Calibrate what a 1 through 4 actually means before four interviewers disagree about it in the debrief

You are writing the interview scoring rubric for a role — the calibration document a panel reviews together in a 15-minute pre-loop meeting, before any interviews happen. This is not the per-candidate scorecard; it is the shared reference that defines what each score level means, so that a "3" from one interviewer means the same thing as a "3" from another.

ROLE
Customer Success Manager, mid-market segment

COMPETENCIES TO CALIBRATE
Handling an angry customer escalation, identifying an upsell opportunity organically, and cross-functional advocacy for a customer's feature request.

A REAL PAST CANDIDATE ANSWER PER COMPETENCY, IF AVAILABLE
For the escalation question, one past candidate said they'd 'apologize and offer a discount immediately' with no diagnostic questions asked first.

COMMON DISAGREEMENT THIS PANEL HAS HAD
Two interviewers disagreed on whether a candidate who admitted they'd never used our specific CRM tool should be scored down on the upsell competency.

For each competency, write a short definition of what it means specifically in this role (not a dictionary definition of the competency in general), then four behavioral anchors for scores 1 through 4, each describing what an interviewer would actually observe or hear, phrased so two different people watching the same answer would independently land on the same score. If a real past candidate answer is provided for a competency, use it as a worked example: state what score that answer would earn under this rubric and exactly why, since a worked example calibrates a panel faster than an abstract anchor alone. Where the past-disagreement note describes a specific point of confusion (e.g., the panel couldn't agree whether a candidate who used an AI tool during a live coding round should be scored down), resolve that exact question explicitly in the rubric rather than leaving it as unaddressed edge-case ambiguity that will resurface in the next loop.

STRUCTURE THIS AS A CALIBRATION DISCUSSION, NOT A FORM
Unlike a scorecard, this document should read as guidance to be discussed and adjusted by the panel before it's finalized — include one open question per competency where the anchors are genuinely debatable, so the panel has something concrete to align on rather than silently rubber-stamping a document nobody actually agreed to.

OUTPUT FORMAT
One section per competency: definition, four scored anchors, worked example (if given), and one open calibration question. End with a short paragraph on how this rubric relates to but differs from the scorecard interviewers will fill out per candidate.