From the monthly NPPES public file on download.cms.gov. CMS publishes the full ~7M row file plus a deactivation list every month under an open license. SleekRank reads the file URL directly, so the data ships from CMS not from your theme repository. The file is in CSV format.

CMS publishes the full NPPES file monthly, typically in the first week of the month. New enumerations and deactivations happen daily on the live NPPES portal, but the public file rolls them up monthly. Setting cache_duration to monthly aligns SleekRank's refresh with the CMS publication schedule.

Type 1 NPIs identify individual providers (physicians, nurses, dentists, pharmacists). Type 2 NPIs identify organizational providers (hospitals, group practices, pharmacies, billing services). The data shape is different enough that most sites render them with distinct templates.

Yes. NPPES publishes both practice location and mailing address per NPI, along with practice phone and fax. Map them into a LocalBusiness JSON-LD block plus a visible address card on the page. Suppress fax in the visible card; keep phone and address so the page is useful for patients verifying a clinician.

NPPES never reuses an NPI. When a number is deactivated, it appears in the monthly deactivation file. SleekRank picks up the deactivation on the next refresh; the URL still resolves but renders with a deactivation banner.

Taxonomy codes follow the NUCC Provider Taxonomy Code Set, which is itself published as a CSV on nucc.org. Load NUCC as a secondary data source, join NPPES taxonomy codes against it, and render the human-readable specialty in the template. Without the join the page shows opaque codes that nobody searches for directly.

Yes, but NPPES does not carry an explicit individual-to-organization link. Use the practice address as a soft join: individual NPIs with the same practice address as an organizational NPI usually bill under that organization.

Each resolved row is roughly 2-3 KB in wp_sleek_rank_items. Seven million NPIs translate to a 15-25 GB MySQL footprint, comfortable on a single dedicated server. Most operators start with one state or one Type (individual versus organizational) and scale outward as their crawl budget expands.