The headline numbers
Computed live on every deploy. The loader behind each one is named so you can reproduce it.
Aggregated from public records across 51 jurisdictions — 37,527 multi-source verified, 248,011 license + phone-state matched, 164,839 single-source. These are records, not vetted hireable businesses. totalGoldTierProsNational().
Across 25 states with an ingested official-board roster. totalVerifiedActive().
Every state + DC: the official authority to verify a license yourself, with model. STATE_LICENSE_LOOKUPS.
From 3,411,397 real public building-permit observations across 30 states. indexedMetros().length.
Dated, quarterly-refreshed research pages, each computed from real data. RESEARCH_ARTICLES.length.
MCP tools on /api/mcp plus public machine-readable feeds. MCP_TOOL_COUNT + PUBLIC_API_FEED_PATHS.length.
What these numbers do — and don't — cover
- These are contractor records, not vetted hireable businesses. The 450,377 figure is records aggregated from public sources — about 86% are individual license-holders, with some government entities. By confidence tier: 37,527 are multi-source verified (full NAP + license + corroborating sources), 248,011 are single-source listings that also passed license + phone-state matching, and 164,839 are single-source. Only the first tier clears the "multi-source verified" bar — so we never label the whole set that way.
- Verified-active is roster-scoped. We confirm a license is active against an official board only in the 25 states whose public roster we've ingested. For every one of the 51 jurisdictions we publish the official board-lookup so you can verify a license yourself — but a board appearing in that directory does not mean we've verified a contractor there.
- Matched ≠ active. 423,346 license records are matched in total; that count includes expired and suspended licenses. Only 382,870 normalize to active under the board's own status language.
- The Real Cost Index is a coverage sample. It spans 30 states / 50 metros where we have enough public permits to publish a benchmark — not the whole country — and it is a declared-value median, not a pure price index.
- The permit-activity layer is a matched sample, not a census. 5,004 permits across 554 pros in 16 states have been pulled and pro-matched. It is not a contractor-density or quality map.
- We started in Toledo, Ohio. Our deepest city, county, and neighborhood coverage is in the Ohio footprint where the directory began; national coverage is broad but thinner.
- No invented trust signals. We publish no aggregate star rating, no resold leads, and no fabricated "response time" stats — including on this page.
Every number, with its source
The same figures the open feed serves. Each row names the live loader that produced it.
| Metric | Value | Computed by |
|---|---|---|
| Contractor records aggregated from public sources | 450,377 | totalGoldTierProsNational() — sums gold-tier-counts.json over SUPPORTED_STATES |
| Jurisdictions served | 51 | supportedJurisdictionCount() — SUPPORTED_STATES.length |
| Contractor licenses confirmed active (board) | 382,870 | totalVerifiedActive() — live state-license overlay stats |
| States with a verified-active roster | 25 | verifiedStateCount() — states with a wired verified-active roster |
| License records matched (incl. expired/suspended) | 423,346 | totalVerifiedMatched() — incl. expired/suspended (superset of active) |
| Official board-lookup jurisdictions (50 states + DC) | 51 | STATE_LICENSE_LOOKUPS.length — official board-lookup directory (50 states + DC) |
| Real Cost Index metros | 50 | indexedMetros().length — metros with a published benchmark (NOT meta.metros) |
| Real Cost Index states | 30 | costCoverageStateCount() — states with a published cost benchmark |
| Public building-permit observations indexed | 3,411,397 | loadPermitCostIndex().meta.usableObservations — real public permit valuations |
| Cost benchmarks (trade × place) | 432 | loadPermitCostIndex().meta.benchmarks — distinct (trade x place) median+P25/P75 |
| Permits in the pro-matched sample | 5,004 | loadPermitCoverage().totalPermits — real (non-synthetic) pro-matched permits |
| Pros in the permit-matched sample | 554 | loadPermitCoverage().totalPros — distinct pros with >= 1 matched permit |
| States in the permit-matched sample | 16 | loadPermitCoverage().stateCount — states in the matched-permit sample |
| Original computed studies | 31 | RESEARCH_ARTICLES.length — original computed studies |
| Agent-callable MCP tools | 46 | MCP_TOOL_COUNT — agent-callable tools registered on the /api/mcp server |
| Public machine-readable feeds | 68 | PUBLIC_API_FEED_PATHS.length — public machine-readable feed paths |
Why we publish our own numbers
Most directories ask you to trust a marketing claim. We'd rather hand you the figures and the method. A number is only useful if it's reproducible, so each stat here is backed by a named loader over committed public-record data, and the identical numbers are available as a machine-readable feed for AI engines, journalists, and partners to cite directly.
That's also the honest answer to the AI-citation question: an assistant can't see a contractor's Google reviews, but it can read a board-verified-active license count, a permit-derived cost median, and a feed that says exactly what it does and doesn't cover. Extractable, dated, and honest beats inflated every time.
Get the data
The open feed serves the same numbers as this page, computed by the same loaders. Free to reuse under CC BY 4.0 with attribution.
Spot a number that looks off?
We'd rather you tell us than trust the directory less. Email press@profixdirectory.com.