メニュー
Use when designing, auditing, or migrating a performance management system — covers cycles, ratings, separation of conversations, and stage fit.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Use when designing or applying progressive discipline — covers the escalating sequence, documentation, gross misconduct exceptions, and consistency.
Use when conducting or scoping workplace investigations — covers intake, scope, interview protocols, evidence, findings, and outcomes.
Use when auditing or designing the candidate's journey — covers communication, time, transparency, and the link between candidate experience and employer brand.
Use when designing career frameworks, level rubrics, IC and management track parity, and promotion processes.
Use when planning organizational change — covers communication architecture, sequencing, manager enablement, and predictable failure modes.
Use when training managers in coaching skills — covers GROW, listening, asking vs. telling, and growth mindset framing.
| name | performance-management-systems |
| description | Use when designing, auditing, or migrating a performance management system — covers cycles, ratings, separation of conversations, and stage fit. |
Bock's central insight from Work Rules!: performance, compensation, and development are three different conversations. Conflating them forces comp to crowd out development. Separate them in calendar and in process.
| Headcount | Cycle | Format |
|---|---|---|
| <30 | Ongoing 1:1s; no formal | Quarterly check-ins |
| 30–80 | Annual lightweight | Single rating; informal calibration |
| 80–250 | Annual or semiannual | Calibrated; structured ratings |
| 250–1000 | Annual or semiannual | Calibrated panels; multi-input |
| 1000+ | Semiannual | Full enterprise: bench reviews, succession |
Don't impose annual reviews at 30 people. Don't leave 500-person companies with 1:1 vibes only.
4-point (recommended): Below / Meets / Strongly Meets / Exceeds. Forces meaningful discrimination.
5-point with neutral middle: most ratings land at 3; information-poor.
9-box (performance × potential): avoid for performance ratings; conflates two different things; "low potential" labels stick.
Forced distribution: high cost, mixed evidence; can corrode collaboration. Default: don't force, but watch for severe skew (90% "exceeds" suggests inflation).
Calibration is not "making everyone agree." It's ensuring consistency in how rubrics are applied across managers.
Performance is one input to comp; not the only input.
Avoid lockstep formulas (rating × multiplier):
Do:
Honest only if there's a real path to success.
Honest PIP:
Theatrical PIP (avoid):
Theatrical PIPs corrode trust company-wide.
When fixing a broken system:
Don't redesign everything at once. Sequence change.
career-leveling skill — Level rubrics anchor ratingsperformance-calibration skill — Calibration mechanicsfeedback-frameworks skill — Ongoing feedbackcompensation-philosophy skill — Comp linkage