scoring-method.md
title: Scoring method description: How to force-rank and weight the criteria, the 1 to 4 anchors, must-haves, and how to handle missing evidence.
Scoring method
Force-rank, then weight
- List the six to eight criteria.
- Rank them with no ties. If two feel equal, ask: "If a candidate is strong on one and weak on the other, which candidate would we rather hire?" That answer breaks the tie.
- Spread 100 points across them in rank order. Each criterion must get fewer points than the one above it. The top criterion usually takes between a fifth and a third of the points.
- Mark up to two must-haves. A must-have scored 1 is a fail regardless of the total.
- Write down the date, the version and who agreed it. If the scorecard changes mid-process, every candidate scored so far gets re-scored against the new version.
The weights are yours. They encode what this seat needs, which no outside template knows.
The 1 to 4 anchors
Four points, no middle score. A middle score lets interviewers avoid a call.
| Score | Meaning | The evidence sounds like |
|---|---|---|
| 4 | Has done this, at this level, repeatedly | Several specific examples with numbers, names of roles and the outcome |
| 3 | Has done this, close to this level | One strong example, or several at a slightly different scale |
| 2 | Adjacent experience, would need to stretch | Has done something similar in a different motion, and can explain the gap |
| 1 | Has not done this, or the evidence contradicts the claim | Vague, hypothetical ("I would...") or numbers that do not add up |
Evidence rules
- Every score cites the answer that earned it, in a phrase or a quote.
- A hypothetical answer ("I would start by...") can never score above 2. The scorecard measures what happened.
- A claim with no example is not assessed, not a 3. Send the question to the next interviewer.
- When two interviewers score the same criterion two or more points apart, that gap goes to the debrief as the first item, before any totals.
Weighted total
Multiply each score by its weight, add them up, and divide by 4. The result is out of 100. A criterion not yet assessed counts as 0 and is flagged on the summary, so a gap in the evidence never looks like a strength. Treat totals within five points of each other as a tie, and break the tie on the top-ranked criterion.
Rescoring
The scorecard is only as good as its last update. After every hire, look back at month six: which criteria predicted the outcome and which did not. Move the ones that did up the ranking next time.