Match and Merge Rules in Informatica MDM: A Complete Hands-On Guide

One of the most powerful capabilities of Informatica MDM is its ability to identify duplicate records and merge them into a single golden record. This process is driven by Match and Merge rules.

If you’re preparing for real-time projects or interviews, mastering Match & Merge in Informatica MDM is essential. In this tutorial, we’ll walk through the concepts, configuration steps, and practical examples.

What are Match and Merge Rules?

• Match Rules: Define how MDM determines whether two records represent the same entity (e.g., same customer).
• Merge Rules: Define how duplicate records are consolidated into one trusted record (golden record).

Example:

• Two records:
  • John Smith, Email: john.smith@gmail.com
  • J. Smith, Email: john.smith@gmail.com
• Match Rule: Same email + similar name.
• Merge Rule: Pick the most trusted data from both and create a golden record.

Key Concepts to Understand

1. Tokenization

MDM breaks down data into tokens for comparison (e.g., “John Smith” → “John”, “Smith”).

2. Trust and Survivorship

• Trust = priority of source system (e.g., CRM > ERP > Excel).
• Survivorship = how final golden record values are selected (highest trust wins, most recent update wins, etc.).

3. Match Types

• Exact Match – Example: Email ID must be the same.
• Fuzzy Match – Example: Names like “Jon” and “John” are considered similar.
• Probabilistic Match – Scores records based on similarity.

Step-by-Step: Configuring Match and Merge in Informatica MDM

Step 1: Create Match Columns

1. Open Hub Console → Schema → Match Columns.
2. Select the columns (e.g., Email, FirstName, LastName, Phone).
3. Define match key types: Exact / Fuzzy / Phonetic.

-- Example: Match Column Setup
Column: EMAIL_ADDRESS
Match Type: Exact
Column: LAST_NAME
Match Type: Fuzzy (Edit Distance: 2)

Step 2: Define Match Rules

1. Navigate to Match Rule Set in Hub Console.
2. Create a new Match Rule with logical conditions.

Example Rule:

• IF (EMAIL = Exact Match) OR (FIRST_NAME + LAST_NAME = Fuzzy Match)
• THEN consider records as duplicates.

Match Rule Example:
(MatchColumn.EMAIL = Exact) OR (MatchColumn.FIRST_NAME = Fuzzy AND MatchColumn.LAST_NAME = Fuzzy)

Step 3: Configure Merge Settings

1. Go to Merge Settings.
2. Choose how records will merge:
   • Auto Merge: System automatically merges based on rules.
   • Manual Merge: Requires Data Steward approval.
3. Define Trust Levels per source system.

-- Example Trust Settings
CRM: 90
ERP: 80
Excel Import: 50

Result: If the same field is present in CRM and ERP, MDM picks the CRM value (higher trust).

Step 4: Run Match and Merge Job

• Execute Match Job from Hub Console.
• Review Matched Sets in the Match Results.
• Execute Merge Job to create Golden Records.

Example: Customer Data Deduplication

Customer ID	Name	Email	Source
CUST101	John Smith	john.smith@gmail.com	CRM
CUST202	J. Smith	john.smith@gmail.com	ERP
CUST303	Jon Smyth	john.smith@outlook.com	Excel

After Match & Merge

• Golden Record:
  • Name: John Smith (from CRM – highest trust)
  • Email: john.smith@gmail.com
  • Source Systems Linked: CRM, ERP, Excel

Best Practices for Match & Merge

• Always profile data quality before configuring match rules.
• Use standardization (via IDQ) for names, addresses, phone numbers.
• Start with conservative rules, expand gradually to avoid false merges.
• Monitor match scores and thresholds to fine-tune accuracy.
• Keep audit logs for merge history (important for compliance).

Common Issues Developers Face

• Over-merging: Too loose match rules causing unrelated records to merge.
• Under-merging: Too strict rules leaving duplicates unmerged.
• Performance Lag: Poor indexing or complex fuzzy logic slows jobs.
• Trust Conflicts: Wrong survivorship when trust levels are misconfigured.

FAQs

Q1. What happens if two records match but trust scores are equal?
Informatica applies survivorship tie-breakers (e.g., most recent update, highest source priority).

Q2. Can we undo a merge in Informatica MDM?
Yes, Data Stewards can unmerge records if needed, restoring original records.

Q3. What is a typical threshold for fuzzy matches?
Usually between 0.7–0.9, depending on data sensitivity.

Q4. How do I test match rules before applying to full dataset?
Use Sample Match option in Hub Console with small test datasets.

Q5. Is Match & Merge different in Customer 360 vs Product 360?
The underlying process is the same, but attributes and survivorship logic differ based on domain.