Finding 01
The Wage-Level Shift is real, but running the wrong way.
The wage-weighted lottery was designed to reward higher-paid positions. The data shows the opposite direction — and the rise is entirely driven by new, unknown-tier employers, not by established sponsors changing behavior.
+41% cross-industry Level I share (15.0% → 21.0%). Established large-tier sponsors actually reduced their Level I share (6.9% → 6.2%).
Finding 02
Filing discipline moved through a stress cycle, not a breakdown.
The Certified-Withdrawn rate spiked during FY2023 lottery uncertainty, then declined below baseline once the framework clarified. Adaptive behavior, not compliance failure.
CW rate: 5.64% → 6.50% → 5.17% (FY22 → FY23 peak → FY25). Employers used CW LCAs as a hedging instrument, then abandoned the tactic.
Finding 03
The Template Problem is measurable and concentrated.
The Index introduces a new public metric — title_top10_share — measuring the percentage of a cohort's filings that use one of the 10 most common job title strings. Templated filing at scale is the behavioral signature USCIS adjudicators are increasingly equipped to challenge.
33.4% of large-tier Professional & Technical Services FY2025 filings use one of just 10 job titles. "SOFTWARE ENGINEER" alone accounts for nearly 10%.
Finding 04
Extensions are no longer routine, and cost is concentrated in mid-market sponsors.
Continuing denial rates rose materially across every top sector between FY2022 and FY2023. Small-tier sponsors face 2–3x the denial rate of large sponsors in every sector. Mid-market employers who assume scale protects them are wrong.
Information (Tech) denials +84%. Manufacturing +67%. Finance +62%. Small-to-Large denial ratio peaks at 3.22x in Information.