Compensation Data & Salary Benchmark Methodology
A transparent breakdown of how JobCollate aggregates, sanitizes, and validates compensation data across global technology roles and local metropolitan labor markets.
1. Primary Data Sources
JobCollate calculates real-time salary benchmarks using a multi-tiered data collection protocol. Unlike platforms that rely exclusively on unverified, self-reported user surveys, our methodology prioritizes legally binding, employer-disclosed compensation figures:
- Mandatory Pay Transparency Filings: Roles published in jurisdictions with statutory pay transparency mandates (including California SB 1162, New York City Local Law 59, Colorado EPEWA, and EU Pay Transparency Directive).
- Verified Direct Job Feeds: Explicit salary ranges published directly by employers on certified ATS portals (Workday, Greenhouse, Lever).
- Government Labor Statistics: Cross-referenced with quarterly macroeconomic wage surveys from the US Bureau of Labor Statistics (BLS OEWS), Eurostat Labor Cost Surveys, and the UK Office for National Statistics (ONS).
2. Data Cleaning & Statistical Hygiene
Raw job postings frequently contain formatting noise, hourly vs. annual discrepancies, or extreme outliers. All ingested salary records undergo automated four-step statistical sanitation:
Hourly, weekly, and monthly rates are converted to standard full-time annual base equivalents (assuming a 2,080-hour work year), unless explicitly designated as part-time or contractor arrangements.
We apply interquartile range (IQR) bounding (1.5 × IQR below Q1 and above Q3) to discard placeholder values (e.g. $1–$1,000,000 ranges intended to circumvent transparency requirements).
International salaries in EUR, GBP, CAD, CHF, and TRY are benchmarked in both original domestic currency and standardized USD purchasing-power parities (PPP).
Non-standard employer titles (e.g., "Software Wizard", "MTS 2", "Code Ninja") are mapped via canonical taxonomy matching to standardized industry roles (Senior Software Engineer, DevOps, etc.).
3. Sample Threshold & Confidence Scoring
To prevent misleading statistical extrapolation on thin samples, JobCollate enforces strict statistical confidence thresholds:
- Minimum Sample Rule: Any role-location pair displaying specific salary distribution percentiles must have at least 15 verified postings within the trailing 90-day window.
- Dynamic Fallback: If a localized city sample is below threshold, our system falls back to national metropolitan medians with explicit labeling (*"Based on national median data"*).
4. Total Compensation vs. Base Salary Disclosures
JobCollate explicitly distinguishes between Base Salary (guaranteed cash compensation) and Total Direct Compensation (TDC) (incorporating equity grants, signing bonuses, and performance incentives). Unless specified as TDC, all headline ranges on JobCollate programmatic pages reflect annual base pay.
5. Questions & Feedback
Researchers, academic institutions, and employers seeking to audit or license JobCollate anonymized labor market dataset can reach our research desk at: