Intelligence · Module
Master Data Management
One golden record for each customer, supplier, product and employee
Master Data Management is planned to link source records from every system to a golden record, using deterministic and fuzzy matching, field-level survivorship and steward review. Merges and unmerges are compensatable Actions with an undo token.
Why duplicates cost so much
When one customer appears five times across CRM, Billing, Service Desk and an imported ERP, every report double-counts and every agent works on a partial picture. A light version, MDM-lite for contacts and companies, runs in the Platform from W1. The full Master Data Management add-on is planned for W4.
Matching that understands Indian data
Source records from One3 apps, Data Fabric datasets and connectors are linked to a golden record with cross-reference ids. Match rules combine deterministic keys (E.164 phone, email, domain, GSTIN, masked PAN, SKU, HSN/SAC plus name) with ML fuzzy matching using blocking keys and a confidence score, including transliterated Indic names. Identifiers used for matching are hashed with a tenant-salted key before storage.
Survivorship and stewardship
Survivorship works field by field from the source-of-truth configuration, with primary and secondary sources and conflict policies. For consent, the most restrictive value wins. Proposed merges, splits and conflicts go to steward task queues with SLAs and assignment by domain. Automatic merging is limited to the exact-match policy; probable matches always go to a person. Vendor and employee matches need finance or HR approval.
Safe merges
Merge and unmerge are compensatable Actions. They re-point role rows and relations in one transaction and return an undo token. Bulk match jobs run on imports, migrations and new connectors with dry-run statistics before anything changes.
Beyond duplicates
Hierarchies cover corporate groups, parent and subsidiary relationships and legal entities. Reference data management covers countries, Indian states and union territories, currencies, HSN/SAC codes and units (authoritative sources to be verified). Golden ids can be published to external systems with reverse lookups through the API. Quality dashboards show the duplicate rate, match precision measured from steward decisions and completeness per domain. External verification of identifiers such as GSTIN through a licensed provider is planned later.
Features
What Master Data Management includes
Source records and cross-references
Records from apps, datasets and connectors linked to a golden record.
Deterministic match keys
E.164 phone, email, domain, GSTIN, masked PAN, SKU and HSN/SAC plus name.
ML fuzzy matching
Blocking keys and confidence, including transliterated Indic names.
Field-level survivorship
Primary and secondary sources with conflict policies; most restrictive wins for consent.
Steward task queues
Proposed merges, splits and conflicts with SLA and domain assignment.
Compensatable merge and unmerge
One-transaction re-pointing of roles and relations with an undo token.
Bulk match jobs
Dry-run statistics on imports, migrations and new connectors.
Quality dashboards
Duplicate rate, match precision from steward decisions and completeness per domain.
Hierarchies
Corporate groups, parent and subsidiary, legal entities.
Reference data
Countries, Indian states and UTs, currencies, HSN/SAC codes and units.
Golden id publication
Cross-reference sync and reverse lookups via API.
Vendor and employee matching
Matches that need finance or HR approval.
Workflow
Match, steward, survive
- 1
Source records arrive from apps, datasets or connectors
- 2
Match rules score candidate pairs
- 3
Exact-policy matches merge; probable matches go to a steward
- 4
The steward approves, splits or rejects
- 5
Survivorship builds the golden record field by field
- 6
The merge runs as a compensatable Action with an undo token
AI
AI skills and Digital Employees
Fuzzy matching
Scores probable duplicates including transliterated names; people decide probable matches.
Survivorship conflict explanation
Explains why a field value won.
Threshold tuning
Proposes match threshold changes from steward decisions for human approval.
Digital Employees that use this module
They act through the same governed Actions as people; risky steps wait for approval.
Frequently asked questions
Is there any MDM before W4?
Yes. MDM-lite for contacts and companies is planned to run in the Platform from W1.
Can a merge be undone?
Yes. Merges are compensatable Actions with an undo token.
Are identifiers like PAN stored in plain text?
Match keys are hashed with a tenant-salted key before storage, and PAN is used in masked form.
See Master Data Management in your workflow
Tell us about your team and the outcomes you want. We will walk you through the apps, the Digital Employees and the plan for your industry.