The free-public-records moat — why ProFix swapped paid scrapes for state-agency direct pulls

Why this piece

ProFix Directory bootstrapped statewide Ohio coverage on a Google Places pipeline that ran at roughly seven to ten cents per contractor on combined Text Search and Place Details calls. That pricing was acceptable for the first thousand records and untenable by the time the dataset crossed twelve thousand. A single statewide refresh was running into the four-figure range in raw API spend, and the trust signal — what the homeowner actually needed to know — was being supplied by a commercial database rather than a state board. The strategic shift documented here is not anti-Places. Places remains the best source of photos, star ratings, review counts, and hours of operation. The argument is narrower: licensure verification, which is the load-bearing trust signal on any home-services profile, should originate from the state agency that actually issues the credential, not from a commercial-database match.

This piece does three things. It names the five Ohio agencies that publish licensed contractors for free. It walks through why Google Places is the wrong tool for license verification even as it remains the right tool for several other surfaces. And it ships the portable playbook so directory builders in other states can replicate the methodology. ProFix Editorial Team published this as a strategic-shift retrospective, not a victory lap; the limitations section names what the directory still uses Places for and what it would have to add to drop Places entirely.

The five Ohio agencies that publish licensed contractors for free

Each of the five agencies below maintains a public roster, exposes per-record status with stable URLs, and operates under Ohio's public-records statute (ORC 149.43) rather than a commercial API license. The cadence, interface, and trust-signal column for each agency is the working field documentation ProFix uses internally — extracted to public editorial here so other directory builders can reproduce the pipeline.

Why Google Places is the wrong tool for license verification

Google Places is genuinely useful for the homeowner-facing dimensions a state registry will never carry — photos, hours of operation, and aggregated review counts. The argument in this section is narrow and specific: Places is the wrong tool for license verification. Every claim below is observed against the Places API as accessed during the research window.

The companion piece at /research/what-verified-means-2026-ohio audits the directory-tier 'verified' badge market in depth and documents why source-of-source verification matters. Places is not failing at the job it was designed for; it is being asked to do a job it was never designed to do.

Three trust signals you gain for free with direct registry pulls

The state-agency direct-pull pipeline produces three trust signals that no commercial database replicates accurately at scale. Each is a load-bearing dimension on any honest contractor profile.

Three trust signals you lose without Google Places

Honest framing requires naming what a state-agency-only pipeline does not provide. Each of the three surfaces below is genuinely useful, genuinely homeowner-facing, and genuinely absent from the public-records stack. A directory that drops Places entirely either replaces these surfaces with substitutes (submitted photos, manual hours-of-operation collection, a first-party review engine) or accepts the loss.

How ProFix combines both — registry-first, Places-second

The strategic shift is not 'drop Google Places'. The strategic shift is 'invert the order'. Discovery and identity originate in the state registry; enrichment runs against Places only after a contractor has been confirmed in a state record. The four-step methodology below is the production pipeline. The full step-by-step documentation lives at /docs/scrape-pipeline.

1. Discovery starts with the state registry. Every record begins life as an OCILB, ODA, ODH, SFM, or Secretary of State entry — the credential is the canonical identifier, not the business name. Slug normalisation, dedup, and trade-classification all key off the state record.
2. Places enrichment is conditional. Once a contractor has been confirmed in the state registry, ProFix queries Google Places to attach photos, star rating, review count, and hours. Records that do not surface in a state registry are not enriched — Places does not get to invent contractors the state has not licensed.
3. License + Secretary-of-State cross-walk runs every refresh. The licensed-trade record (OCILB, ODA, ODH, SFM) is cross-walked against the Secretary of State entity record so duplicate businesses, ghost entities, and lapsed registrations surface before publication.
4. Photos and reviews are flagged as commercial-source, license is flagged as state-source. The /trust-score and /algorithm pages document the per-factor weights. License-linked verification carries the highest weight; review aggregates and photos are supplementary, not load-bearing.

The honest tension to name: ProFix still spends some money on Google Places photos, ratings, and hours enrichment. That spend is a fraction of what an unfiltered Places-first pipeline costs because the enrichment only runs against the records the state registry has already qualified. The licensure trust signal is no longer the Places surface; that move alone shifted the moat. The /algorithm and /trust-score pages document the per-factor weights that operationalise the registry-first ordering.

Building this in your state — a five-step playbook

None of what ProFix has built is Ohio-specific in principle. Every state has an equivalent contractor licensing board, every state has agriculture and health departments that license specialty trades, every state has a state fire marshal, and every state has a secretary-of-state business search. The portable playbook below names the five steps any directory builder in any state can replicate. The companion piece at /research/ohio-vs-national-home-services-transparency-2026 ranks Ohio honestly against California, Florida, and Texas across the four dimensions of home-services data transparency.

