The formula

score = Σ (dimension score × weight). Within a dimension, skill score = min(student level ÷ required level, 1.0), weighted by how critical that skill is to the role — critical 3, important 2, optional 1.

Weights are set per role, so the same student scores differently for a Business Analyst and a Java developer. Bands: 80+ ready, 65–79 almost ready, 50–64 needs improvement, below 50 high risk. All of it is configurable.

Points recoverable

Next to every gap is the number of readiness points closing it returns. That figure is not an estimate — the engine re-scores the student with that one skill lifted to the required level and reports the real difference. An automated test asserts the displayed number equals the actual delta, and another asserts that completing a plan produces exactly the score the plan promised.

From answers to levels

For each skill: level = (difficulty of questions answered correctly ÷ difficulty of all its questions) × 5, rounded to the nearest half level. Three easy questions right and one hard one wrong gives 2.0, not the 3.75 that counting questions would give. A skill with fewer than three answered questions is left alone rather than scored from thin evidence, and the student is told which ones were skipped.

What is excluded

If a role states no requirement in a dimension and the student has no evidence in it, that dimension is dropped and its weight redistributed — a student is never penalised for something nobody assessed. Eligibility (CGPA, backlogs, branch) is plain rule logic applied separately, never scored and never inferred.