Understanding How Match Scores Works

What “Match Score” Means:

For each potential pair, Reltio evaluates your positive match rules. Each rule contributes two numbers you configure: scoreStandalone (the base score if that rule matches) and scoreIncremental (extra points added when other rules also match). The Match Score is calculated as: take the highest scoreStandalone among matched rules, then add the scoreIncremental from every other matched rule that fired. This score helps stewards prioritize review; it does not trigger merges by itself—your rule types/severities and any negative rules govern auto-merge behavior.

Quick Example

If you have:
    •    Rule A (Email exact): 70 standalone, 30 incremental
    •    Rule B (Name+Address fuzzy): 60 standalone, 25 incremental
    •    Rule C (Phone+Name): 50 standalone, 20 incremental

When B and C match: score = max(60,50) + 20 = 80.
When A, B, and C match: score = 70 + 25 + 20 = 115.
Higher scores suggest stronger overall evidence, but merges still follow your configured severities and negative rules.

Important Nuance: Relevance-Based Matching:

If you use Relevance-based matching, the UI shows a Relevance score (0–1) produced by attribute comparators. That is a different mechanism from the standalone/incremental scheme above. Don’t mix or compare these numbers across approaches.

Practical Guidance:

Assign a higher scoreStandalone to strict, high-precision rules (e.g., exact email or ID). Use scoreIncremental to reward corroborating evidence from different attributes. Avoid “weaker copies” of existing rules (e.g., a version missing one attribute), which can inflate scores without adding new signal. Treat Match Score as a prioritization tool for stewarding queues; keep merge control in rule severities/negative rules. If you need targeted queues, the Potential Matches API supports filtering by Match Score to focus review on the ranges that matter.