How New York Behavioral Health Teams Verify Provider Listings in Salesforce
Why Do New York Behavioral Health Provider Listings Need Careful CRM Review?
In this anonymized composite scenario, a New York behavioral health network team finds stale provider-location-plan listing rows in Salesforce after roster imports. An illustrative 15,000-row review shows how admins verify practice location and network participation, preserve legitimate multi-site relationships, check directory-facing reports, and protect approved corrections with preview, backup, bulk undo, and restore. This is not a named customer result.

Why New York Provider-Directory Evidence Points to This Data Problem
The New York Attorney General’s 2023 investigation of mental-health provider directories called 396 listings across 13 plans and obtained appointment offers from only 56; it classified 86% of the contacted listings as “ghosts” because the provider was unreachable, outside the listed network, or not taking new patients. That audit measured a sample of directory contacts, not the duplicate rate of any Salesforce database. The state’s Provider & Health Plan Look-Up guidance adds the exact record-grain warning: a provider may practice at several locations, and a plan’s network may cover some locations but not others. For a New York CRM used to prepare provider directories, the critical unit is therefore a provider–practice location–health plan listing with a verified status and date. A stale active row can inflate a network report or drive an incorrect directory export, while merging all rows by provider name or NPI can erase real location distinctions. Neither source is evidence of a ZaapIT customer outcome.
Use Case: An Old Practice-Location Listing Remains Active
A behavioral health network team imports monthly rosters into Salesforce custom Provider Listing records. A clinician’s NPI appears on an older Manhattan office row and a current Brooklyn office row. The source feed updates the Brooklyn row but does not retire the Manhattan row because the address spelling and source record ID changed. A directory-facing report still includes the Manhattan location as active for one plan, although its participation status has not been reverified. The clinician is not a duplicate person; the stale row is a separate provider–location–plan association that needs review.
The team first determines which source owns each fact: NPI for clinician identity, a normalized address and practice-site identifier for location, the plan’s current roster for network participation, and a separately verified appointment-availability field. It preserves the Brooklyn association and any other genuinely active offices. It does not equate one NPI with one location or treat an outdated phone number as proof that the clinician no longer practices anywhere. A joint AMA–CAQH provider-directory analysis also recommends distinguishing locations where clinicians regularly see patients from other contracted locations, reinforcing this site-level review.
| Salesforce grain | Verification key | Unsafe shortcut |
|---|---|---|
| Clinician Contact or Provider record | NPI and authorized identity evidence | Merge different clinicians on similar names. |
| Practice location | Normalized address, site ID, and current office confirmation | Assume all addresses under one NPI are duplicates. |
| Provider–location–plan listing | Plan roster, effective date, active status, and verification date | Carry a prior location’s active network flag forward. |

How the Admin Profiles and Corrects the Listing Rows
- Define the target population. Review an illustrative 15,000 custom listing rows with clinician NPI, location key, plan ID, effective dates, source file, last verification date, status, and linked provider and site records. Record pre-change counts by plan and geography.
- Group likely conflicts. Use ZaapIT’s filtering and matching workflow to find repeated provider–location–plan combinations, alternate address strings, and conflicting active flags. Review candidates rather than automatically merging rows on NPI alone. The CMS Doctors and Clinicians data dictionary likewise treats individual clinician, enrollment, group, and address as a composite data grain, with clinicians legitimately appearing on multiple lines.
- Confirm current facts. Compare with the plan’s current roster and approved provider outreach. Retire or correct a listing only after the authorized owner confirms its network and location status. Escalate unverified appointment availability instead of inventing it.
- Preview and back up. Preview field-level changes and related reporting effects in ZaapIT. Back up the affected rows and relationship mappings, then process a small approved batch before expanding to the rest of the reviewed population. Test supported bulk undo and restore on a controlled subset, since operational safety depends on recovery from the specific change being made.
- Stop reimport drift. Normalize the provider–site–plan composite key, retain feed provenance and effective dates, and quarantine source rows that conflict with verified status. Make old status values expire or trigger re-verification rather than silently flowing into the next directory cycle.
Salesforce’s matching-rule documentation explains its standard rules for Accounts, Contacts, and Leads, with custom rules available. Its Flow duplicate-check guidance can update or skip a matching record depending on configuration. Neither automatically verifies a custom provider–location–plan row’s real network participation. The composite key and review policy have to be designed for this use case.
Validate the Directory Feed and the Recovery Procedure
Re-run the same Salesforce report filters used for the directory feed. Count active listings by plan and location, compare them with the verified roster, and sample corrected Manhattan rows plus preserved Brooklyn rows. Check that each active listing has a current verification date and that no genuinely distinct practice site disappeared. Compare a small set of directory exports with the CRM source rows; a cleaner CRM report alone does not prove the public directory has updated.
Document the number of approved corrections, unresolved exceptions, and any changes to provider coverage counts as results of the illustrative workflow only. Test whether ZaapIT’s supported bulk undo or restore returns field values and relationships for a sample batch. If an operation has different recovery behavior, record the manual steps before a broader run.

Make Provider Listing Corrections Reviewable
Use ZaapIT to profile Salesforce listing rows, preview approved corrections, back up affected records, and test bulk undo and restore where supported before a directory feed is refreshed.
Frequently Asked Questions
Why can one New York clinician have several legitimate directory rows?
A clinician can practice at multiple locations and may participate in a plan at some sites but not others. Verify each provider–location–plan association separately.
Does a matching NPI prove two provider listings are duplicates?
No. An NPI identifies the provider; it does not identify a single practice location or plan relationship. Compare the location, plan, effective dates, and source evidence.
What should an admin check before marking a listing active?
Confirm the practice site, plan participation, effective dates, and verification source. Appointment availability requires its own current confirmation.
How can the team recover from a mistaken bulk listing update?
Preview the intended change, back up affected rows and relationships, run a small batch, and test supported bulk undo or restore before a wider correction.
Useful ZaapIT Resources
Sources
- New York Attorney General: Inaccurate and Inadequate provider-directory investigation (2023)
- New York State Provider & Health Plan Look-Up FAQ
- AMA and CAQH: Improving Health Plan Provider Directories
- CMS Provider Data Catalog: Doctors and Clinicians file and data grain
- Salesforce Help: Matching Rules
- Salesforce Help: Check for Duplicate Records in Flow