Ten digits. Intelligence-free. Permanent. Issued by CMS.
The National Provider Identifier was mandated under the HIPAA Administrative Simplification provisions and is issued by the Centers for Medicare and Medicaid Services 1 . The number is 10 digits, intelligence-free (the digits do not encode specialty, geography, or role), and permanent. A provider keeps the same NPI across employer changes, geographic relocations, marriage-name changes, and specialty transitions. The permanence is what makes the NPI work as the primary data key for entity reconciliation; downstream directories can pull, refresh, and update the surrounding data without losing the identity link.
Type 1 NPIs identify individual health care providers (physicians, dentists, nurse practitioners, physical therapists, sole-proprietor practices). Type 2 NPIs identify organization health care providers (group practices, hospitals, clinics, ambulatory surgical centers, anywhere an organization bills under its own name). A multi-physician practice typically carries both: each attending physician holds a Type 1 NPI, and the practice entity holds a Type 2 NPI. Billing flows pick the correct NPI per payer contract and per claim type. The Schema.org markup uses the Type 1 NPI on each Physician node and surfaces the Type 2 NPI on the MedicalBusiness node where the practice entity bills as the organization.
npiregistry.cms.hhs.gov. The upstream source the directory ecosystem reads.
The National Plan and Provider Enumeration System is the public CMS-maintained registry of every issued NPI 2 . NPPES returns the provider's legal name, the 10-digit NPI, the primary and secondary taxonomy codes (specialty), the practice addresses, the phone number, and the last update timestamp. Third-party directories pull the NPPES feed to populate physician profiles: Healthgrades, Zocdoc, Vitals, Doximity, and the rest of the medical-directory ecosystem all start with NPPES data. Drift between the NPPES record and the practice's current operating state propagates through the directory ecosystem until the NPPES record refreshes.
The architectural consequence is that NPPES is the upstream lever for cross-directory consistency. A practice that relocates, updates phone, transitions a physician to a new taxonomy, or registers a new attending has to update NPPES first. The downstream directories refresh on their own publication cycles (typically days to weeks). Any directory profile that drifts further (a Healthgrades profile updated by a previous staffer six years ago) gets reconciled at the directory level after NPPES is current. Maintaining NPPES current is the diagnostic-phase action item that runs before any directory-level work begins. The order matters: directory-level edits without an NPPES update get partially overwritten on the next pull.
NPI in the entity graphPhysician.usNPI and the sameAs chain.
The Schema.org Physician type provides the usNPI property as a direct, machine-readable linkage to the CMS National Provider Identifier registry 3 . The property is a 10-digit string. Including usNPI in the Physician JSON-LD bypasses the need for search engines to infer identity from name and address alone. Google's entity resolver reads the property as a typed identifier and follows it to the NPPES record for structured cross-reference.
The standard architectural pattern places usNPI as the primary identifier on each Physician node and uses sameAs as an array linking to the NPPES profile URL, the ABMS verification page 5 , the state medical board licensure page, the AMA Physician Masterfile reference where present 4 , and any high-authority directory profile the physician appears in (Healthgrades, Zocdoc, Vitals, Doximity, US News for hospital-affiliated specialists). The chain consolidates the physician's identity for Google's entity resolver and transfers the off-site authority of the cited surfaces to the on-site author byline.
The AMA Physician Masterfile predates the NPI (established 1906) and tracks physicians from medical-school entry through licensure and practice 4 . The AMA dataset and the NPPES dataset cover overlapping populations through different historical lenses. Third-party aggregators use both for triangulation. The NPI is the primary key for current operating state; the Masterfile carries the historical continuity that validates the foundational identity. The sameAs chain references both where applicable.
Without the chain, Google reads a fragmented physician identity. The on-site editorial competes for ranking on YMYL medical queries without the underlying entity signal that would let it surface. The chain is the load-bearing mechanism. Adding the chain to existing JSON-LD is the highest-impact architectural action in most diagnostic engagements.
The NPI sits at the foundation of the broader medical SEO work at Praxis. The identifier feeds the schema layer, the directory layer, and the editorial-author byline layer. The full architecture runs on the entity-graph reconciliation the NPI anchors.