The transition to decentralized architectures and AI integration places new demands on corporate data quality. For large enterprises with fragmented data across ERP, CRM, and ECM systems, the main challenge is often not the technology stack, but the lack of a clear organizational structure. Without defined roles for data owners and data stewards, establishing stable Master Data Management (MDM) is impossible.
Organizational challenges and the Data Mesh concept
Attempts to resolve reference data inconsistency solely through software without role allocation lead to broken data schemas. While using an API Gateway helps manage transport and access, it does not solve issues with the logical quality of information.
Data Mesh architecture proposes shifting data responsibility directly to business domains. The "Data as a Product" concept implies having a dedicated owner, defined SLAs, and clear data contracts that prevent unpredictable failures during system integration.
Division of responsibility: Data Owner and Data Steward
Successful MDM implementation relies on a clear distinction between strategic and tactical levels of governance:
- Data Owner — a business representative or domain leader who owns data as an asset. They are responsible for the financial and operational consequences of data quality, and approve business rules, glossaries, and SLAs.
- Data Steward — a specialist who provides day-to-day control. They translate business requirements into technical specifications, monitor quality metrics, and configure validation schemas.
To minimize risks, especially before feeding data into AI models, it is advisable to use international standards such as the NIST AI RMF 1.0 framework, which helps identify data lineage and establish accountability.
Automating Data Governance rules
Organizational regulations require technical implementation. System integration from Intecracy Group based on the UnityBase low-code platform allows automating master data management through the following built-in mechanisms:
- Domain metadata: describing the domain model through unified metadata for automatic REST API generation and operation under a single technical contract.
- Row-level security (RLS): flexible configuration of access rights down to the level of individual records or attributes.
- Audit trail: automatic logging of any changes in master data using the DataHistory tool.
Combining business responsibility for data with technical control and automated data contracts allows enterprises to build a predictable and resilient IT landscape.
What changes for the sector
For modern enterprises, neglecting organizational data governance leads to broken data schemas, integration failures, and compromised AI initiatives. Without clear ownership, businesses face severe operational and financial risks due to poor data quality, making it impossible to scale decentralized architectures like Data Mesh effectively.
Next steps
- Establish clear roles: Appoint Data Owners to take financial and operational responsibility for data assets, and Data Stewards to manage day-to-day quality control and technical validation.
- Adopt international standards: Use frameworks like NIST AI RMF 1.0 to trace data lineage and secure accountability before feeding data into AI models.
- Automate governance: Implement low-code solutions like the UnityBase platform from Intecracy Group to automate master data management through domain metadata, row-level security, and automatic audit logging.
Prepared by a Software Ukraine member. Original publication.