Match Score Explained

The match score is a number from 0–100. It is a weighted sum of three components.

Components

Hard match (40%)

Binary pass/fail on each hard filter the company set:

• Location / remote policy
• Contract type (full-time, part-time, contract, etc.)
• Minimum years of experience
• Required skills (all must be present)

A candidate who fails any hard filter receives a score of 0 regardless of personality fit. They do not appear in the recruiter’s ranked list.

If all hard filters pass, the hard match contributes its full weight to the final score.

Personality fit (40%)

Euclidean distance between the candidate’s 47-dimensional PUC (Psychological Unit of Competency) vector and the job’s target vector, converted to a similarity score.

The job’s target vector is derived from:

• The company’s org profile (Big Five traits from E1–E4)
• The per-job personality weight config (E5–E9)

Lower distance = higher score. The conversion is linear, normalized against the maximum observed distance in the candidate pool.

Predictive signal (20%)

A heuristic score derived from pattern-matching against past hiring outcomes for similar candidates in similar roles. This component activates as a learned model (XGBoost) once there are at least 100 historical outcomes with AUC > 0.65; until then it falls back to a simple heuristic.

The predictive component improves automatically as your company collects more hiring outcomes.

Final score formula

score = (hard_match_passes ? 1 : 0)
        × (0.40 × hard_score + 0.40 × personality_score + 0.20 × predictive_score)

Where hard_score is 1.0 when all hard filters pass.

Confidence and transparency

Every score comes with a convergence confidence indicator. If the candidate’s MALA profile has low coverage (e.g. they only answered 30% of assessment questions), the personality component carries a confidence penalty. Recruiters see this as a “Incomplete Profile” badge rather than a falsely precise score.

A defensible 70 is more valuable than an unaccountable 90.