A Deterministic Provider-Targeting Engine for Cash-Pay Prescriber Acquisition

Challenge

The platform wanted to find and reach roughly the 5,000 highest-value cash-pay-oriented prescribers to seed a paid-acquisition campaign. The underlying data made that hard: source records were fragmented across five vendor clinic databases with inconsistent identifiers and addresses, and the provider list carried almost no contactable emails. Before a single ad dollar could be spent responsibly, the targeting list had to be defensible (every provider on it traceable to the rules and source data that put it there), not a black-box score the marketing team could not explain.

Approach

We built a five-step deterministic data pipeline that did the matching, merging, and scoring as transparent, documented rules. AI was confined to the reading and language work at build time (reconciling messy vendor records, normalizing free-text addresses and specialties, drafting acquisition copy and channel notes). The decisions (which records are the same clinic, which prescriber maps to which clinic, and how each provider scores on cash-pay propensity) ran as versioned, auditable code and tables that a person can inspect and reproduce.

• Deduplicated 14,750 clinic records across five vendors using fuzzy name and normalized-address matching, collapsing inconsistent identifiers into a single clinic spine.
• Merged demographic and utilization overlays and consolidated 27 common fields with a source-priority waterfall, so every field has a known provenance and a known winner when sources disagree.
• Matched 517K Medicare Part D prescribers to clinics with a conservative strategy (exact matches accepted automatically, fuzzy matches flagged for review rather than auto-merged) to keep false joins out of the targeting list.
• Designed a transparent cash-pay propensity scoring model driven by specialty, geography, drug-cost and volume quartiles, and independent-versus-hospital signals, expressed as documented rules and lookup tables rather than an opaque model.
• Ran an email-acquisition strategy that empirically tested channels and produced ad-platform-ready upload files for Google, Meta, and LinkedIn.

Impact

• Delivered a ranked master list of 5,000 cash-pay-focused providers, with each provider’s rank traceable back to the source data and scoring rules that produced it.
• Matched 12,434 Medicare providers to 2,089 clinics through the conservative exact-plus-review approach, giving the platform a vetted prescriber-to-clinic graph rather than a loose name list.
• Sized the opportunity with a TAM, SAM, and SOM framing to ground spend decisions (these are modeled market estimates, not booked revenue).
• Produced empirically tested email-acquisition channels plus a costed paid-append recommendation and an ROI model, so the email-coverage gap had a concrete, priced path forward rather than an assumption.
• Left the matching and scoring logic documented and reproducible, so the list can be refreshed and re-justified as vendor data changes.

Capabilities demonstrated

• Multi-vendor entity resolution: deduplicating and reconciling clinic records across five inconsistent sources into one clean spine.
• Source-priority field consolidation with explicit provenance across 27 fields.
• Large-scale prescriber-to-clinic matching (517K Part D prescribers) with a conservative, review-gated join strategy.
• Transparent, rules-and-tables propensity scoring instead of opaque model output.
• AI confined to extraction and language while decisions run as deterministic, auditable code.
• Go-to-market activation: ad-platform-ready audience files plus a costed email-append and ROI recommendation.

Anonymized by design: client names stay off the narrative per our reference policy. Figures that are modeled, small-sample, or targets are identified as such above.