Skip to content
FonteumPublic-records evidence

Print-friendly export. To save as PDF: open browser Print → choose “Save as PDF”. The print-CSS suppresses gradients and renders link URLs inline.

Fonteum Methodology

v1.6 · last revised 2026-07-18 · canonical: fonteum.com/methodology

Aligned with W3C PROV-DM + FAIR data principles. Free to cite with attribution.


01 · Data sourcing

Where the data comes from.

Fonteum covers named US federal, state, sanctions, and global public-register sources. Dated platform coverage is published from the platform-stats snapshot; active registry status does not establish a fresh or complete load.

The generated public catalog documents a subset of the active registry. Source pages state acquisition method, publisher cadence, and limitations for the entries they document.

02 · Coverage scope

What we include, and what we don't.

Coverage is limited to the named records actually loaded from participating sources; it is not a complete census of US exclusions, providers, contractors, companies, or enforcement records.

Identifiers are source-specific: NPI where a source publishes it, CCN for supported CMS facilities, UEI/CAGE for federal contractors, and each register's own key elsewhere. An NPI is not a universal identity key and does not establish licensure or credentialing.

03 · Update cadence

Observed loaded dates, not schedule promises.

When checked July 12, 2026, the serving tables showed: NPPES newest system updated_at 2026-06-10; PECOS newest source date 2026-06-18; OIG LEIE 2026-05-08; supported Care Compare tables 2026-05-07; and PBJ work date 2025-06-30.

Publisher cadence and configured jobs do not prove a later load completed. Public copy uses the source or serving-table date actually observed and can report a date as unavailable.

04 · Provenance architecture

A nullable fourteen-field schema.

The provenance schema defines fourteen possible fields, including source, URL, dataset id, snapshot date, methodology, last-checked time, confidence, availability, pipeline version, license, and coverage bounds. Population varies by source, endpoint, record, and field; values can be null.

A direct July 12 count found 26,211,234 provenance rows. 26,211,220 carried a valid-shaped source-file SHA-256 and 14 did not. Deterministic fact-to-attestation and fact-to-witness-signature linkage was 0.

05 · Retained history

History is source-specific.

Closed row versions observed July 12 were UK sanctions 9,442, procurement awards 38, provenance claims 20, OFAC SDN 10, and UN sanctions 1; EU sanctions had 0.

Sources without retained versions cannot be reconstructed to arbitrary past dates. Historical endpoints can return no prior version even for a supported subject.

06 · State Medicaid exclusions

13 state Medicaid jurisdictions represented, not national coverage.

As of August 11, 2026, the platform-stats snapshot reported 22,917 state Medicaid exclusion rows across 13 distinct jurisdictions. Other jurisdictions are not represented as covered.

07 · Nursing-home penalties

Records and facilities use different denominators.

The April 2023–March 2026 table has 16,832 rows across 6,919 facilities and $467,344,898 in positive fines. The median positive fine record is $14,113; median cumulative fines per fined facility are $34,272.

2,553 payment-denial records cover 1,950 distinct facilities. They are records, not 2,553 facilities.

08 · External timestamp proofs

Stored status is not mainnet evidence.

Production held 162 OpenTimestamps proof records. Independent validation against authentic Bitcoin-mainnet headers found 0 valid proofs. Stored operational status and block-height values are therefore not presented as external confirmation.

09 · Research methodology

Methods and downloads are study-specific.

A published study states its named sources, sample definition, temporal coverage, joins, limitations, and reproduction method. Downloads appear only where that study publishes them; no universal CSV path is promised.

Direct counts, derived rates, and modeled values are labeled according to the study's actual method. A study snapshot does not imply the serving table or upstream publisher refreshed later.

10 · Corrections policy

How to report an error.

Send to corrections@fonteum.com or use /contact. We acknowledge within two business days and publish accepted corrections to /corrections-log. Substantive corrections update the affected sentence with a dated marker.

11 · Citation guidance

How to cite Fonteum Research.

Use the citation block on the individual study or dataset page. Recommended form: Fonteum Research, "<Study Title>," <Month YYYY>. https://fonteum.com/research/<slug>. Include the named upstream source and snapshot stated on that page.

12 · What we do not claim

Disclaimers in writing.

Fonteum does not independently attest to provider credentials, establish that a provider is qualified or in good standing, assign a risk score, or treat a zero-match screen as a clearance when coverage is stale or indeterminate.

Counts describe the loaded Fonteum data at the stated observation, not a representative sample of a broader market. [corrected on 2026-07-12]


