Open methodology
The complete instrument: every item, the scoring model, the band cut-offs, and the limitations — published so it can be inspected, criticised, and improved.
Most leadership assessments are black boxes. You answer the questions, surrender an email address, and trust the score. You cannot see how it was calculated, whether the items measure what they claim to, or what would move your number.
ALX takes the opposite bet. The questions were never the valuable part — anyone can write thirty questions. What matters is whether the instrument holds up to inspection, and whether the data it produces, aggregated across thousands of leaders, tells us something true about how leadership is actually changing. Publishing the methodology invites the criticism that makes version 2 better than version 1.
It also means anyone — a leader, a team, an entire organisation — can see exactly what they are being measured against before they answer a single question.
ALX measures behaviour: what a leader does, and how consistently, when leading humans and AI together. It does not measure personality, potential, intelligence, or values. It is not a psychometric instrument in the clinical sense and does not claim validation against long-run performance outcomes — see Limitations.
The working definition underneath the instrument:
Five pillars, derived from field work across twelve organisations rather than from a literature review. Each is a capability you can observe in what a leader does in an ordinary week.
| Pillar | What it captures | Items | Weight |
|---|---|---|---|
| 01 · AI Fluency | How effectively you use, model, challenge, and scale AI as a core leadership capability — not just as a productivity tool. Without this pillar, none of the others fully function. | 6 | 30% |
| 02 · Human + Agent Orchestration | How effectively you design work that blends human judgment and creativity with AI execution — building systems that outperform either working alone. | 6 | 25% |
| 03 · Decision Velocity | How effectively you use AI to make faster, higher-quality decisions — and create the conditions for your team to do the same. | 6 | 20% |
| 04 · Talent Multiplication | How effectively you amplify what your people achieve — using AI to elevate individual and team performance beyond traditional management limits. | 6 | 15% |
| 05 · Learning Agility | How consistently you — and your team — update, adapt, and evolve your leadership approach as the AI landscape continues to shift. | 6 | 10% |
Thirty items, six per pillar, all answered on the same five-point agreement scale. Every item is a first-person behavioural statement — "I have actively redesigned workflows in my team to incorporate AI tools" — and you rate how strongly it describes you.
Median completion time: 8–10 minutes. Every respondent answers all thirty items; none are optional or branching. Respondents also select a lens before starting — Individual Leader, People Manager, or HR / L&D Leader — which frames the report, not the questions: all three answer the same thirty items.
Verbatim from the live instrument at alx.agentic-leaders.com. What you read here is what respondents answer.
Deliberately simple. Complexity in scoring is where black-box assessments hide. This is the exact computation the live instrument runs:
pillar_raw = sum of 6 item responses → range 6–30
pillar_score = round(pillar_raw / 30 × 100) → range 20–100
ALX = round( 0.30 × AI Fluency
+ 0.25 × Orchestration
+ 0.20 × Decision Velocity
+ 0.15 × Talent Multiplication
+ 0.10 × Learning Agility )
The pillars are not weighted equally. AI Fluency carries the most (30%) because it is the foundation — without it, none of the other pillars fully function. Orchestration carries 25% as the highest-leverage skill of the next decade. The weights descend from there.
Be clear about what the weights are: a design judgement from field experience, not an empirical finding. Version 1.0 asserts them; the accumulating dataset earns the right to revise them. If the data shows a different pillar predicts the outcomes that matter, the weights will change in a future version and the change will be documented.
One consequence of the formula worth stating plainly: because a pillar's raw score divides by 30 rather than being rescaled from its true minimum of 6, the lowest achievable pillar score is 20, not 0. The overall ALX therefore also bottoms out at 20. This is documented rather than hidden; a future version may renormalise the scale, and if it does, the change and its effect on historical scores will be published.
Worked example. A leader's six Orchestration responses are 4, 3, 3, 2, 4, 3 → raw 19 → 19/30 × 100 = 63. Five pillar scores of 73, 63, 67, 60, 70 combine as 0.30×73 + 0.25×63 + 0.20×67 + 0.15×60 + 0.10×70 = 67 — Integrator.
| Band | Range | What it means |
|---|---|---|
| Orchestrator | 76–100 | Operating at the frontier — leading humans and AI together with clarity, intent, and measurable impact. Not just ready for the agentic enterprise; helping to define what it looks like. |
| Integrator | 51–75 | Actively embedding AI into your leadership practice and seeing real impact — ahead of the majority of leaders today. The challenge now is moving from individual capability to multiplying it across your team and organisation. |
| Adopter | 26–50 | Beginning to use AI in your work and experimenting with new approaches. Adoption is not the same as transformation: the next step is moving from using AI as a tool to leading with it as a capability. |
| Bystander | 0–25 | Aware that AI is changing how organisations operate, but not yet meaningfully integrating it into how you lead. The risk is not ignorance — it is inaction. |
The cut-offs are even quartiles of the score range — a design choice, not an empirical finding. If the accumulating data shows the meaningful behavioural breaks sit elsewhere, the cut-offs will move in a future version and the change will be documented.
Stated here because an instrument that hides its limitations is asking you to trust it instead of inspect it.
ALX asks leaders to rate their own behaviour on an agreement scale, and people over-report desirable behaviour — agreement scales are especially exposed to this. Two candidate improvements are on the table for future versions: rewording items to behavioural-frequency anchors, and adding a team-rated variant, which is where the honest signal usually lives. Until then, read individual scores as self-assessment, not audit.
Respondents currently self-select — they found the instrument through my writing or network, which skews the early data toward the AI-curious. Until the dataset is large and organisationally diverse, aggregate numbers describe the sample, not the leadership population. This is why no aggregate findings are published on this site yet: as of v1.0 the response base is in the dozens, and dozens is a pilot, not an index.
ALX has not been validated against downstream performance outcomes. It measures behaviours that field experience suggests matter. Whether high scorers actually outperform is a claim the longitudinal data must earn.
The instrument encodes what leading with AI looks like in 2026. Some items will age. That is what versioning is for.
ALX follows plain versioning: item wording changes, scoring changes, or band changes each trigger a new version, documented in a public changelog. Scores are always reported alongside the version that produced them.
Individual responses are stored against the respondent's email for their own report, and enter the aggregate dataset anonymised. The aggregate feeds the annual Agentic Leadership Index, the first edition of which publishes in January 2027 — with its sample size and composition stated plainly, whatever they are.
Burleigh-Sheard, G. (2026). ALX™ — The Agentic Leadership Index, v1.0: Open methodology. agentic-leaders.com/alx
The instrument may be cited, critiqued, and used for individual self-assessment freely. Organisational deployment and benchmarking against the ALX dataset are licensed — enquire here.
You've seen exactly what it measures. Eight minutes to find out where you sit.