<\/span><\/h2>\nThis is for recruiters and recruiting ops teams cleaning provider lists and matching records before<\/em> outreach\u2014especially when you\u2019re merging multiple sources and fighting duplicate records<\/strong>.<\/p>\nIf you\u2019ve ever had a candidate reply \u201cwrong person,\u201d or you\u2019ve burned time chasing a shared clinic line, you\u2019re dealing with an identity resolution<\/strong> problem, not a messaging problem.<\/p>\n<\/span>Quick Answer<\/span><\/h2>\n\n- Core Answer<\/dt>\n
- Anchor each physician to NPI, then confirm with license matching (state + license number) before merging records or attaching contact channels for outreach.<\/dd>\n
- Key Insight<\/dt>\n
- NPI is typically more stable than contact methods; licenses disambiguate same-name clinicians and document multi-state practice footprints.<\/dd>\n
- Best For<\/dt>\n
- Recruiters\/ops cleaning lists and matching records before outreach.<\/dd>\n<\/dl>\n
\nCompliance & Safety<\/strong><\/p>\nThis method is for legitimate recruiting outreach only. Always respect candidate privacy, opt-out requests, and local data laws. Heartbeat does not provide medical advice or legal counsel.<\/p>\n<\/blockquote>\n
<\/span>Framework: The \u201cRight Person\u201d Checklist: NPI \u2192 License \u2192 Specialty \u2192 Location \u2192 Contact<\/span><\/h2>\nIn recruiting, the cost of a bad match shows up fast: wrong-person calls, spam complaints, wasted recruiter hours, and messy suppression history.<\/p>\n
\n- NPI<\/strong>: stable anchor identifier for a clinician record.<\/li>\n
- License<\/strong>: disambiguates duplicates and confirms multi-state footprint.<\/li>\n
- Specialty<\/strong>: sanity check for role fit and collisions.<\/li>\n
- Location<\/strong>: helps separate similar identities and route territories.<\/li>\n
- Contact<\/strong>: only after identity is resolved; then optimize for connectability and deliverability.<\/li>\n<\/ul>\n
The trade-off is\u2026<\/strong> faster auto-merges versus wrong-person risk. The workflow below keeps speed while putting hard stops where collisions happen.<\/p>\n<\/span>Step-by-step method<\/span><\/h2>\n<\/span>Step 1: Write down your matching definition (so merges are consistent)<\/span><\/h3>\nMatching definition (required)<\/strong>: Two records represent the same physician<\/strong> when the NPI<\/strong> matches exactly and<\/em> at least one license matches on (state + license number). Use name\/specialty\/location as secondary validation, not as primary keys.<\/p>\nRequired identifiers definition (required)<\/strong>: The minimum fields to attempt a deterministic match are: NPI (10-digit), license state, license number, and at least one of (full name or last name + first initial). If license is missing, the record is a candidate match only and must not be auto-merged.<\/p>\n<\/span>Step 2: Normalize inputs (most matching failures start here)<\/span><\/h3>\n\n- NPI<\/strong>: store as a 10-character string; preserve leading zeros.<\/li>\n
- License number<\/strong>: strip spaces and punctuation; preserve leading zeros if present.<\/li>\n
- License state<\/strong>: normalize to two-letter state code.<\/li>\n
- Name<\/strong>: split first\/middle\/last\/suffix; normalize suffixes (Jr, III).<\/li>\n
- Specialty<\/strong>: map to your internal taxonomy; don\u2019t compare raw text.<\/li>\n<\/ul>\n
<\/span>Step 3: Run deterministic matches first (high confidence)<\/span><\/h3>\n\n- NPI exact match<\/strong> \u2192 same-identity candidate.<\/li>\n
- Within NPI match, license matching<\/strong> on (state + license number) \u2192 confirm identity and capture multi-state licenses.<\/li>\n<\/ol>\n
License status handling:<\/strong> if your source includes license status, treat \u201cactive\u201d as a stronger confirmation signal than \u201cunknown,\u201d but still validate with at least one secondary check (specialty, location, or name\/suffix). If status is missing, store the license but don\u2019t let it be the only confirmation.<\/p>\n<\/span>If\/Then routing summary (use this in your ETL)<\/span><\/h3>\n\n- If<\/strong> NPI matches and at least one license matches (state + number), then<\/strong> auto-merge.<\/li>\n
- If<\/strong> NPI matches but license conflicts, then<\/strong> route to review (collision risk).<\/li>\n
- If<\/strong> license matches but NPI is missing, then<\/strong> hold as candidate match and backfill NPI through legitimate sources.<\/li>\n
- If<\/strong> only name\/specialty\/location match, then<\/strong> do not merge; collect required identifiers first.<\/li>\n
- If<\/strong> contact channels match but NPI differs, then<\/strong> treat the channel as shared and keep identities separate.<\/li>\n<\/ul>\n
<\/span>Edge case: individual vs organizational NPI<\/span><\/h3>\n\n- Don\u2019t merge an organization record into an individual physician record just because a clinic address or phone overlaps.<\/li>\n
- Keep separate identity types in your data model (individual clinician vs organization) and only compare like-to-like during matching.<\/li>\n
- If your source doesn\u2019t label type clearly, route to review rather than auto-merge when the record looks like a facility (generic name, shared switchboard, multiple specialties).<\/li>\n<\/ul>\n
<\/span>Step 4: Route collisions to a review queue (the \u201ctwo Dr. Smiths\u201d mini-case)<\/span><\/h3>\nCollision<\/strong> = two different people whose records appear mergeable until you check identifiers.<\/p>\nMINI_CASE: two Dr. Smiths.<\/strong> Your CRM has \u201cJohn Smith, MD\u201d from a conference list. Your ATS has \u201cJohn A. Smith\u201d from a referral. Same specialty, same metro. A rushed merge creates one record, and your team calls the wrong person\u2014twice\u2014because the phone number belonged to the other Smith.<\/p>\nHow NPI license matching prevents this:<\/p>\n
\n- You treat conference\/referral entries as candidate matches until anchored to an NPI<\/strong>.<\/li>\n
- Once NPI is present, you require at least one matching license (state + number) before merging.<\/li>\n
- If NPI matches but license conflicts, you flag a collision and route to review instead of auto-merge.<\/li>\n<\/ul>\n
<\/span>Step 5: Implement a golden record merge policy (explicit example)<\/span><\/h3>\nA merge policy is how you stop \u201clast import wins\u201d from corrupting identity. Here\u2019s a concrete example you can adopt.<\/p>\n
![\"54405\"](\"http:\/\/hc.heartbeat.ai\/wp-content\/webp-express\/webp-images\/uploads\/2026\/02\/npi-license-matching-3702a5dc.png.webp\")
<\/p>\n