clearMDM Focus Areas

Blocking Keys. Are composed field values used to create coarse groupings of records to which fine grained matching rules are applied.

Blocking Key Input Fields. Should be stable identifiers with a high % of accurate population across the dataset.

For example, the 2 records below may be assigned the key ACMEUK based on the normalised Company Name and Country.

Acme Inc, 199 North Ave, London, WC2N 5AE, UK (ACMEUK)

Acme Ltd, North Ave, London, Great Britain (ACMEUK)

Blocking Key Configuration.
- Blocking Key Settings [Target Object]
- Multiple Blocking Keys
Blocking Key Match Value. The complete Blocking Key value can be used to group records for matching, or a set number of characters.

SMITHJOHN = Blocking Key (LastName:5+FirstName:5) == 10 chars.

SMITHJ = Blocking Key Match Value == 6 chars.

Increased Sensitivity. The shorter the key value the larger the groupings become which increases sensitivity (actual matches found across the dataset) at the cost of processing efficiency.
Dynamic Adjustment. clearMDM dynamically increases the number of characters to ensure processing is within the execution limits imposed by the Salesforce platform.

Matching Rules. Identify matching records using logic which reflects the dataset characteristics.

Matching Settings [Target Object]
- Matching Rules & Cross Field Matching
- Matching Thresholds
- MDM Status and [Is Active for Matching?]
Data Sources and Matching Settings [Data Source]
Data Stewardship Conditions [Target Object]
Matched Record Pairs

UI Tools.
- Matching Test
- Potential Matches
- Find Matches & Match Analysis
- Data Stewardship UI

Fuzzy Match 2 Records. Calculate a record level fuzzy match score based on how similar each attribute is and their relative weighting.

If Record1 or Record2 value is null add Null Score to the Running Total
Calculate Edit Distance between Record1 and Record2 values

Edit Distance == number of character changes required to align values

Maximum Edit Distance == maximum possible number of changes required

Deterministic Matching Rules. Prevent matches between records where attribute-based conditions are not met.

E.g. Gender, Brand (ring fenced data) or Market (staggered rollout)

Partition Data Sources. Records with a low quality score can be dynamically assigned to an inactive data partition.
Rejected Record Pairs. Prevent future matches between specific records, typically following a Reject decision by a Data Steward.

Matching Rule Validation. It is recommended that matching rules are validated empirically using the Matching Test function with diverse high/mid/low score examples. Matching Statistics can be used for global validation (i.e. across the dataset).
Iterative Improvement. Rules are managed via the Settings UI and should be validated and refined over time as the dataset characteristics evolve.
MDM Synchronisation. Can be used to maintain stable groups where records may no longer match due to rule changes.

Practical Considerations.
- Rule changes are not applied retrospectively.
- Whenever rules are modified by adding or removing a referenced field the related Data Sources must be re-saved to refresh field mappings.
- Object and field level permissions are strictly checked by MDM operations.

Rollback

Merge Results. Each time a new group is formed or updated a Merge Result is recorded for the Master Record.
Merge Timeline. Merge Results form a timeline of merge events for the group and support:
- Data lineage visualization.
- Field level restore.
- Point-in-time rollback.
Data Retention. Merge Result data is managed by data retention policy.

To find out more about clearMDM please do not hesitate to contact us at hello@clearMDM.com

Article last reviewed: 2025-01-05

Feature Quick Start Guides