Success Story

Closing the Gaps in HCP Master Data with AI-Powered Enrichment

The Challenge

For a global pharmaceutical company, HCP master data held little beyond an ID, a name, a primary specialty, and a basic location – leaving most of the attributes the business actually needed were still missing.

While the records were usable as identifiers, they offered limited support for targeting, engagement, and analytics further downstream. Key challenges included:

Missing Contact Data Verified email addresses and direct phone numbers were absent for most records, limiting outreach and campaign reach.

Incomplete Demographics and Affiliation data Institutional relationships, specialties, and demographic detail were too sparse to support precision targeting or KOL identification.

Manual Exception Management Low trust in the master record meant data stewards spent their time resolving exceptions rather than improving data quality at the source.

Limited Enrichment Traceability. Where enrichment had been attempted, the link between source data, confidence, and the final record was not consistently auditable.

The team needed a way to systematically complete HCP profiles from sources the company already trusted, extend to external sources only where necessary, and keep every change explainable.

Five-stage HCP enrichment flow: a fragmented record moves through discover, match and extend to become a trusted record
Discover, match, then extend only where internal sources fall short – each stage narrowing the gap between a fragmented record and a trusted one.

Our Solution

Chryselys designed a four-stage enrichment pipeline that starts with data the company already owns and reaches outward only where internal sources fall short.

A systematic approach from discovery to a validated, audit-ready record:

Discovered high-value sources

Assessed internal transactional sources – including PubMed, Informa, and Voice of Customer – alongside the existing data lake to identify where each source added enrichment value.

Matched against internal transaction data

Linked each HCP to internal transaction records using full name as the merge key, applied separately for every target attribute.

Extended through GenAI web enrichment

Where internal data left gaps, used GenAI-driven web scraping – targeted sites first, generic search second – to complete the picture.

Scored, validated, and loaded

Ran every candidate through a confidence-scoring decision box weighing merge-attribute match strength, recency, and frequency, with an LLM resolving ties for strong matches only. Potential matches went to data stewards for review; weak matches were stored for audit but never applied. Validated records were written back to Reltio with a full audit trail.

Business Impact

The enrichment approach consistently outperforms transactional matching alone, with affiliation, phone, and email match rates landing above 99%, and specialty match above 91%.

Closed the largest data gaps as affiliation, phone, and email match rates consistently land above 99%, with specialty match above 91%.

Outperformed transactional-only matching on every measured attribute, more than doubling the share of profiles meaningfully enriched.

Kept manual review targeted, with review load held under 10% of enriched records, reserved for the confidence band that genuinely needs steward judgment.

Strengthened governance through confidence scores, stored reasoning, and full audit trails for every enrichment decision.

Improved downstream engagement potential by giving segmentation, personalization, and omnichannel campaigns a more complete and trustworthy record to work from.

Scaled a proven data onboarding approach across 30+ countries, enabling the onboarding of 100,000+ records in 48 hours and reducing duplicate records and the manual effort needed by 90%

Created a reusable enrichment asset, ready to extend market by market as new pilots roll out.

Every insight can lead to meaningful impact

Whether you’re exploring a new opportunity, addressing a complex challenge, or rethinking what’s possible with data and AI, we’re here to help.

Other Success Stories