Methodology
How the readiness score is calculated
Every capability is rated on one 1-to-5 ladder, weighted into a pillar score, capped by a gate where a critical gap exists, and shown to clients as a letter grade.
The 1 to 5 scale
Every capability in the Ejentic AI Strategy Health Check is rated on one ladder of 5 levels, and each level has written wording so a rating means the same thing in any organization. Levels 1, 3 and 5 are described in full; 2 and 4 sit between them.
Level 1
Initial
Level 2
Emerging
Level 3
Defined
Level 4
Managed
Level 5
Optimized
How the pillars are weighted
Capability ratings are averaged by weight into a pillar score, and pillar scores are weighted into the overall figure. The weights are fixed for every engagement, running up to 5, so two organizations — and the same organization a year apart — are measured on the same instrument.
| Pillar | Weight | What it asks |
|---|---|---|
| 1 · Business strategy and AI value | 5 | Where AI attaches to the business, who owns it, and whether the value is proven. |
| 2 · Data readiness | 4.8 | Is the data fit for an AI use case to consume? |
| 3 · AI security and trust | 5 | Five domains protecting the data, the models and the people using them. |
| 4 · People and change | 4.2 | Communication, collaboration and capacity when AI enters the work. |
| 5 · AI operating governance | 4.4 | How governed the AI already running is, across five process domains. |
Why a gate can cap a pillar
A weighted average can hide a catastrophic gap, so 5 gates are applied after the average is taken. Where a gate fires, the pillar is held at its ceiling no matter how well the rest of the pillar scored, and the report says which gate did it.
DLP / ingestion gate
Caps at 2.0Data protection scores 1 — proprietary data reaches models with no control.
RAG authorization gate
Caps at 2.5Identity and access scores 1 — retrieval runs on credentials broader than the user.
Guardrail / execution gate
Critical flagThreat defence scores 1 — raw input reaches the model and raw output reaches systems.
Data classification gate
Caps at 2.5Classification scores 1 — you cannot govern data you have not classified.
Change capacity gate
Caps at 2.5Load honesty scores 1 — AI added on top of full workloads with no protected learning time. The single strongest stall predictor.
How scores become grades
Grades are a reading aid over the unchanged score, not a second calculation. The decimal is always shown alongside, and the bands never move.
| Grade | Score |
|---|---|
| F | 1.00 to 1.74 |
| D | 1.75 to 2.49 |
| C | 2.50 to 3.24 |
| B | 3.25 to 3.99 |
| A | 4.00 to 4.74 |
| S | 4.75 and above |