are sound, and the most expensive thing to over-invest in when they are not. The lower weight reflects that asymmetry, not a view that tooling is unimportant.
The full calibration logic, with citations, lives on the methodology page. The weights are pinned in code at src/index/dimensions.ts in the public repository.
What an Arqmetrica score actually tells you
Your score is not a grade. It is a map. The overall number tells you roughly where on the European mid-market spread your organisation sits. The dimension breakdown tells you why — and what kind of intervention will move it.
A company with a low overall score but strong governance and tooling, and weak ROI and people, has a fundamentally different problem from a company with the same overall score driven by weak governance and strong ROI. The first one needs to install measurement discipline and build capability. The second one needs to install governance before the next EU AI Act review cycle. They cannot be served by the same playbook, and the dimension breakdown is what tells you which playbook applies.
This is the practical reason we report dimensions explicitly rather than collapsing them into a single grade. A grade tells you how you compare. A map tells you what to do.
Try it
The Arqmetrica AI Maturity Index takes about ten minutes, returns your overall score, your dimension breakdown, your peer benchmark in your industry × employee-band cohort, and your three highest-leverage moves drawn from our catalogue of 24. The result is private, the methodology is fully published, and the scoring code is in the public repository. There are no hidden adjustments and no proprietary multipliers.