Healthcare sales, credentialing and staffing teams buy a list of practices that just appeared in NPPES — not an 8-million-row dump of the entire National Provider Identifier registry. The official NPI Registry API cannot search by enumeration date and caps each query at 1,200 rows. CMS already publishes those new and updated NPIs as a small weekly zip (about 6–15 MB).
This walkthrough uses the New NPI Numbers: Newly Registered Medical Practices & Doctors Actor (quarrylane-data/nppes-npi-new-provider-leads) to read that official weekly incremental file, keep rows whose enumeration date falls in your window, and return one normalized row per NPI. There is no CMS login and no API key. The Actor never downloads the roughly 1 GB monthly full file.
You need an Apify account and an API token from Console → Settings → API & Integrations. Keep the token in a secret or credential vault — never in a public workflow, gist or blog comment.
Why this list matters
Medical device, lab, imaging and specialty-pharma reps want new clinics in a state or taxonomy. Credentialing and payer-network teams watch a specialty or territory. A weekly Clay, n8n or Make job can drop those practices into a CRM before the rest of the field notices them.
By default the Actor returns organizations only (NPI-2): legal business name, practice and mailing address, phone, taxonomies and the official NPI Registry URL. Individual (NPI-1) records are public federal data but are consumer-adjacent, so they are opt-in (entityType: "individuals" or "both") and limited to professional fields. Leave the default on unless you have a clear reason to include individuals. EIN and tax IDs are never included. An NPI being issued does not mean the provider is licensed or credentialed.
Run it once from curl or Python
run-sync-get-dataset-items starts a run, waits for it to finish, and returns the dataset items in one response. The HTTP call times out at 300 seconds; this Actor's default timeout is 5 minutes, and a weekly-file pull is usually much faster. Use a tilde (~) between the username and the Actor name.
Input fields from the schema:
-
daysBack(default 21, so a run still overlaps the latest weekly file), orstartDate/endDateasYYYY-MM-DD -
entityType— defaultorganizations;individualsorbothare opt-in -
taxonomies— NUCC codes (for example207Q00000X) or text such asFamily Medicine -
states,cities— primary practice location -
activeOnly— defaulttrue; skips deactivated NPIs -
onlyNew+monitorKey— monitor mode -
maxItems— default 25, maximum 20,000; free Apify plans are capped at 15 results per run
CMS currently hosts only the latest weekly incremental zip. Older weekly URLs return 404. The feed can lag the live registry by up to about a week, and a custom date window that does not overlap the published week returns no rows. If you leave the dates at their defaults and the 21-day window ever misses the listed file, the Actor falls back to that file's own week so a default run still returns data.
The snippets below match a succeeded run on 13 Nov 2026 that asked for organizations in Texas over the last 21 days:
export APIFY_TOKEN="YOUR_APIFY_TOKEN" curl -sS -X POST \ "https://api.apify.com/v2/acts/quarrylane-data~nppes-npi-new-provider-leads/run-sync-get-dataset-items" \ -H "Authorization: Bearer $APIFY_TOKEN" \ -H "Content-Type: application/json" \ -d '{ "daysBack": 21, "entityType": "organizations", "states": ["TX"], "activeOnly": true, "maxItems": 10 }' import json, os, urllib.request TOKEN = os.environ["APIFY_TOKEN"] url = ( "https://api.apify.com/v2/acts/" "quarrylane-data~nppes-npi-new-provider-leads/" "run-sync-get-dataset-items" ) body = { "daysBack": 21, "entityType": "organizations", "states": ["TX"], "activeOnly": True, "maxItems": 10, } req = urllib.request.Request( url, data=json.dumps(body).encode(), headers={ "Authorization": f"Bearer {TOKEN}", "Content-Type": "application/json", }, method="POST", ) with urllib.request.urlopen(req, timeout=300) as resp: items = json.loads(resp.read()) print(len(items), "practices") print(json.dumps(items[:2], indent=2)) Add "taxonomies": ["Family Medicine"] (or a NUCC code) when you want a specialty cut.
Sample output
These two organization records came from that Texas run on 13 Nov 2026 (2:39pm Australia/Brisbane). The weekly file covering 2–8 Nov 2026 was NPPES_Data_Dissemination_110226_110826_Weekly_V2.zip. Organization records also include an authorizedOfficial object (name, title, phone); those personal-name fields are omitted here so the example stays on practice-level business data. A few null and individual-only fields are omitted for brevity.
