Match rules are the heart of Identity Resolution. They define which field combinations identify records as belonging to the same person. In this article, we'll explore how matching works and strategies for effective rule design.

How Matching Works

Identity Resolution uses a cascading match approach:

All records from all sources are collected
Rule 1 is applied—records that match are grouped together
Remaining (unmatched) records move to Rule 2
This continues through all rules
Records that don't match any rule remain as separate identities

📘
Note
Once a record matches a rule, it stops processing. This is why rule order matters—place your highest-confidence rules first.

Anatomy of a Match Rule

Each match rule specifies:

Component	Description
Rule Number	The order in which this rule is evaluated (1, 2, 3...)
PII Fields	The fields that must match for records to be grouped

For example, a rule using Email + First Name + Last Name means: "Group records together if they share the same email AND first name AND last name."

Designing Effective Rules

Start Strict, Then Broaden

Your first rule should be your most confident match:

Rule	Fields	Confidence
1	Email + First Name + Last Name	Highest
2	Phone + First Name + Last Name	High
3	First Name + Last Name + Postal	Medium
4	Email only	Lower

This approach ensures:

High-confidence matches happen first
Broader rules only catch what stricter rules missed
You minimize false positives (incorrectly merging different people)

Choose Fields Wisely

Different fields have different matching characteristics:

Field	Uniqueness	Data Quality	Notes
Email	High	Usually good	Best single identifier, but people have multiple emails
Phone	High	Variable	Formatting differences can cause misses
First + Last Name	Low alone	Good	Common names cause false positives; combine with other fields
Postal Code	Low	Good	Too broad alone; useful as a supporting field
Address	Medium	Variable	Formatting variations can cause misses

Understanding Match Results

After running Identity Resolution, the Summary tab shows how your rules performed.

Current Status

Metric	Description
Source Records	Total records processed across all source tables
Golden Records	Number of unified profiles created
Distinct Match Configurations	Number of match rules applied

Matched Records On

The bar chart shows how records matched across your rules:

Each rule displays its count and percentage (e.g., Rule_1: 21 records, 36.21%)
Unmatched records didn't match any rule and remain as individual profiles
Colors help visualize the distribution across rules

Data Quality

This section highlights potential issues in your data:

Metric	What It Means
Duplicate Emails	Records with the same email found across sources
Conflicting Records	Records where sources have different values for the same field

The Conflicted Data in Columns table shows which fields have conflicts and how many — useful for identifying data quality issues upstream.

Interpreting Your Results

High Rule 1 matches — Your strictest rule is effective
Many unmatched records — Consider adding broader rules
High conflict counts — Review source data quality or adjust precedence

Troubleshooting

Too Many Unmatched Records

If most records aren't matching:

Check PII coverage — Do your sources actually share common fields?
Review data quality — Are there formatting differences (e.g., "[email protected]" vs "[email protected]")?
Add broader rules — Consider rules with fewer required fields

📘
Note
Identity Resolution normalizes common variations (case, whitespace), but significant formatting differences may need data cleaning upstream.

Suspicious Merges

If unrelated people are being merged:

Tighten rules — Add more required fields
Check for data issues — Placeholder values like "[email protected]" can cause false matches
Review source precedence — Ensure high-quality sources are prioritized

Missing Expected Matches

If records you know belong together aren't matching:

Verify field mapping — Are the PII fields mapped correctly?
Check for nulls — Records with null values in match fields won't match
Review rule coverage — Do your rules cover this combination of fields?

Ready to understand your results? Continue to Output Tables to learn about the Final Identity Table and other outputs.