Rubric scoring
Rubric scoring is a method of grading an interview where each predefined criterion gets its own numeric score and its own written justification, rather than collapsing the whole conversation into one composite rating.
Direct answer
A rubric is a written list of criteria with a score scale per criterion and (in the strong version) example answers for each band of the scale. Rubric scoring applies the rubric to a candidate's interview: each criterion gets a 0–5 or 1–4 score plus a short justification, ideally with a transcript quote attached. It matters because per-criterion scoring tells you why a candidate is being advanced or rejected — and that "why" is what regulators, candidates, and your own future hiring panel need. Who needs it: any team that wants comparable signal across interviewers.
In detail
What a defensible rubric contains
Four moving parts: (1) criteria — usually 3–6 dimensions you actually care about for the role, (2) a scale per criterion, often 1–4 or 0–5, (3) anchor descriptors per band ("1 = no example, vague answer" up to "4 = specific situation, owned outcome, second-order detail"), and (4) weights per criterion if some matter more than others. The first three are non-negotiable; weighting is optional but it forces an uncomfortable conversation about what the role actually values, which is the conversation worth having before the first interview.
When it applies
Rubric scoring is most useful when you will compare candidates side-by-side, when more than one person will look at the result, or when you may need to defend the decision later. It is less useful when there is one decision-maker, one candidate, and no audit trail — but even there, a rubric forces clearer thinking and is cheap to keep.
Common misconceptions
A rubric is not a checklist. A checklist has yes/no items; a rubric has graded scales with descriptors. Rubric scoring is also not the same as an overall "1 to 10 interview score" — that's exactly what rubric scoring is meant to replace, because a single number hides which criterion the candidate lost on. Finally, an AI agent's rubric score is not a hiring decision — it is structured input the recruiter uses to decide who advances.
Related concepts
Rubric scoring is the scoring half of a structured interview. When an AI conducts the interview, it produces a rubric score that goes into the broader AI first-round interview report. If that score "substantially assists" a hiring decision, the system is an AEDT under NYC Local Law 144 and the per-criterion scores are the audit material a bias auditor uses to compute the adverse impact ratio.
How Vettika handles it
Vettika expects a per-criterion rubric and produces per-criterion scores with a quoted justification from the transcript. No mystery composite. The model version that scored a given session is pinned in the audit log — see compliance for the audit-log surface.
Try a first-round interview, no card.
Bring your rubric. Get back per-criterion scoring with quotes from the transcript.
Run your first 3 free