Skip to main content

pay-bands-market-data

Use when setting or revising pay bands — covers band structure, market data sources, compa-ratios, and band integrity.

インストールへ移動

ソース情報

リポジトリ
ccashwell/agentic-hr
ソースの最終更新活動
2026年4月25日 23:49
検出された SKILL.md の言語
英語
スター
0
フォーク
0

インストール方法

デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。

ソースファイルを確認

インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。

SKILL.md を表示中

SKILL.md
ソースの指示 · 読み取り専用プレビュー
name
pay-bands-market-data
description
Use when setting or revising pay bands — covers band structure, market data sources, compa-ratios, and band integrity.
# Pay Bands and Market Data ## What a Pay Band Is A pay band defines the cash compensation range for a (level × job family × geography) combination. Bands have: - **Min**: minimum acceptable pay for the level - **Mid**: target pay; anchored to market percentile - **Max**: ceiling; out-of-band pay requires exception Width: typically 15–30% spread (max is 15–30% above min). Wider widths give flexibility; narrower widths force promotion to grow comp. ## Compa-Ratio Compa-ratio = actual pay ÷ midpoint. Used for diagnostics: - **>1.10**: above midpoint; appropriate for proven-at-level - **0.90–1.10**: in-band; expected range - **<0.90**: below band; investigate Aggregate compa-ratios useful for: - Spotting under-paid groups - Demographic gaps - Manager patterns (always low / always high) ## Market Data Sources | Source | Strengths | Weaknesses | |--------|-----------|------------| | **Pave** | Strong tech coverage; modern UX | Submission base skews tech / startups | | **Comprehensive** | Broad sector coverage | Setup heavier | | **Radford** | Tech enterprise standard; level-mapped | Expensive; dated participant model | | **Option Impact** (Compensia) | Equity benchmarks at early stage | Equity-focused | | **Salary.com / Glassdoor** | Free / low-cost | Submission bias; unverified | | **Levels.fyi** | Public, level-mapped tech data | Submission bias toward high-comp candidates | | **Survey participation** | Improves your data quality | Time investment | When choosing: match the source's submission base to your hiring market. A SaaS startup's data isn't reliable for hardware manufacturing. ## Setting Bands 1. Pick percentile target per level (per `compensation-philosophy`) 2. Pull market data for each (level × function × geo) cell 3. Apply percentile to get midpoint 4. Set min and max based on width policy (e.g., ±15% from mid) 5. Sanity-check overlaps between adjacent levels (some overlap is fine; large overlap means levels aren't differentiated enough) 6. Geo zones applied (Tier 1 SF/NY/Seattle; Tier 2 major US; Tier 3 lower-cost; international per local market) 7. Refresh annually; off-cycle if market moves materially ## Out-of-Band Pay Any actual pay outside the band is an exception. Reasons that justify: - New hire bridging large unvested equity from prior employer - Acquisition / acqui-hire compensation preservation - Counter-retention for genuinely exceptional, hard-to-replace talent - Geography pay preserved post-relocation policy change Reasons that don't: - "We had to to close them" (band is wrong, or this isn't the right candidate at this level) - Founder verbal commitment - Manager pressure - "We've always paid Y this much" Document every exception. Audit quarterly. ## Band Integrity Audit Quarterly: - Distribution of comp within bands by job family / level / geo - Out-of-band exceptions - Compa-ratio by demographic - Manager-by-manager comp distribution - New-hire offers vs. existing employees at same level (compression risk) Compression: when new hires get paid as much or more than longer-tenured existing employees at the same level. Common during hot markets; corrodes trust. Address via market refresh of existing employees, not by underpaying new hires. ## Band Communication - **Internal bands**: consider sharing with employees by their own level - **Job posting ranges**: legally required in many US jurisdictions - **Range on offer**: standard practice - **Band rationale**: how were these set? ## Common Failures - No bands; ad-hoc comp - Bands set once, never refreshed - Bands routinely violated (bands are ornamental) - Compression unaddressed - Manager sets pay without comp partner review - Geo strategy invented per case - Single data source over-relied on - Bands too wide (no upgrade pressure) or too narrow (forces unnecessary promotion) ## Cross-References - `compensation-strategist` agent - `compensation-philosophy` skill - `pay-equity-analysis` skill - `offer-strategy` skill - `career-leveling` skill ## Key References - Bock, L. (2015). *Work Rules!* - Practitioner work from Pave, Comprehensive, Radford - Public references: Buffer's published bands; Holloway Guide
GitHubで見る