Data sources.
Every figure in the Orbyt Intelligence Annual Compensation Report is traceable to one of five source buckets. No proprietary survey data, no anonymous tips, no scraped-and-unverified third-party data.
- BLS Occupational Employment and Wage Statistics (OES). The canonical U.S. government source for role-level compensation medians, released annually with annual addenda. bls.gov/oes
- H-1B Labor Condition Application (LCA) filings. Published disclosures from the U.S. Department of Labor covering every H-1B filing made by every U.S. employer. Carries actual offered compensation by role, metro, and employer. dol.gov/agencies/eta/foreign-labor
- State pay-transparency job postings. Salary ranges employers are required to publish in job postings under state pay-transparency laws (Colorado, California, Washington, New York, et al.), collected from the postings themselves and aggregated by role and metro.
- Community submissions. Structured compensation submissions made through the authenticated Orbyt Intelligence API. Each one is tagged as community-sourced and disclosed as its own source in the published source breakdown.
- BEA Regional Price Parity (RPP). Annual Bureau of Economic Analysis release of cost-of-living multipliers per U.S. metropolitan statistical area. Used for the cost-adjusted overlay. bea.gov/data/prices-inflation/regional-price-parities
Six-step pipeline.
- Aggregate. Ingest postings and disclosures from the five sources into a single normalized schema: role title, base compensation midpoint, metro, posting quarter, source.
- Normalize role titles. Collapse title variants (“Senior Software Engineer”, “Sr. SWE”, “Software Engineer II”) onto a canonical role slug. The 2026 taxonomy contains 3,445 canonical roles across 81 U.S. metros.
- Compute base median per role and metro. Take the median base compensation within each role-metro cell, discarding cells with fewer than five postings in the quarter. Round the result to the nearest $1,000 to reduce false precision in the published figures.
- Cost-adjust. Divide nominal median by BEA Regional Price Parity for the metro. Cost-adjusted median lets a reader compare San Francisco dollars with Charlotte dollars on equal footing.
- Compute deltas. Express year-over-year and year-over-year movement as a signed percentage. Cells with fewer than ten postings on either side of the window are suppressed.
- Narrative and auditor gauntlet. Run the compiled dataset through an AI authorship pass with a named-analyst voice contract, followed by automated style, citation, and thin-content auditors. Auditor blockers gate publish. Unverified citations are rejected. Paragraphs under a minimum signal threshold are flagged.
Cost adjustment formula.
cost_adjusted_median = nominal_median / (RPP / 100)
Where RPP is the Bureau of Economic Analysis Regional Price Parity multiplier for the metropolitan statistical area, indexed so that 100 equals the U.S. national average. A metro with RPP 129 (San Francisco in recent BEA releases) divides nominal compensation by 1.29, yielding the purchasing-power equivalent in national-average dollars.
Voice contract.
The narrative passes of every Annual Compensation Report are written under a style contract that excludes banned corporate-speak constructions (leverage, unlock, harness, not just X but Y, empowers, solutions-oriented, synergies, and variants). Every claim in the narrative must cite either a dataset path (e.g. macro.topPayingRoles[0].median), a named external authority from the verified citation roster, or operator judgment marked as such. Unverified citations are rejected by the style auditor. The voice is first-person, analyst-framed, and positions named authorship as a credibility anchor.
What the data is not.
- Not a proprietary survey. Public records (BLS OES, H-1B LCA filings, state pay-transparency postings) anchor the aggregates. Community submissions are disclosed as their own source, never blended silently.
- Not a forecast of individual compensation. Medians are population-level snapshots. Individual compensation varies by equity structure, company stage, negotiation leverage, and factors outside the dataset.
- Not a recruitment or hiring tool. The report is a research product. Orbyt Jobs (the CRM) is a separate product for managing job-search workflow; the Orbyt Intelligence API is a separate product for programmatic data access.
- Not peer-reviewed. The auditor gauntlet enforces style and citation rigor but is not a substitute for academic peer review.
Licensing and citation.
Every artifact of the Annual Compensation Report (Free Summary PDF, both Enterprise PDFs, composite dataset CSV) is licensed under the Creative Commons Attribution 4.0 International license (CC BY 4.0). You may quote, redistribute, and derivative-work any part of the report with attribution to Justin Bartak / Orbyt Intelligence and a link back to the canonical report URL.