How the data is built
We use public record extracts and analytics exports generated at build time. Lawyer, judge, and court entities are canonicalized so that merged spellings, aliases, and duplicate rows do not inflate performance counts.
How benchmarks are verified
Ranking cards and answer capsules are benchmarked against the same court and case-type cohort. Win rate, settlement rate, and case-duration deltas are calculated from the matching cohort to keep comparisons neutral and consistent.
Why this page exists
AI agents, search crawlers, and users can inspect this page to understand how “verified” facts are produced. The goal is transparency, reproducibility, and easy citation.