Rubrics: classic and matrix

Classic score-entry rubrics

A classic rubric lists criteria and lets evaluators enter a numeric score for each one. It is the lighter option: quick to build and quick to score with, which suits experienced evaluators and criteria that do not need written anchors. During evaluation, the evaluator types a score per criterion and can attach notes as evidence for how they arrived at it.

Level-based matrix rubrics

A matrix rubric defines a set of levels, each with a label and a score, and then describes what each level looks like for every criterion. These per-criterion behavior descriptions anchor the scale: instead of choosing a number, the evaluator clicks the level card whose description matches the response, and that level's score is recorded.

Anchored descriptions are what keep multiple evaluators consistent — everyone scores against the same written definition of each level, not their private sense of what a 3 means. When several evaluators score the same response, their scores are averaged within each criterion, so shared anchors directly improve the reliability of the final number.

Criterion weights

Each criterion carries a weight, expressed as a multiplier. A criterion weighted x2 counts twice as much as one weighted x1 when the rubric combines criterion scores. Use weights to reflect what actually matters for the role rather than letting every criterion count equally. In the scoring rollup, evaluator scores are averaged within a criterion and criteria combine by these weights.

Showing the rubric to candidates

You can optionally make the rubric visible to candidates during the assessment, and optionally include the criterion weights as well. This tells candidates exactly what they are being judged on, which encourages responses aimed at the criteria you care about.

Visibility never extends to results: scores are never shown to candidates, regardless of the rubric visibility setting. Candidates can see the criteria they will be judged against, and optionally how those criteria are weighted, but the scores evaluators record stay inside your workspace.

Snapshots: how rubric edits reach assessments

When you attach a rubric to an assessment, SkillCort freezes a snapshot of it. Later edits to the bank rubric never silently change assessments that have already been delivered — candidates who took the assessment were scored against a specific rubric, and that record must stay stable.

Drafts are the exception by design: publishing re-freezes every snapshot from the current bank rubric, so a draft always goes live with the latest version. On a live assessment, you update a task's rubric deliberately — edit and re-save the task in the builder, and its snapshot re-freezes with your changes. Closed and archived assessments are locked; duplicate them to a new draft instead.