You are here:
Filter Source Records out of Identity Resolution
Control which source records enter the identity resolution processing pipeline by defining filter conditions on your ruleset's primary data model object. Filters act as a preprocessing gate, so records that match your exclusion conditions never enter the unification pipeline. Using filters helps keep unified profiles clean and reduces credit consumption.
Filters are intended for broad, partition-level segregation. They apply to the primary data model object of a ruleset and aren’t designed for record-level control on high-cardinality fields such as IDs.
Use Cases
- Separate B2B and B2C accounts—If your data space includes both business and consumer accounts in the same Account DMO, add a filter such as RecordType = "B2B" to your B2B ruleset. Business accounts stay out of consumer unification, and consumer accounts stay out of B2B unification.
- Run a sample dataset to test match rules—Before running identity resolution on your full dataset, filter by data source to process only a representative sample. Evaluate your match and reconciliation rule results, refine the configuration, and then remove the filter to process the full dataset.
- Isolate verified data sources—If you have a high-velocity or lower-quality data source alongside highly verified CRM data, use a filter to include only the verified source in a given ruleset.
Limits and Considerations
Each ruleset supports up to 3 filter groups with up to 3 field conditions per group. Each filter group applies to the primary data model object of the ruleset.
Use AND/OR logic to combine conditions within and across groups. Standard operators are available for building conditions.
Adding or changing a filter condition on an active ruleset triggers a full reprocessing of the ruleset. To minimize the number of full reruns, pause the ruleset's automated schedule, make all configuration changes, including filter changes, at once, and then resume the schedule.
