Read your first results

How scoring works

Evaluators score each response against the shared rubric. With a matrix rubric, you click the level card whose description matches the response, and that level's score is recorded; with a classic rubric, you enter a numeric score. Evaluator notes attach to scores as evidence, and multiple evaluators can score the same response — within a criterion, their scores are averaged.

Two rules keep scores honest: if a task has both a rubric score and an auto score, the rubric wins; and an item without an answer key auto-scores as null, meaning it needs manual review — it is never counted as zero.

AI assistance supports, never decides

AI can summarize long responses and interview transcripts and suggest draft criterion-level scores, including from transcripts. Every AI output carries model and version provenance, and a named person confirms or overrides each suggestion — AI is decision support, never the decision-maker. In interview evaluation, only the candidate's own words count as evidence; accent and fluency are excluded by design.

Compare candidates on the Decision Board

The Decision Board places candidates side by side on weighted role-fit. The role-fit percentage comes from a weighted rollup: evaluator scores are averaged within each criterion, criteria combine by rubric weights, task scores roll up to skills, and skill contributions are weighted by skill weight times section weight. Risk flags appear as context for human review — they never reject anyone automatically.

Drill into the evidence

No number on the board is a black box. Open a candidate's Evidence link to see the work behind the score: their actual task outputs, the rubric notes evaluators attached, and the integrity timeline of events from the attempt. When a role-fit figure surprises you, this is where you go — read the work behind the score before you decide.

Export reports

Two exports cover most needs. The candidate evidence report ties role-fit to the blueprint, groups tasks by section when the assessment has multiple sections or weights (with weight badges and a weighted-scoring note), and includes evaluator notes and the integrity timeline — as a printable, white-label PDF. The audit pack, exported from the Reports page, is one self-contained decision file: structure and weights, plain-language methodology, every score with its named evaluator and timestamp, notes, integrity events and resolutions, rollups, and the decision log.