Merger

Purpose

The merger resource defines how to create golden records.

Configuration

Merge Types

Merge Type	Description	Use Case
RANDOM	Random selection by ID hash	No preference
CONSTANT	Select record with specific value	Trust specific source
DATE	Most recent record wins	Prefer latest data
NUMBER	Highest number wins	Use numeric priority
SCRIPT	Custom weight calculation	Complex logic

Merge Sort

HIGHEST_WEIGHT - Prefer values from highest-weighted records
LOWEST_WEIGHT - Prefer values from lowest-weighted records

Merge Behavior

Single-value columns:

Select value from highest/lowest weighted record
Based on merge sort strategy

Array columns:

Union all unique values from all records
Preserves all data

Nested datasets:

Merge based on identity matching
Recursive merge logic

Example Merger Configuration

JSON

{
  "type": "merger",
  "id": "customer_merger",
  "dataset": "customer_dataset",
  "mergeType": "DATE",
  "mergeSort": "HIGHEST_WEIGHT",
  "weights": [
    {
      "column": "email",
      "weight": 2.0
    },
    {
      "column": "full_name",
      "weight": 1.5
    }
  ]
}

Best Practices

Trust recency - Use DATE merge type for frequently updated data
Consider source - Use CONSTANT to prefer specific sources
Array preservation - Arrays automatically preserve all values
Test merges - Review golden records for quality