Loop configuration
These apply to every candidate in this loop. Set them before scoring; the tool snapshots them at the first score and flags any later change on every sheet.
Weights (optional)
Equal weights are a decision, not a finding. The unweighted total is always the score of record; a custom weighting is shown alongside it, never instead. If you change weights, set them before the loop, write down why, and apply them to every candidate at the level.
Reporting bands (optional)
Define any ranges before the loop. One rater's sheet cannot rank candidates inside a two-point band, so read totals as groups.
Candidate details
Level, weights, and bands apply to every candidate in this loop and are set under Settings.
Curiosity and self-starting
Do not open with a question here. This block is structurally different on purpose, and it produces an evidence check rather than a score.
Read what they've made, shipped, written, or fixed. Then ask about the hardest part of it.
Follow-ups
"Who asked you to do this?" · "What did you stop doing to make room for it?" · "Who uses it now?" · "What did you learn that you didn't expect to?"
Failure modes
Initiative described rather than shown. An artifact they can't discuss below the summary level.
Score sheet
| Scored | Score |
|---|
Do not revise a score on your own sheet once it is written. Do not write an overall impression. If more than one person interviews, score independently before comparing. A one-point gap between raters is noise; anything wider is a real split — return to the evidence, never average, and reach a number you both sign.
Candidates in this browser, ranked within each level.
Compare candidates inside a level; an 11 from an analyst loop and an 11 from a director loop are different measurements. Candidates a point or two apart are not distinguishable by this instrument — resolve a tie with the rest of the loop's scored dimensions, never with a feel, and never with the artifact check.