Glossary

Data stewardship

Data stewardship is the governance and day-to-day care of data to keep it accurate, usable, consistent, and controlled.

Data Integration, Security and Trust Data Integrity, Version Control and Audit Trails MES, ERP, PLM and Data Integration

Data stewardship commonly refers to the governance and day-to-day care of data so it remains accurate, consistent, usable, and appropriately controlled over time. It includes defining who is responsible for data quality, metadata, access rules, issue resolution, and lifecycle handling across systems and processes.

In manufacturing and regulated operations, data stewardship often applies to master data, transactional records, quality data, equipment data, specifications, and traceability records that move between systems such as MES, ERP, PLM, QMS, LIMS, and reporting tools. It is not the same as data ownership, database administration, or cybersecurity alone, although it works alongside each of those functions.

What it includes

Defining data elements, business meaning, and acceptable values
Maintaining data quality rules and resolving data issues
Managing metadata, naming conventions, and reference data
Clarifying responsibilities for creation, review, change, and archival
Supporting controlled use of data across workflows, reports, and integrations

How it appears in operations

Operationally, data stewardship shows up in activities such as keeping part masters aligned between ERP and MES, maintaining approved reason codes for downtime or nonconformance, reviewing duplicate supplier records, or ensuring genealogy and batch data are captured in a consistent format. In regulated environments, stewardship also supports reliable evidence trails by reducing ambiguity, version drift, and inconsistent record handling.

Common confusion

Data stewardship is often confused with data governance. Data governance is the broader framework of policies, decision rights, and oversight. Data stewardship is the operational role or practice that applies those rules to specific data domains.

It is also different from data custody or IT administration, which focus more on technical hosting, storage, backup, and infrastructure. A steward is typically concerned with business meaning, quality, and proper use of the data.

You do not validate a process drift model by showing good accuracy on historical data alone. In a customer-regulated aerospace program, validation usually requires a documented intended use, verified data lineage, measurement-system adequacy, change control, and evidence that model outputs are bounded, reviewable, and compatible with existing quality and production controls.

Can I build AI models directly on my MES database without a data warehouse?

Yes, in some cases, but it is usually a poor default for regulated manufacturing. Directly training on an MES database can create performance, data quality, traceability, and validation problems. It may work for small read-only use cases, but production AI typically needs a governed data layer, whether that is a warehouse, lakehouse, or curated replica.

What is the relationship between configuration control and engineering change management?

Configuration control and engineering change management are tightly linked but not identical. Configuration control defines and governs the approved product and process baselines over time, while engineering change management is the formal workflow for proposing, assessing, approving, and implementing changes that alter those controlled configurations. In regulated, long-lifecycle manufacturing, both must integrate with PLM, ERP, MES, and QMS, and be supported by traceable, validated workflows rather than ad hoc changes, to avoid gaps between design intent, released documentation, and actual as-built/as-maintained records.

How do we ensure MES data is trusted for KPI reporting?

You cannot “declare” MES data trustworthy for KPI reporting; you have to design, prove, and continuously maintain that trust. This requires clear KPI definitions, data lineage and traceability, controlled integrations, validation of calculations, tight change control, and ongoing reconciliation against physical reality and other systems.

Related Glossary

Daylight savings time (DST)

A seasonal clock change that shifts local civil time, affecting timestamps, schedules, and system synchronization.

Data profiling

Data profiling is the analysis of data structure, content, and quality to understand what data contains and how reliable it is.