Fonteum Methodology

01 · Data sourcing

Where the data comes from.

Every published figure starts from a public, reproducible federal source. Primary inputs: CMS NPPES, CMS PECOS, CMS Care Compare, HRSA HPSA, BLS OEWS, BEA Regional, U.S. Census Bureau population estimates — the canonical authorities for U.S. healthcare provider identity, Medicare facility quality, shortage-area exposure, and healthcare workforce + economic context.

Each source is registered in /sources with refresh cadence, fields used, limitations, and an explicit display ceiling.

02 · Coverage scope

What we include, and what we don't.

We track businesses with a physical US presence across the network's category groups. Inclusion test: real verifiable address, publicly accessible business record, ≥1 independent review source.

Exclusions: online-only operators, lead-gen shells, businesses without a reviewable public footprint. Coverage gaps documented per study.

03 · Update cadence

How often we refresh.

Listings refresh on a rolling 30-day cadence; ratings/reviews refresh weekly for the highly-rated segment. Research snapshots are dated and versioned per study.

04 · Credential checks

What "highly rated" means on Fonteum.

For healthcare provider records normalized across federal CMS sources and healthcare workforce data sources, the credential status we surface is what those public sources themselves publish — facility-level records (CMS NPPES, CMS PECOS), quality-and-outcomes records (CMS Care Compare modules), workforce records (HRSA HPSA, BLS OEWS). We do not assert credentials we cannot trace to a federal public source.

We do not claim independent vetting. We claim publication of what independent federal sources have measured.

05 · Corrections policy

How to report an error.

Send to corrections@fonteum.com or use /contact. We acknowledge within two business days and publish the resolution to /corrections-log. Substantive research corrections update the affected article and append a dated note.

06 · Editorial standards

No pay-to-rank.

Rankings are derived from public ratings and credentials; not influenced by commercial relationships. Featured slots are clearly labelled. Research content is never sponsored.

07 · Research methodology

Sample sizes, methods, disclosed limits.

Each study declares sample size, inclusion criteria, temporal coverage, and methods. Datasets are published as downloadable CSVs at /research/data/<slug>.csv. Where studies use correlations or modeled estimates, the methods name the estimator, assumptions, and uncertainty.

08 · Source-pack ingestion methodology

How a public dataset becomes a wired source.

Five-stage pipeline: (1) versioned manifest (authority + license + field list, reviewed against §125 restricted-source list); (2) bounded dry-run pilot + match-rate validation; (3) confidence calibration + low-confidence exclusion; (4) registry write to src/lib/brand/sources-registry.ts; (5) public surfacing via SourceChip + ProvenanceCard.

09 · Entity matching

Connecting profiles to source-pack records.

Deterministic match on stable identifier (NPI / state license / CMS Certification Number) with probabilistic fallback on (name, address, taxonomy). §95 confidence tier assigned at match time. Low-confidence matches stay internal.

10 · Provenance field architecture

The four-piece contract.

Every public field carries source name + last-checked ISO date + confidence tier + display permission. Aligned with W3C PROV-DM (Entity / Activity / Agent decomposition) and the FAIR data principles (Findable · Accessible · Interoperable · Reusable). Wire shape at /data-platform/schema; UI shape at /data-provenance.

11 · Confidence scoring

Tiering today + on the roadmap.

Tier high requires stable-identifier match plus address agreement. Tier medium requires multi-field agreement without an identifier. Tier low covers residual ambiguity and is not displayed publicly. Threshold is deterministic + manually reviewed today; learned-model approach is on the roadmap once labeled gold sets are in place.

12 · Historical change detection

Versioned at the field level.

We version the value, not the surrounding metadata. Change record = (provider_id, field, previous_value, new_value, source, observed_at, snapshot_id). Cadence per source family: CMS NPPES daily, CMS PECOS weekly, HRSA HPSA monthly, CMS Care Compare quarterly, BLS OEWS + BEA Regional annually.

13 · Reproducibility

How to re-run any published study.

Every research study ships its CSV at /research/data/<slug>.csv with a methodology block declaring sample size + inclusion criteria + temporal coverage. Federal-source studies have aggregation scripts in scripts/research/. Indexed-dataset studies document group-by aggregations in the technical appendix.

14 · Citation guidance

How to cite Fonteum Research.

Each study page carries a copy-paste citation block at /research/<slug>#cite. Recommended form: Fonteum Research, "<Study Title>," <Month YYYY>. https://fonteum.com/research/<slug>. Dataset: https://fonteum.com/research/data/<slug>.csv. AP-style short and BibTeX variants available on each study page.

15 · What we do not claim

Disclaimers in writing.

Counts describe the Fonteum indexed provider dataset (26,417 healthcare-confirmed providers across 6 federal public-record sources), not a representative sample of the U.S. healthcare provider data landscape. We do not own / moderate Google reviews. No pay-to-rank. No unsupported v-word adjective in user-facing copy.