One bar. Written down first.

how the system behaves, not a benchmark

Interviews Designed for Fairness.

Great interviews do not happen by chance, they are structured. The rubric is written from the role before anybody applies, so the standard cannot drift as the day wears on. Scores come with the evidence attached, which means you can argue with them.

How the interview gets built.

• 01
  The role is read first

  The job description is turned into the skills and tools the role actually needs, before a single question exists. The interview is derived from the job, not from a library of questions someone liked.

• 02
  Sections, then questions

  An opening round to settle in, a middle round where material appears on screen and gets discussed aloud, and a closing round. Roles that call for it also get a walkthrough of the candidate’s own resume, a whiteboard, a coding section, or a mock sales call.

• 03
  Every question arrives with its rubric

  Two to four named criteria, the signals a strong answer contains, and a time budget. All of it written once, before anybody sat down, so the standard cannot drift as you get tired of reading.

• 04
  The rubric does not move once the interview starts

  Criteria are fixed before anybody sits down, so an answer is judged against the standard the role asked for rather than against the last person you spoke to.

How an answer gets scored.

• 1 to 5
  Per answer, against that question’s rubric

  Each named criterion is scored separately, and a quote from the answer is attached to the score. You are never asked to trust a number on its own.

• 0 to 100
  For the interview as a whole

  Average quality, weighted by how much of the interview was actually reached.

• 50
  For a question never reached

  Unanswered is not zero. A short interview reads as inconclusive and leans back on the paper, rather than being scored as a failure.

• -1
  Only when slow and wrong

  Running long costs a point only past one and a half times the budget, and only when the answer was weak anyway. Thinking slowly is not the same as not knowing.

The room, and what watches it.

Interview integrity is off by default and turned on per organisation or per role. When it is on, the candidate is told exactly what is watched before they agree to anything, and the session is recorded with that stated up front.

• advisory
  Recorded, never blocking

  More than one face, no face at all, where the eyes are pointed, framing, a phone in shot, a tab switch, a second monitor. All of it is logged for you to look at. None of it stops the interview or touches a score.

• blocking
  The Guard app, on the candidate’s own machine

  It watches the operating system for windows deliberately hidden from screen capture, for screenshots, and for capture permissions being granted mid-session. If it is not paired, has gone quiet, or has seen something strong, the room does not open.

• never
  A signal that changes a score

  Nothing the proctor sees feeds any number. Blocking a room is not rejecting a person, and the decision stays with you either way.

What it will not do.

• It will not rewrite the rubric after reading an answer.
• It will not score culture fit, leadership, ownership or go-getter. The criteria validator rejects them before ranking starts.
• It will not put a human in the room. You watch the recording afterwards, at speed, with the transcript beside it.
• It will not let a proctoring flag move a score, or reject anybody on its own.