How to read the Decision Board and evidence reports

What you'll need

• An assessment with completed, confirmed evaluations
• Familiarity with the rubric and section weights you configured
• The stakeholders who will take part in the decision
• Any integrity events reviewed or ready for review

Step 1: Treat role-fit as a weighted starting point, not a verdict

The headline number on the Decision Board is role-fit: a percentage that reflects each candidate's weighted performance across the assessment. It is the right place to start reading — it orders the field and surfaces who deserves your attention first — and the wrong place to stop. Role-fit summarizes evidence; it is not the decision.

Two candidates separated by a few points are not meaningfully ranked; they are an invitation to look closer. The board exists so you can compare shapes of performance, not just heights of bars, and the steps below are how you do that.

Step 2: Know where the number comes from

Role-fit is not a black box, and reading it well means knowing its arithmetic. Within each rubric-scored task, evaluators are averaged per criterion first, then the criteria combine according to their rubric weights. Task scores roll up to skills and sections, weighted by skill weight × section weight, and role-fit reflects those weights.

Auto-scored tasks contribute their score out of points; where a task has both an answer key and a rubric, the rubric wins. And an auto-score of null means the task has no answer key and needs manual review — it is not a zero. If a candidate's role-fit looks oddly low, check for unresolved manual-review items before drawing conclusions.

Step 3: Drill from any score into the underlying evidence

Every score on the board is a doorway. Click into it and you reach the evidence underneath: the candidate's actual task outputs, the rubric notes evaluators attached to their scores, and the integrity timeline for the session. This drill-down is the core habit of reading the board well — the number tells you where to look, the evidence tells you what you are looking at.

Use it whenever a score surprises you, in either direction. A low score on a strong candidate might trace to one evaluator's note about a specific miss; a high score deserves the same scrutiny before it anchors your decision. If an AI-drafted summary or suggestion contributed, the provenance is logged — and a named person confirmed or overrode the score.

Step 4: Compare candidates per competency, not just overall

The overall ranking hides trade-offs that matter. Compare candidates competency by competency: one may lead on technical execution while another leads on written judgment, and which gap matters depends on the team they would join — context the board cannot know.

This is also where your weighting choices come home. Section weights (×N) and skill weights shaped role-fit, so a candidate who wins overall won it on the competencies you declared most important. If the per-competency view keeps contradicting your instincts, that is worth examining — either your instincts are being corrected by evidence, or your weights need rethinking for the next assessment.

• Where does each finalist rank on the highest-weighted competency?
• Is anyone's overall score carried by a single strong section?
• Do close overall scores hide opposite competency profiles?
• Does the evidence behind the deciding competency hold up on drill-down?

Step 5: Read risk flags as context for human review

Integrity signals — a lockdown violation, second-screen detection, a copy/paste event — appear on the board as risk flags. They are context for human review, never auto-rejects. A flag means "look at this session more carefully", and the integrity timeline shows exactly what happened and when, so you can judge the event against the candidate's actual work.

Review each flag on its merits: a brief focus loss during a long task reads very differently from a pattern of events around every difficult question. Record your resolution either way — flags with documented resolutions become part of the decision record, which protects both the candidate and the process.

Step 6: Use the candidate evidence report for people outside the board

When a hiring manager or client needs to see a candidate's results without touring the Decision Board, generate the candidate evidence report — a print-ready PDF that can be white-labeled with your branding. It packages the candidate's performance and the evidence behind it in a form a stakeholder can read on their own.

Reach for it for per-candidate conversations: a debrief with the hiring manager, a client presentation, a panel discussion about one finalist. It is deliberately narrower than the audit pack — one candidate, presented for human reading, not the full defensibility record.

Step 7: Export the audit pack when the decision is recorded

The audit pack is the other export, built for a different question: not "how did this candidate do?" but "can we defend how this decision was made?" It is one self-contained decision file containing the assessment structure and weights, the rubric versions used, every score with its named evaluator and timestamp, the scoring methodology in plain language, AI provenance, integrity events with their resolutions, and the decision log.

Export it when the decision is final — for compliance, for a client who commissioned the assessment, or simply as the durable record your future self will thank you for. Because it is self-contained, it answers questions months later without anyone reconstructing what the assessment looked like at the time.

Step 8: Record the decision with its reasoning

Close the loop by recording the decision itself: who was selected or advanced, and the reasoning in terms of the evidence — which competencies decided it, how flags were resolved, why a lower role-fit candidate advanced if one did. The named people on every score already give the record its accountability; the decision log completes it.

This final step is what turns an assessment from a scoring exercise into a decision file. Every earlier step — weights, rubrics, calibration, drill-downs — exists so this record is consistent across candidates and explainable after the fact. Write the reasoning as if a candidate or an auditor will read it someday, because with the audit pack exported, they genuinely can.