(Advanced) Rule Evaluation Order

This article explains in detail how MatchLogic evaluates Master Record Rules and Overwrite Rules internally. Understanding the evaluation model helps you design rule sets that produce predictable, correct golden records.

Master Record Rules: Win-Count Model

Master determination uses a win-count model across all active rules:

1. For each active Master Record Rule, the system evaluates every record in the group against the rule's operation (Longest, Max, Most Popular, etc.).
2. The record with the best value for that field is awarded one win.
3. After all rules have run, each record in the group has a win count.
4. The record with the highest win count is designated the master.

Example: You have five active Master Rules. Record A wins three of them (FirstName longest, Email present, City most popular). Record B wins two (LastName longest, Phone present). Record A becomes the master with 3 wins.

Tie-Breaking

When two records have equal win counts, the tiebreaker is record position within the group. The record that appears earliest (by original data source load order) wins the tie. To avoid relying on load order, add a deliberate tiebreaker rule — for example, a rule on a timestamp field using the Max operation to prefer the most recently updated record.

Rule Order Has No Effect on Win Counts

For Master Record Rules, the order of rules in the list does not affect the outcome. Each rule independently contributes one win to the tally. Changing rule order changes nothing about which record becomes master — only the number of active rules and their operations matter. See Activating and Ordering Rules for more.

Overwrite Rules: Sequential Evaluation

Overwrite Rules are evaluated in a strictly sequential, top-to-bottom pass. Each rule is independent:

1. Rule 1 runs. If its condition is met, the master's field value is overwritten.
2. Rule 2 runs. If its condition is met, it overwrites the value (potentially overwriting what Rule 1 just wrote).
3. This continues for all active rules in order.

Order Matters for Overwrite Rules

Unlike Master Rules, overwrite rule order does affect the final output. Consider this scenario:

• Rule 1 (top): Email field — From Best Record, No Condition. Writes the best-scoring record's email.
• Rule 2 (below): Email field — From Master, Field Is Empty. Only writes the master's email if the field is empty.

Because Rule 1 runs first and always writes a value (No Condition), Rule 2's condition (Field Is Empty) is never met and Rule 2 has no effect. Reversing the order would change the outcome — Rule 2 would write the master's email if empty, and Rule 1 would then potentially overwrite it unconditionally.

Interaction Between Master and Overwrite Phases

The two execution phases are independent jobs:

• Master determination runs first and designates a master for each group.
• Overwrite rules then read those master designations and the current golden record values to apply field-level corrections.

If you re-run Master Rules (e.g., after modifying them), master designations may change. You should then re-run Overwrite Rules as well so that From Master and From Best Record operations reference the updated masters.

Data Source Filtering

Both rule types allow you to restrict which data sources participate in the evaluation. Records from excluded sources are not considered when calculating wins (Master Rules) or when sourcing overwrite values (Overwrite Rules). This effectively gives you per-source priority control without needing a separate rule per source combination.