Per-skill ratings give you a rating for each skill tested during the interview, along with the number of questions that rating is based on, so you can spot strengths and gaps at a glance rather than digging through a transcript. If an interview is structured to test three skills (say SQL, communication, and system design), the report shows a separate rating for each of those three skills, and next to each one, exactly how many questions that rating reflects.
That question count matters more than it might seem. A strong rating built on five questions is a far more reliable signal than the same rating built on just one, and this distinction is surfaced directly rather than being hidden behind a flat score with no supporting context.
Per-skill ratings live inside the AI-generated report, in the same section as the technical and communication scores, sentiment data, transcript, and recording. This is separate from the human-filled interviewer scorecard, which holds the interviewer's own independent judgment. In practice, when you open a candidate's report after an interview, expect the per-skill ratings to appear as a distinct panel, likely near the top alongside the headline scores, functioning as the quick "strengths and gaps" view before you drill into the full transcript or question-by-question detail.
This feature is built specifically with technical hiring panels in mind, so that engineers can debate the same evidence rather than relying on individual memory of how the conversation went. The exact visual layout of this panel (whether it's shown as a chart, a list, or bar-style ratings) hasn't been published in detail, so if the presentation format matters for how your team will actually use it day to day, that's worth confirming with a live demo. In short: per-skill ratings are a standard part of every report, they include a question count for context, and they sit inside the automated AI report alongside your other scoring data.
