Skip to main content

Merge

The Merge module combines matched records into golden records, using survivorship rules to decide which values become the trusted master view.

Why Merging Matters​

Once records match, which values survive: the most recent address or the most complete? Configurable survivorship rules decide.

What You Can Do​

Define Survivorship Rules

Choose which values survive by recency, completeness, source priority, or custom logic.

Review Before You Commit

See the suggested golden record beside every source record before anything is written.

Handle Conflicts

Disagreeing fields are highlighted, and alternative values are recorded for audit.

Maintain Lineage

Track which source records fed each golden record.

Survivorship Strategies​

Each semantic attribute mapped from two or more tables can have its own strategy on the Merge rules page.

Not Prioritized

The default: take the first non-null value

Source Priority

Order tables so the highest-priority source wins

Most Recent

The most recently updated value wins

Best Match

The value from the highest-scoring match wins

Key Capabilities​

Field-Level Survivorship​

Use different strategies per field, such as recency for addresses and source priority for clinical data.

Golden Record Creation​

Produce one master record per entity with the best values from all sources.

Bidirectional links between golden and source records show where data came from.

The Review Queue​

Clusters below the auto-merge threshold wait as merge candidates. A steward compares source records, the score, why they matched, and the proposed golden record, then approves, rejects, or splits the cluster. Approvals can be undone, every decision requires a comment, and candidates can be approved in batches. Single-record clusters are approved automatically.

Golden Record Provenance​

Each golden record records how it was created (auto-merge, approval, rejection, or split), who approved it, and its source candidate.

Consistency Safeguards​

Golden records are written before approvals are recorded, and a reconciler heals any drift. Merge won't start while Match is still running on the same profile.

Real-Time 360° Views​

Search merged golden records on the Profiles page, where they are also available to downstream systems.

Runs Where Your Data Lives​

Merge runs natively on Databricks, Snowflake, or Microsoft Fabric, writing all golden records into your own catalog or lakehouse.

Business Impact​

Single Source of Truth

One authoritative record per patient, provider, or member.

Data-Driven Decisions

Trustworthy data for confident decisions.

Operational Efficiency

Stop reconciling conflicting data by hand.

Who Benefits​

  • Clinical Teams: Complete, accurate patient records
  • Executive Leadership: Trusted dashboards and reports
  • IT Teams: A single API endpoint for master data
  • Compliance Teams: Proven accuracy and lineage for audits
  • Analytics Teams: Reliable models on golden records

The MDM Journey Complete​

Merge completes the skyMDM pipeline:

  1. Connect - Bring data together
  2. Clean - Standardize and correct data
  3. Validate - Meet quality thresholds
  4. Enrich - Add context and attributes
  5. Map - Apply semantic labels
  6. Match - Identify related records
  7. Merge - Create golden records

The result is a unified, trustworthy view of your healthcare data.