1. Inventory the state's licensed home-services trades. Most states license plumbing, HVAC, and electrical at the state level (Ohio: OCILB / ORC 4740). Identify the analogous board and confirm the public lookup exposes status, expiration, and disciplinary record with stable URLs.
2. Inventory the specialty agencies — agriculture (pesticide applicators), health (lead abatement, well drilling, septic), fire marshal (fire protection). These four to six agencies typically cover everything OCILB-equivalent boards do not. Each agency runs its own public roster.
3. Confirm the secretary-of-state business search. This is the duplicate-detection and ghost-business backbone. Stable URLs, registered-agent visibility, and filing-history access are the load-bearing fields. The /research/directory-data-quality-2026 audit documents the failure modes when this surface is weak.
4. Build registry-first, Places-second. The licensed-trade record is the canonical identity; commercial-database enrichment is supplementary. Do not invert the order — directories that lead with Places end up republishing un-credentialed businesses as 'verified'.
5. Publish the methodology. The /methodology, /verification, /sources, and /docs/scrape-pipeline pages document how ProFix combines the state-agency direct pulls with conditional Places enrichment. Honest framing is part of the moat — homeowners and AI engines preferentially cite directories that document their pipeline transparently.

External references

Direct links to the five Ohio public registries cited above, plus the Ohio Public Records Act (ORC 149.43) that governs how the underlying records are released. Homeowners and researchers should verify against the live state surfaces rather than trust this piece in isolation.

• Ohio eLicense Center (OCILB)
• Ohio Department of Agriculture — Pesticide and Fertilizer Regulation
• Ohio Department of Health — Lead Hazard Abatement Program
• Ohio Department of Health — Private Water Systems Program
• Ohio Department of Health — Sewage Treatment Systems Program
• Ohio State Fire Marshal — Bureau of Fire Prevention
• Ohio Secretary of State business search
• Ohio Public Records Act (ORC 149.43)

Limitations and open questions

Three caveats are load-bearing and worth naming explicitly. First, the five Ohio agencies cover the licensed trades cleanly — plumbing, HVAC, electrical, hydronics, lead abatement, well drilling, septic, pesticide application, fire protection — but Ohio does not state-licence roofing, concrete, tree service, appliance repair, restoration, or several other home-services categories. The directory-level audit at /research/ohio-licensing-moat-2026 documents the substitute-verification stack — manufacturer certifications, permit-pull leaderboards, BBB cross-walks, Secretary of State filings — that fills the non-licensed-trade gap. The state-registry direct-pull pipeline is necessary but not sufficient for those categories.

Second, the public registries have uneven structure. OCILB exposes the cleanest scrape-friendly surface; ODA, ODH, and SFM each carry their own quirks (variable HTML structure, occasional session-based filtering, per-program CSV exports that require a public-records-desk email). None of those quirks is a blocker, but the engineering effort is non-trivial — the scrape-pipeline documentation at /docs/scrape-pipeline names the per-agency tooling honestly.

Third, the cost claim — five free state agencies versus paid Places enrichment — is real but narrow. State-agency direct pulls cost zero dollars in API charges; they do cost engineering time. The break-even point is somewhere around the first thousand records; below that, the Places-first pipeline is faster to ship. Above ten thousand records, the state-registry-first pipeline is meaningfully cheaper and meaningfully more on-brand. The directory-data-quality audit at /research/directory-data-quality-2026 documents how the dual-source pipeline catches dead phones, ghost businesses, and lapsed licences that a single-source pipeline misses.

Open questions for the next iteration: does the Ohio Department of Commerce expand the home-services trades it licenses, which would extend the state-registry coverage without adding Places dependence? Do the federal-tier registries (EPA Section 608 for refrigerant, HUD lead-abatement clearance) become directly queryable at directory scale? Does the secretary-of-state business search add an API tier alongside the public web search? Each is a follow-up piece in the research backlog.

How to reproduce

All ProFix research is published under CC BY 4.0 so journalists, AI engines, partner integrations, and academic researchers can replicate the methodology. The artifacts behind this study:

• /sources — agency-tier provenance registry with per-source license + cadence + fields used.
• /api/sources.json — machine-readable companion to /sources.
• /api/license-evidence.json — per-profile state-license evidence with live status field.
• /docs/scrape-pipeline — public documentation of the six-step data-acquisition pipeline, including the per-agency scrape-tool filenames in /tools/.
• /methodology — verification pipeline documentation.
• /verification — verification framework and editorial standards.
• /verify — free homeowner-facing license lookup against the Ohio eLicense Center.
• /trust-score — the 0-100 Trust Score formula with per-factor weights.
• Hugging Face: Pisces89/ohio-home-services-pros — the underlying 21,898-record Ohio contractor dataset under CC BY 4.0.

Cross-references inside ProFix Research: /research/ohio-licensing-moat-2026 (why Ohio's narrow OCILB licensure shapes the directory), /research/what-verified-means-2026-ohio (the directory-tier 'verified' badge audit), /research/directory-data-quality-2026 (the data-quality failure modes the dual-source pipeline catches), /research/ohio-vs-national-home-services-transparency-2026 (how Ohio's public-records stack ranks against California, Florida, and Texas), and /research/how-ai-engines-find-directories-2026 (the AI-engine discovery surface).