[ { "recordId": "1982504668", "npi": "1982504668", "entityType": "NPI-2", "entityTypeLabel": "Organization", "status": "Active", "enumerationDate": "2026-11-06", "lastUpdateDate": "2026-11-06", "certificationDate": "2026-11-06", "organizationName": "ADYT DENTAL PLLC", "otherNames": [ "Avant Dentistry" ], "isOrganizationSubpart": false, "practiceAddress": "909 LEGACY DR # 100", "practiceCity": "PLANO", "practiceState": "TX", "practiceZip": "75023", "practicePostalCode": "750238202", "practiceCountry": "US", "practicePhone": "2146999828", "primaryTaxonomyCode": "122300000X", "primaryTaxonomy": "Dentist", "taxonomies": [ { "code": "122300000X", "description": "Dentist", "primary": true } ], "npiRegistryUrl": "https://npiregistry.cms.hhs.gov/provider-view/1982504668", "source": "NPPES", "sourceFile": "weekly-incremental", "mailingAddress": "PO BOX 1715", "mailingCity": "ALLEN", "mailingState": "TX", "mailingZip": "75013", "scrapedAt": "2026-11-13T04:39:42.676Z", "sourceWeekStart": "2026-11-02", "sourceWeekEnd": "2026-11-08", "sourceFileName": "NPPES_Data_Dissemination_110226_110826_Weekly_V2.zip" }, { "recordId": "1851291637", "npi": "1851291637", "entityType": "NPI-2", "entityTypeLabel": "Organization", "status": "Active", "enumerationDate": "2026-11-06", "lastUpdateDate": "2026-11-06", "certificationDate": "2026-11-06", "organizationName": "VALENCIA FAMILY MEDICINE PLLC", "isOrganizationSubpart": false, "practiceAddress": "10520 MONTWOOD DR", "practiceCity": "EL PASO", "practiceState": "TX", "practiceZip": "79935", "practicePostalCode": "799352703", "practiceCountry": "US", "practicePhone": "9152199306", "practiceFax": "8334557975", "primaryTaxonomyCode": "207Q00000X", "primaryTaxonomy": "Family Medicine Physician", "taxonomies": [ { "code": "207Q00000X", "description": "Family Medicine Physician", "primary": true } ], "npiRegistryUrl": "https://npiregistry.cms.hhs.gov/provider-view/1851291637", "source": "NPPES", "sourceFile": "weekly-incremental", "mailingAddress": "10520 MONTWOOD DR", "mailingCity": "EL PASO", "mailingState": "TX", "mailingZip": "79935", "mailingPhone": "9152199306", "scrapedAt": "2026-11-13T04:39:42.676Z", "sourceWeekStart": "2026-11-02", "sourceWeekEnd": "2026-11-08", "sourceFileName": "NPPES_Data_Dissemination_110226_110826_Weekly_V2.zip" } ] Map organizationName, npi, practiceCity, practiceState, practicePhone, primaryTaxonomy, enumerationDate and npiRegistryUrl into your CRM or Clay columns.
n8n: a weekly scheduled feed
CMS publishes the incremental file weekly, so a daily job often returns the same week until the next zip appears. A weekly schedule is the natural cadence; a daily monitor still works if you turn on onlyNew (see below).
- Add a Schedule Trigger (for example, Monday at 07:00).
- Apify node (
@apify/n8n-nodes-apify): add an Apify credential, set the operation to Run Actor and get dataset items, selectquarrylane-data/nppes-npi-new-provider-leads, and paste the JSON input above. - Or use an HTTP Request node: Method
POST, the same URL,Authorization: Bearer …,Content-Type: application/json, body as JSON. Set the node timeout to at least 300 seconds. - Split Out the array and write each practice to Sheets, a CRM or Slack.
Make.com
Connect the official Apify app with OAuth or an API token. Add Run an Actor, choose quarrylane-data/nppes-npi-new-provider-leads, paste the input, and set "Run synchronously" to Yes for this relatively short job. Follow it with Get Dataset Items, mapping the default dataset ID from the previous module.
If you prefer one step, call run-sync-get-dataset-items from an HTTP module. Use Watch Actor Runs plus Get Dataset Items only if you start the run elsewhere or expect to exceed Make's sync timeout. Trigger the scenario weekly once the mapping is correct.
Clay
HTTP API as source is the clean way to import a new list of practices: Method POST, the same URL and Bearer token, body the JSON above. Map the fields listed earlier.
You can also add an enrichment: search Apify → Run Apify Actor, or add an HTTP API enrichment column with the same POST details. Clay's HTTP API docs use /ColumnName inside a quoted string to pull in another column (for example a state code). Keep maxItems small if the column runs once per row — a table-wide enrichment that starts a new Actor run on every row will multiply cost. For a weekly territory list, one source pull (or one control row) is enough.
Monitor mode: later runs return only new NPIs
Set "onlyNew": true and a stable "monitorKey" (for example "tx-orgs-weekly"). Use a different key per filter set.
The first run bills up to maxItems as standard provider-record events and remembers every other current match in the window without charging. Later scheduled runs return only NPIs not seen before and bill the new-provider event instead. Typical weeks contain a few thousand new organizations nationally; a state or taxonomy filter is much smaller. Updates to older NPIs in the same weekly file are ignored unless their enumeration date is also in the window.
What it costs
Pay per result, with no monthly rental of the Actor:
| Event | When it is charged | Price |
|---|---|---|
provider-record | Each newly enumerated NPI in a standard run, including the first monitor run (up to maxItems) | $0.005 ($5 per 1,000) |
new-provider | Each NPI delivered on a later onlyNew run | $0.025 ($25 per 1,000) |
Platform compute is usually a fraction of a cent because the Actor only downloads the weekly zip. Set Max cost per run for a hard budget.
Worked example: a first Texas organizations run with maxItems 10 costs $0.05 in events (the run that produced the sample above delivered 10 rows). A later weekly monitor that returns, say, 80 new Texas organizations costs $2.00 in new-provider events. On the free Apify plan each run is capped at 15 rows, which is enough to inspect the schema.
Try the Actor
The Actor is on the Apify Store: quarrylane-data/nppes-npi-new-provider-leads. Start with organizations in one state, then add onlyNew and a weekly schedule.
Built by Quarrylane Data. Not affiliated with CMS or NPPES. Data comes from the public weekly files at download.cms.gov/nppes.

