Skip to main content

Healthcare · 16 weeks · A healthcare provider

Unified customer data platform

4 CRMs + 3 billing platforms → one source of truth. 87% duplicates resolved. Privacy-by-design with full GDPR audit trails.
Client
A healthcare provider
Sector
Healthcare
Duration
16 weeks
Region
DE / EU
Year
2026

Client name anonymized under NDA. Industry, technical approach, tools, and measured outcomes are reported as-is. Named references available on request.

# RESULTS

Duplicates resolved
87%
Source of truth
1
End to end
16 wk
Privacy-by-design
GDPR

# THE-CHALLENGE

4 CRMs, 3 billing platforms, no unified view. Every department had its own version of the patient — and none of them agreed. Marketing sent duplicate mailings. Billing couldn't reconcile accounts. Clinical teams worked from partial records.

The root cause wasn't just technical fragmentation — it was organizational. Each department had chosen its own system over the past decade. Nobody had authority to merge them, so they'd built point-to-point integrations that silently broke and produced conflicting data. A patient could have three different addresses across four systems, and nobody knew which was current.

# THE-TRANSFORMATION

// before

  • 4 CRMs and 3 billing platforms, none of them talking to each other
  • Marketing sent duplicate mailings, billing couldn't reconcile
  • Patient #10042 in CRM ≠ Patient #10042 in billing
  • GDPR compliance was "we think we're fine"

// after

  • Single source of truth across all 7 systems
  • 87% of duplicate records identified and merged
  • Full GDPR audit trail — compliance is documented, not assumed
  • Downstream systems pull from unified platform via API

# OUR-APPROACH

Data archaeology

Audited all 7 source systems. Mapped entity relationships, identified overlap, and cataloged every field that contributed to the "who is this patient?" question.

Identity resolution

Built a probabilistic matching pipeline in Python — not just name + DOB, but address history, contact patterns, and billing identifiers. Validated against a 500-record manually verified sample.

Platform build

Unified data layer in Snowflake with pseudonymization, consent tracking, and full GDPR audit trails. Fivetran for ingestion, dbt for transformation, Terraform for infrastructure.

API & handover

Created API endpoints so downstream systems pull from the unified platform instead of each other. Documented everything. Trained internal team to maintain the pipeline.

Three of the seven systems had overlapping patient IDs in different formats. The worst case: two systems used the same ID field name but assigned them from different sequences. Patient #10042 in the CRM was a completely different person than Patient #10042 in billing. We caught it during validation — if we hadn't, the unified platform would have merged two strangers' medical records.
— From the engagement

# TECH-STACK

SnowflakedbtPythonFivetranTerraformAPI integrationsPrivacy-first architecture
We had several systems that never really fit together. Juri and Oleks guided us through the project step by step and involved data protection and the business units from the very beginning. In the end we had a solution everyone could live with.
— Head of DigitalizationHealthcare provider

# OPERATING-CONTEXT

Operating constraint

Patient data meant zero tolerance for record mis-merges — a false positive in identity resolution could mean combining two strangers' medical records. Every matching rule required sign-off from the compliance lead. GDPR audit trails had to be in place before a single record was merged, not retrofitted after.

Adoption & rollout

Downstream systems (CRM, billing, clinical) were re-pointed to pull from the unified platform via API instead of from each other — no big-bang cutover. Rollout sequenced by department over 4 weeks with rollback paths at each step. Internal data team took over ongoing pipeline ownership 4 weeks before the engagement ended; we stayed on call through the first month of independent operation.

Common questions about this project

  • How do you unify customer data across systems without wrong merges?
    Conservative identity resolution: every matching rule was signed off by the compliance lead, validated against known cases, and biased toward not merging when in doubt — in healthcare, a false-positive merge means combining two strangers' records. 87% of duplicates were resolved; the remainder stayed deliberately separate rather than risk a bad merge.
  • How does GDPR fit into a customer data platform?
    Privacy by design, in the literal order of work: audit trails were in place before the first record was merged, not retrofitted after. Every merge is traceable — who, when, by which rule — which is exactly what a supervisory authority or an internal audit asks for.
  • How do you migrate downstream systems without a big-bang cutover?
    The CRM, billing and clinical systems were re-pointed one by one to read from the unified platform via API, sequenced by department over four weeks with a rollback path at each step. The internal team took over pipeline ownership four weeks before the engagement ended.

Have a similar challenge?

A 30-min call. We'll tell you within 24h whether we're the right fit.

 book-call

// or write: hello@saloid.com · gräfelfing · de