Specialty labels contradict each other across sources. "Practice size" means something different to every data provider. And a name that looked current six months ago might belong to a physician who's since retired or switched health systems.
Teams doing healthcare B2B research, account planning, recruiting, or referral development run into this constantly. A general workforce estimate from AAMC won't tell you how many cardiologists with 25+ affiliated physicians operate in a specific metro area. You need a method, not a download button.
One researcher can do this well for twenty physicians. The problem arrives at two thousand, across a whole specialty nationwide, where every inconsistent label and every stale affiliation multiplies. This guide walks through defining your criteria, choosing sources, matching records correctly, and keeping the list usable after launch at that volume.
Key Takeaways
- Lock down specialty and size definitions before you collect a single record
- Combine authoritative sources instead of relying on one directory
- Build deduplication, verification, and privacy review into the process, not as an afterthought
- Match your build method (manual, vendor, or automated) to your list's size and refresh needs
- Document the source and date behind every field so reviewers can trust the data later
How to Build a Physician List by Specialty and Size
Step 1: Define the List's Purpose and Inclusion Criteria
Start by writing down what the list is actually for. Recruiting outreach needs different fields than referral mapping or account scoring for a sales team.
Decide the unit of record early:
- Individual physician: one person, regardless of where they practice
- Practice location: a physical site that may house multiple physicians
- Medical group or health system: the parent organization
- Physician-organization relationship: a linked record connecting a person to a specific site
Pick a size measure and stick with it. Mixing physician count with revenue or bed count in the same filter produces nonsense results.
Write exclusion rules upfront and flag these before they enter the working file:
- Retired physicians
- Duplicates
- Out-of-scope geographies
- Records you can't verify
Step 2: Select and Collect Source Data
No single source covers everything. The NPPES registry run by CMS gives you identity, taxonomy codes, and practice addresses, but NPI issuance doesn't confirm a physician is currently licensed or actively practicing.
Layer in additional sources depending on what you need:
- State licensing boards for current license status and disciplinary history
- Specialty-board directories for certification (not the same as a self-reported taxonomy code)
- Health system and practice websites for affiliation and location details
- Licensed commercial healthcare datasets for broader coverage and standardized fields
For every source, record the name, access date, terms of use, and exactly which field it supports. A public directory listing is not proof of current affiliation, and it's definitely not permission to contact someone.
Step 3: Normalize, Match, and Categorize Records
Different sources format the same physician differently. "Dr. J. Smith, MD" and "Jonathan Smith, M.D." need to become one standardized record before anything else works.
Build separate fields for:
- Primary specialty versus subspecialty
- Taxonomy code or source classification (keep the original label alongside the mapped one)
- Practice name, organization size, and location count
- Confidence or verification status
Match on multiple signals: name, NPI, specialty, address, and license number together, not name alone. Route anything uncertain to a human reviewer instead of forcing a merge. This is also where you preserve the raw source value next to the normalized one, so anyone reviewing the record later can see how a classification decision was made.
Step 4: Validate, Publish, and Maintain the List
Before the list goes anywhere near a CRM, check for:
- Duplicates
- Inactive licenses
- Conflicting specialty labels
- Physicians tied to more than one organization
Assign each record a status (verified, partially verified, needs review, or excluded) with a documented reason. Test your filters against a sample batch. If a "large cardiology groups" filter pulls in solo practitioners, something's broken upstream.
A structured workflow helps here. Intellectible's connected workflow pattern pulls from multiple sources, scores records against defined fit criteria, pauses for human review on anything ambiguous, and routes approved results into a CRM or project database.
Records live inside a project-scoped PostgreSQL database with source dates and change history intact, so an owner can trace any field back to its source. That doesn't remove the need to validate legal, privacy, and security requirements for your use case, but it does take manual spreadsheet merging out of the picture.
When Should You Build This Type of List, and What Do You Need First?
A specialty-and-size list earns its cost when generic physician directories are too broad for what you're trying to do. Account planning, recruiting campaigns, referral development, and targeted outreach all need repeatable segmentation that a flat directory can't give you.
Before starting, get these in writing:
- A data dictionary defining every attribute
- Inclusion and exclusion rules
- A source plan naming which systems supply which fields
- Geographic boundaries and specialty taxonomy
- A practice-size definition
- A named list owner and refresh schedule
Systems and Tools
At minimum, you need:
- A database or spreadsheet with filtering capability
- Record-level source tracking
- Duplicate detection
- A review workflow for flagged records
- Access controls
- An export or CRM integration path
Spreadsheets work for a few hundred records; anything larger needs a real database.
Source Data Requirements
Source data has to be current enough to support both specialty classification and organization-level relationships. Some sources, notably individual license registries, identify a physician reliably but say nothing about practice size. Don't force a size estimate out of a source that was never built to provide one.
Data Stewardship and Compliance
Someone needs to own data stewardship: source licensing terms, privacy review, outreach permissions, suppression requests, and compliance with applicable U.S. communications rules. This isn't a one-time checklist item. It's an ongoing responsibility tied to whoever owns the list.
Key Parameters That Affect Physician List Quality
List quality comes from the definitions you apply during collection, not the record count you end up with. Four variables matter most.
