Provider Directory Intelligence / Reconciliation System

Provider directories are assembled from internal records, NPPES, payer/network files, and credentialing systems that were never designed to agree. Demographic mismatches, specialty mismatches, NPI mismatches, and participation discrepancies show up as patient-facing errors, audit findings, and claims friction. Exception handling is still mostly manual.

Directory inaccuracy is not a marketing problem. It affects patient access, regulatory directory requirements, network adequacy, claims, and provider experience. If the systems disagree, the public record is wrong — and someone has to prove which source is authoritative.

Analysts compared extracts in Excel, chased mismatches by email, and updated one system without writing back to the others. There was no durable exception object, no source-of-truth ranking, and no routing for who owns a specialty mismatch vs. an NPI mismatch.

• Reconciliation pipelines that compare internal provider data against NPPES, payer/network data, and other credentialing/enrollment sources.
• Structured exception records instead of ad-hoc spreadsheet tabs.
• Routing by discrepancy type — demographics, specialty, NPI, network participation — so the right operational owner sees the case.
• A systems view of directory accuracy: not a one-time cleanup, a recurring comparison.

Compare → classify mismatch → open an exception → route → resolve against an authoritative source → write the decision back. The comparison is deterministic. Judgment stays with the operator when sources conflict.

Sources are labeled by access type. Restricted commercial or federal feeds are never presented as live credentials.

Mismatches are classified, not auto-corrected. An NPI mismatch, a specialty/taxonomy mismatch, and a network-participation discrepancy have different owners and different evidence requirements. The system proposes the discrepancy; it does not silently overwrite the directory.

This case study is an architecture for reconciliation pipelines on Salesforce / Apex, informed by directory and data-integrity work in operations. It is not a claim of a production multi-payer directory product or of restricted-source access.

• Apex-oriented comparison jobs and exception objects as the intended implementation path.
• Focus on classification and routing — the operational bottleneck — rather than a generic 'data quality dashboard.'

Editorial workflow layouts — not screenshots of a claimed production product.

1. View 01

   Exception queue grouped by mismatch class (NPI, specialty, demographics, participation).

2. View 02

   Side-by-side source panel: internal value vs. NPPES vs. payer file, with a required resolution note.

3. View 03

   Editorial diagram of the comparison pipeline rather than a mocked consumer app.

• Directory work is provider-operations work. Treating it as a reporting afterthought is why the same mismatches return every cycle.
• Source conflict is a workflow problem: who is allowed to decide, with what evidence, and where the decision is stored.

• Fold reconciliation exceptions into the Provider Ops Platform so directory mismatches and enrollment issues share one case model.