Methodology changelog

  • 2026-07-18Source authority now follows the issuer named on the record. (Doctrine). Reason: A global source-authority claim assigned pages to a short agency list even when their actual records named different issuers..
  • 2026-07-18Source lifecycle and state jurisdiction are separate query-reproducible measures. (Schema). Reason: The previous live-source field conflated registry, loaded-data, and recency concepts and omitted data-bearing snapshot IDs that lacked a crosswalk..
  • 2026-07-15Public freshness now requires a proved canonical content transition. (Doctrine). Reason: The freshness audit found that cross-namespace rollups and workflow timestamps could make unchanged or unproved serving data appear freshly updated..
  • 2026-07-14Indexed study figures now require cap-safe SQL and fail closed when a derivation is unsupported. (Research rule). Reason: Source re-ingests and date corrections changed production data after several studies were computed, and at least one earlier headline used a PostgREST response capped at 1,000 rows..
  • 2026-07-12Public counts, freshness, history, and cryptographic claims tied to production evidence. (Doctrine). Reason: A production SQL audit found that several public phrases combined publisher cadence, catalog size, file hashes, and snapshot attestations into broader claims that the serving data did not support..
  • 2026-07-12Snapshot commitments move from count-grain hashes to sealed row-level Merkle roots. (Schema). Reason: A count-only preimage could not detect changes to values when the number of rows stayed constant. Preserving and labelling the old evidence is necessary because silently replacing it would be indistinguishable from tampering..
  • 2026-07-12Exclusion screening becomes tri-state and fails closed on stale coverage. (Schema). Reason: The OIG serving table remained on the 2026-05-08 release while signed snapshot 906 covered 2026-07-01, allowing post-May excluded NPI 1689095580 to receive a false negative..
  • 2026-05-03Data Graph Visual v2 — moat as a one-second picture. (UX / surface). Reason: Previously the homepage carried a scroll-narrative pipeline that predated the source-provenanced moat, and /data-provenance's hero schematic was abstract — no source names visible, no clickability. The audit recommended a stronger graph that makes the moat readable in one second..
  • 2026-05-03Sources Library v2 — status field + restricted-sources doctrine. (Doctrine). Reason: The first version of the source library conflated tier (display contract) with deployment status. AZ ROC + CSLB were tagged Tier-2 even though no public-records request had been processed; their reality is 'manifest registered, awaiting CSV'. Surfacing pending and explicitly-not-used sources is a stronger trust signal than implying everything is live..
  • 2026-05-03/press rebuilt as a journalist + data-user landing page. (UX / surface). Reason: .
  • 2026-05-03/data-platform — B2B / data-product surface. (UX / surface). Reason: Journalists, investors, and B2B prospects landing without a vertical-specific intent had no surface that explained what Fonteum is as a data company. /research answers vertical questions; /data-platform answers the meta question..
  • 2026-05-03Brand chart palette + StatTable typography polish. (UX / surface). Reason: .
  • 2026-05-03/directories Coverage Atlas — explicit 4-status taxonomy per cell. (UX / surface). Reason: .
  • 2026-05-03Research study template reskinned to brand tokens. (UX / surface). Reason: .
  • 2026-05-03/directories repositioned as Coverage / Network Map. (UX / surface). Reason: .
  • 2026-05-03Single canonical data-snapshot date across the brand hub. (Schema). Reason: .
  • 2026-05-03Research pages carry an explicit AI-citation summary + Limitations panel. (Research rule). Reason: .
  • 2026-05-03Public source library at /sources. (Doctrine). Reason: .
  • 2026-05-03Profile provenance reveal cards on listing detail pages. (Display rule). Reason: .
  • 2026-05-03/data-provenance upgraded to the public Data Graph page. (UX / surface). Reason: .
  • 2026-05-03Homepage repositioned as the source-provenanced provider graph. (UX / surface). Reason: .
  • 2026-05-03CMS Care Compare research bundle (home health + hospice). (Data snapshot). Reason: .
  • 2026-05-03Dialysis facility research snapshot published. (Data snapshot). Reason: .
  • 2026-05-03Nursing-home research snapshot published. (Data snapshot). Reason: .
  • 2026-05-03CMS Care Compare display rules — research vs profile separation. (Research rule). Reason: .
  • 2026-05-03Florida state-board contractor-license display rules (superseded). (Display rule). Reason: .
  • 2026-05-03CMS PECOS Medicare-enrollment indicator display rules. (Display rule). Reason: .
  • 2026-05-03NPPES NPI display rules + non-endorsement doctrine. (Display rule). Reason: .
  • 2026-05-03Source-provenance schema codified. (Schema). Reason: .

For corrections: corrections@fonteum.com or fonteum.com/contact. For data licensing: press@fonteum.com.

What’s on file, by the numbers

Platform snapshot · 2026-08-11

13.4Mproviders & companiesProviders, organizations, owners, and facilities on file
26.2Msource-linked factsSource-linked field facts in the dated platform snapshot
90sources with dataDistinct snapshot source IDs with at least one positive record count
73fresh sourcesDistinct source IDs whose latest positive-data snapshot falls within the preceding 45 days
111sources integratedActive registry rows; integration does not establish a load
13state Medicaid jurisdictionsDistinct states represented in the state-exclusions serving table

Integrated, with-data, and fresh-observation counts are separate. No platform-wide source-completeness count is published. Completeness is source-specific and must be evaluated against the named source's expected scope. State coverage is a separate jurisdiction measure.

Source authority is record-specific

Use the issuer named on the record.

Fonteum spans federal, state, and global public publishers. A source page or returned record identifies its issuer and dataset where that metadata is available. A platform registry count does not assign every page to one authority or establish loaded, fresh, or complete coverage.

Browse source records and their stated limitations →

Reproducible by design

Inspect the evidence each published figure actually supplies.

Source and date

Research pages expose the named public file and observation date where those fields are available. Source-file SHA-256 coverage is separate; facts do not currently link deterministically to signatures.

Available derivation

Studies with a retained release and committed derivation link the SQL or method used. Other studies state the evidence and reproduction limits they actually have.

Daily observations

Dated table row-count observations can detect local drift. They do not imply that an upstream publisher released or Fonteum ingested new data that day.

Named medical review

Reviewed by Jennifer Montecillo, MD, medical reviewer. Non-practicing medical reviewer.

Read the full provenance and attestation methodology →

Request access