Evidence & track record
Weight demonstrated work over interview performance. Charisma is a criterion that sneaks into every score if it doesn't have its own leash.
Decision guide
Interviews measure interviews. A matrix measures the job.
Compare candidates with a weighted hiring matrix: role skills, trajectory, collaboration, and evidence quality — with a stress test that shows exactly what would change the pick.
A typical who should i hire? matrix, scored and totaled. The line under it is the part nobody does by hand: the exact weight change that would flip the winner.
Scrappy riser wins at 72%. Lower Trajectory & learning speed by 2 points and Polished senior wins instead.
Weight demonstrated work over interview performance. Charisma is a criterion that sneaks into every score if it doesn't have its own leash.
You're hiring the person they'll be in 18 months. The senior candidate's skills are priced in; the riser's slope is the arbitrage.
Salary expectations, notice period, timezone. Real, but low weight — the moment logistics outranks skill, you're optimizing the wrong variable.
Comparing candidates head-to-head rewards whoever interviewed last. Score each against fixed criteria: what does the role actually need in month one, quarter one, year one? The matrix keeps the yardstick still while the candidates move.
Five interviewers scoring everything means five people scoring vibes. Assign each interviewer the criteria they actually observed — the matrix aggregates, and disagreements become visible instead of political.
A bad hire costs more than a slow one. Include 'keep looking' with honest scores — the cost of an empty seat, the odds of a better pipeline. If it keeps winning, the matrix is telling you the bar moved.
The opposite: it forces you to name what you're actually evaluating, so candidates aren't rejected for unnamed gut feelings. The criteria are human; the matrix just keeps them honest and consistent across candidates.
Five to seven, and each should map to something observable in your interview process. A criterion nobody evaluates is decoration — delete it or assign it.
A tie is information: both are acceptable, and the decision is now about risk preference, not merit. Flipweight's tie detection shows exactly which inputs are holding the tie — adjust the one you feel least confident about and see if it breaks.