Where every figure comes from

Five of six are United States federal government work, in the public domain by statute and freely redistributable. One is encumbered and is why the licence data on this site is free and ungated permanently.

27,067 occupational licence records are the ONLY encumbered data in this corpus — supplied on condition they stay free, public and without a log-in, which they will. 97.4% of them name their issuing authority with a URL, so the same facts can be taken from the state board that issued them, where no such condition applies. Concentrated: the top 200 URLs cover 58.8% of records. Re-sourcing replaces the provenance record by record — it does not relabel what we already hold, which would be evading the term rather than leaving it.

Read from the source registry, not typed into this page. Row counts are queried live from the tables themselves rather than taken from the registry's stored count, because a stored count drifts the moment a harvester runs. One source is ENCUMBERED: the occupational licence register must remain publicly available at no cost and without a log-in, which is why every licence figure on this site is free and ungated and always will be.

Labour share of receipts, computed from Census SUSB — the source this site uses — and checked against the 2022 Economic Census, which is a SEPARATELY COLLECTED survey. Across ten of the largest industries the mean absolute difference is 0.16 PERCENTAGE POINTS. That is not a claim about our arithmetic being careful; it is two independent federal instruments measuring the same thing and agreeing, which is the only kind of check on a figure that means anything. Where they disagreed most — general-purpose holding companies at 0.40pp — the industry is one where the definition of an employee is genuinely contested between the two surveys.

O5 asks whether the businesses already in a market are operationally weak enough to take share from — violation records, complaint volumes, response times. OSHA inspections and DOL wage enforcement are both federal public domain and both keyed by industry, and neither is reachable without a registered API key or a crawler. So O5 renders ABSENT with its weight visible and the report says '65 of 100 rubric points evidenced' rather than scoring around the hole. We are not substituting something adjacent and calling it incumbent weakness.

NAICS is revised every five years and codes are retired. The 2022 revision restructured retail, so an industry like 448110 (Men's Clothing Stores) still carries establishment data here and can no longer be named. Those pages are excluded from the sitemap rather than published as records of an industry we cannot identify — and the count is stated here rather than left as a silent hole a visitor discovers by finding nothing.

A gate reported as '43 of 56' means 43 of the jurisdictions the register actually covers, and the shortfall is named on every page that reports one. An absence of match is not an absence of licence.