How evaluation works
Every response is scored against the rubric attached to the assessment, which comes in two kinds: a level-based matrix or classic score entry. Because the rubric is shared, all evaluators work from the same criteria and the same definitions. Attached rubrics are frozen snapshots — later edits to the bank rubric never change what evaluators see on an assessment already using it.
Scoring with a matrix rubric
A matrix rubric defines levels, each with a label and a score, and a behavior description per level for every criterion. To score, read the response, then click the level card whose description best matches what the candidate actually did — that level's score is recorded for the criterion. This keeps scoring anchored to observable behavior rather than gut feel, and it makes disagreements between evaluators concrete and discussable.
Scoring with a classic rubric
Classic rubrics use direct numeric score entry per criterion. Enter the score that reflects the response against the criterion's definition. Criterion weights still apply when scores combine, so a heavily weighted criterion moves the task score more than a lightly weighted one. Note that if a task has both a rubric score and an auto score, the rubric score wins.
Notes as evidence
Evaluator notes attach to each task's evaluation alongside the scores, so the reasoning travels with the numbers. A short note explaining why a response earned its level turns a bare number into a defensible judgment — it surfaces later in the candidate evidence report and the audit pack. Every score in the record is attributed to a named evaluator with a timestamp, so the decision file always shows who judged what, and when.
Multiple evaluators on the same response
Multiple evaluators can score the same response independently. Within each criterion, their scores are averaged; criteria then combine by rubric weights into the task score, which rolls up through skills and section weights into the role-fit percentage. Independent scoring plus averaging dampens any single evaluator's bias while keeping every individual judgment visible in the record.
At a glance
- Evaluators score against the shared rubric, so every candidate is judged on the same criteria.
- Matrix rubrics: click the level card whose description matches the response — that level's score is recorded.
- Classic rubrics: enter a numeric score per criterion.
- Evaluator notes attach to each task's evaluation as evidence, and every score is attributed to a named evaluator with a timestamp.
- When multiple evaluators score the same response, their scores are averaged within each criterion.