Performance Review Synthesizer

What it does

It closes the three chronic problems of a performance conversation in sequence. (1) Rater bias — trusting a single evaluator exposes you to halo/horn effects, central tendency, leniency/severity, recency and similar-to-me bias. The skill gathers feedback from 4 sources (manager, peer, direct report, self) and applies 360° Multi-Rater Aggregation: a theme confirmed by ≥2 independent sources is "confirmed", a single-source theme is a "weak signal", and when sources contradict, both poles are shown — contradiction is information, never averaged away. (2) Abstract feedback — non-actionable statements like "should improve communication" are reduced via the SBI (Situation–Behavior–Impact) model into {in which situation, which observable behavior, what impact}; pure judgments that cannot be turned into behavior ("lazy", "good team player") are not evidence and fall into gaps. This also makes feedback legally defensible (observation ≠ label). (3) Jargon + fairness gap — a bias-corrected, source-weighted Performance Synthesis Score (0-100) is produced (bias-flagged raters drop to weight 0.5), the person is placed on a 9-Box (Performance × Potential) grid, and up to three development areas are converted via GROW + SMART/OKR into a 30/60/90-day plan — no goal is written without a "how will we measure it" metric.

When to use it

During the year-end/quarterly review cycle when you have a messy pile of 360° comments; before a promotion/calibration meeting when you need a clean strength-gap map for a person; when a line manager must justify feedback fairly and traceably (decisions with 4857/KVKK exposure); or when establishing a measurable baseline for a coaching/development plan. Not for turning a single score into a promotion/termination rationale — the score only gains meaning with the 9-box position, confidence level, and an evidence chain.

Method / frameworks

• 360° Multi-Rater Aggregation — dilute single-rater bias with cross-source agreement; ≥2 independent sources = confirmed.
• SBI Model (Center for Creative Leadership) — reduce judgment to observable behavior.
• Rater Bias Taxonomy — halo/horn, central tendency, leniency/severity, recency, similar-to-me; flag distribution anomalies (Brown, Inceoglu & Lin 2017 — forced-choice/Thurstonian IRT logic as reference).
• 9-Box Grid (GE/McKinsey → talent management) — two-dimensional Performance × Potential placement.
• GROW (Whitmore) + SMART/OKR — structure the development plan across 30/60/90 days.
• Turkey layer: Labor Law 4857 art. 5 (equal treatment) + objective/written criteria; KVKK art. 4-5-6 + 10-11 (purpose limitation, disclosure, role-not-identity attribution). ISO 30414 human-capital reporting reference on request.