Specialty Taxonomy and Granularity
Specialty names vary wildly across directories. NUCC taxonomy codes are provider-selected and describe classification, not certification or legal scope of practice. Board certification, tracked separately by ABMS member boards, is a different attribute entirely.
A physician might carry a primary specialty plus one or more subspecialties, and conflating the two creates false precision. Map every source label to a documented taxonomy, and never quietly convert a broad specialty into a narrow subspecialty just because a directory implied it.
Practice-Size Measurement
"Large practice" means something different depending on who's asking:
- Physician headcount (including all practice locations)
- Physician or provider FTE, which can differ meaningfully from headcount
- Number of locations
- Hospital beds, for facility-based entities
- Parent-organization scale, separate from local practice size
Pick one primary measure, keep the raw value and its source date, and only create size bands once you have enough consistent data to support them.
Entity Relationships and Deduplication
Physicians work at multiple locations. Groups sit inside larger health systems. Name and credential formats vary by source.
Without a data model that separates people, organizations, locations, and affiliations, you'll either double-count a physician or merge them into the wrong practice entirely. Anchor identity on a stable identifier like NPI, then corroborate with name, license number, and address, not name alone.
Data Freshness and Verification Confidence
Licenses lapse. Affiliations change. Contact details go stale. Assign a last-verified date and confidence rating to each field, and set a refresh priority based on how often that field actually changes and how much it matters for your use case. High-value records deserve more frequent rechecking than the rest of the list.
Common Mistakes, Troubleshooting Issues, and Alternatives
Most physician list problems trace back to a handful of recurring mistakes.
Skipping the definition stage. Teams download a directory before deciding whether they're targeting physicians, practices, locations, or systems. Fix: review your data dictionary and test inclusion rules against sample records before scaling up.
Treating specialty or size labels as self-explanatory. Combining "hospitalist," "internal medicine," and "cardiology" without documenting whether each label reflects training, certification, or marketing copy leads to messy filters. Fix: preserve source labels and map them to a documented taxonomy.
Accepting duplicate or stale records. Joining directories by name alone, without checking license status or update dates, lets outdated records slip through. Fix: match on multiple fields and maintain a suppression queue for anything questionable.
Overlooking privacy and outreach restrictions. Public visibility isn't a license to copy, enrich, or email at scale. The FTC's compliance guide makes clear that CAN-SPAM covers business-to-business email, not just consumer marketing. Fix: review source terms, applicable privacy rules, and suppression requests before any outreach goes out.
With those pitfalls controlled, choose a build method that fits your list size, refresh cadence, and review capacity.
Three Ways to Build the List
| Method | Best for | Trade-off |
|---|---|---|
| Manual research | Small, high-value lists needing nuanced judgment | Slow, hard to standardize, difficult to refresh consistently |
| Licensed data provider | Broad coverage with recurring updates | Subscription costs, coverage gaps, need to validate vendor methodology |
| Automated workflow | Multiple sources, frequent refreshes, CRM integration | Requires source governance and human review, but scales without added headcount |
For teams past the spreadsheet stage, an automated approach handles the heavy lifting. Intellectible's engines connect to multiple source systems, apply specialty and size filters, and flag exceptions for review. Results publish to a working interface or dashboard without requiring a developer to build that pipeline from scratch.
Conclusion
A reliable physician list starts with precise definitions, not a bigger download. Know what you mean by "physician," "organization," "specialty," and "size" before you pull a single record.
After that, four practices decide whether the list still works in six months or becomes another stale spreadsheet:
- Source tracking
- Entity matching
- Verification status
- Privacy review
Choose the simplest method (manual, vendor, or automated) that meets your coverage, refresh, and compliance needs. Don't over-engineer a 200-record list, and don't try to manually maintain a 20,000-record one.
Whatever you choose, the list itself is not the deliverable. The teams that get value out of this run it as a monitored process rather than a one-time pull: sources re-queried on a set cadence, license lapses and affiliation changes picked up as they happen, and review effort spent only on the records that actually moved. Built that way, the list stays accurate across campaigns instead of being rebuilt from scratch every time someone asks for it.
Frequently Asked Questions
What is a good physician-to-population ratio?
There's no single benchmark. It varies by specialty, care model, population health needs, and geography. AAMC's 2025 workforce data reports around 304 active physicians per 100,000 people nationally, but that figure shifts significantly by specialty and region.
How do I define a physician's specialty when sources disagree?
Preserve the original label from each source and map it to a documented taxonomy. Keep primary specialty separate from subspecialty, and flag any conflicting classifications for human review rather than guessing.
What's the best way to measure practice size?
It depends on your purpose. Physician count works for headcount-based segmentation, FTE accounts for part-time staff, and location count or bed count suit facility-focused analysis. Pick the measure your source data actually supports.
How do I remove duplicate physicians from a list?
Match on multiple fields, such as NPI, name, license number, and address, instead of name alone. Use stable identifiers where available, and handle physicians affiliated with multiple organizations as a separate case rather than a duplicate.
How often should a physician list be updated?
Update frequency should match how often each field changes and how much it matters to your use case. License status and affiliations need more frequent checks than general demographic details.
Can I use public physician information for marketing?
Public availability doesn't grant automatic reuse or contact rights. Review source terms, applicable communications rules like CAN-SPAM, suppression requests, and your organization's own compliance policies before any outreach.
