J
JobRush Interview
Dashboard →

How we score

Every interview at JobRush — single, panel, or trial — is scored on the same 5-axis rubric, modeled on the behavioral criteria used by Google, Amazon, and Meta hiring committees. The same axes appear in your live coach chips and your final report card, so the score tells one consistent story.

Why these 5 axes

  • Few enough that you can hold them all in your head while answering.
  • Many enough that an answer can be weak on one axis (e.g. ownership) and strong on another (technical depth) — the whole point of axis-based scoring vs. one number.
  • Real-interview aligned. Maps cleanly to how published interview rubrics from FAANG-tier companies grade candidates. No invented criteria.

Communication

Weight 20%

Clear, confident, concise.

score10

Crisp delivery, zero filler, natural pacing, confident tone, no rambling.

score5

Understandable but uneven — some filler, occasional drift, mixed confidence.

score0

Hard to follow, dominated by 'um/like', no clear sentences.

Technical Depth

Weight 30%

Correctness + tradeoffs.

score10

Names systems precisely, discusses tradeoffs, anticipates edge cases, shows hands-on signal.

score5

Surface-level correctness, generic vocabulary, no tradeoff discussion.

score0

Wrong, hand-wavy, or avoids the technical content of the question.

Structured Thinking

Weight 20%

STAR shape, no rambling.

score10

Crisp arc — context, action, result, learning. Easy to follow without backtracking.

score5

Recognizable shape but loose — jumps around, recovers.

score0

Pure rambling. No discernible structure.

Ownership & Impact

Weight 20%

I, not we — with metrics.

score10

Clearly owned the work, named scope, gave concrete metrics ('cut p99 from 240ms to 45ms').

score5

Some ownership signal but mostly 'we'; vague outcomes.

score0

Entirely 'we'/passive. No outcomes. Could be describing a team they observed.

Role Fit

Weight 10%

Right altitude for the level.

score10

Examples and altitude match the level (senior shows scope/influence, junior shows craft).

score5

Adjacent but mismatched — strong examples from a different altitude.

score0

Off-topic or wrong altitude entirely.

How the verdict works

  • Live coach: each answer scored 0–10 on every axis by a fast turn-by-turn evaluator. You see the chips beneath the coach bubble during the interview.
  • Composite: the per-axis scores are combined using the weights above into one 0–10 number you see in the corner of your report.
  • Coaching threshold: the coach interjects when your composite falls below 7 or any single axis drops below 5. Otherwise it stays quiet with a short acknowledgment so you can keep flowing.
  • End-of-session grader: a slower, more thorough pass re-scores the whole transcript — it can reward recovery from a weak start that turn-by-turn scoring would miss. The report card shows the grader's verdict, not the turn average.
  • Panel mode: each panelist evaluates from their own lens (behavioral, technical, leadership), then we map their scores onto the same 5 axes so the report card looks identical to a single-interviewer report.

The rubric is canonical — defined in code, replayed against a frozen golden set on every change. If the rubric drifts, our test suite catches it before the change ships.