Data governance is not what you think it is
Data governance has acquired an image problem. For many manufacturers, the term still suggests policies, controls and approval processes that slow work down without producing an obvious operational benefit. Yet the growing use of AI, advanced analytics and connected systems is exposing the cost of treating governance as an administrative obligation rather than a core part of how the business operates.
Malgorzata Samuel, Operations Data Standards Manager at BAT, argues that this narrow perception reflects how governance has traditionally been presented. “Data governance is fundamentally about establishing clear ownership, accountability, standards, and processes across the entire data lifecycle,” she says. “Its objective is not to introduce bureaucracy, but to enable consistent, reliable, and business-ready data.”
Governance must begin at source
Manufacturers often attempt to improve data quality at the point of consumption. Analysts correct reports, data teams reconcile conflicting definitions and employees build workarounds around incomplete information. While these interventions may resolve an immediate problem, they leave the process that created it unchanged.
Samuel believes manufacturers need to move their attention further upstream. “In a modern manufacturing organisation, effective data governance extends well beyond the point of data consumption,” she says. “A key success factor is master data governance at source. Too often, organisations focus on managing and correcting data issues in downstream outputs such as reports and analytics, rather than addressing the root causes within operational systems and processes.”
This requires governance to become part of the process through which data is first created, maintained and changed. Employees responsible for product, customer, supplier, asset or production information need clear standards and ownership, while the systems and workflows they use must make correct data creation easier than correction after the event.
“Embedding master data governance at source is one of the most effective ways to drive meaningful data ownership within the business,” Samuel explains. “When accountability sits with those who create and maintain the data, quality improves as a natural outcome of well-defined processes, rather than requiring ongoing remediation.
“Ultimately, effective data governance should be seen as an enabler of operational excellence. When implemented well, it supports better decision-making, improves efficiency, and ensures that data can be trusted across the organisation—moving it firmly away from a perception of bureaucracy towards a strategic business capability.”
AI is raising the cost of weak governance
The consequences of poor data have always been present, but AI magnifies them. An inaccurate field in a report may affect one decision, while the same weakness embedded in an automated model can be repeated across thousands of recommendations or transactions. Manufacturers therefore cannot separate their AI ambitions from the condition and governance of the information beneath them.
“No matter how advanced the technology is, its outputs will only be as reliable as the data that feeds it,” Samuel says. “Inconsistent definitions, poor-quality master data, or unclear ownership can lead to inaccurate insights, reduced trust in AI-generated recommendations, and challenges in connecting processes and systems across the value chain.”
Transactional data is only part of that foundation. Generative AI tools and enterprise assistants increasingly consume procedures, standards, technical documents and internal knowledge when answering questions or supporting decisions. This means manufacturers must govern knowledge assets with the same discipline they apply to operational records.
“Historically, maintaining policies, standards, procedures, and other documentation was often viewed as an administrative task with limited business value,” Samuel says. “Today, that same content is being consumed by AI tools and leveraged as part of enterprise knowledge bases to generate insights, recommendations, and answers.”
Version control, ownership, review cycles and approval workflows now influence the quality of AI outputs. An agent given access to outdated maintenance instructions or conflicting operating standards may generate a confident but unreliable answer. Trusted AI therefore depends on both trusted data and trusted organizational knowledge.
The business must own the data
Governance cannot succeed when it is treated exclusively as an IT responsibility. Technology teams have an essential role, but ownership must reflect where data is created, understood and used. “I believe the answer depends on which aspect of data governance we are discussing, because data governance is a broad discipline that spans both business and technology responsibilities,” . Samuel explains. “There are certainly areas where IT plays a critical governance role, particularly around data platforms, security, access management, integration, and the technical controls required to ensure data can be consumed safely and effectively.
“However, when we look at data governance at source—where data is created, maintained, and managed – the responsibility must sit much closer to the business. In leading manufacturing organisations, Enterprise Master Data teams should own the governance framework—defining standards, policies, roles, controls, and ways of working. However, the data itself should be owned by the business functions that create, maintain, and use it. After all, it is the business that understands the meaning, purpose, and impact of the data on operational outcomes.
“The most successful organisations embed governance into everyday operations by making it part of existing business processes rather than treating it as a standalone initiative. Data ownership is clearly assigned, standards are incorporated into operational workflows, and accountability exists at the point where data is created and maintained. As a result, governance becomes part of how the organisation operates, rather than an additional layer of administration.
“Ultimately, data governance is most effective when it is viewed as a shared responsibility. IT provides the enabling technology and controls, Enterprise Data teams provide the governance framework, but the business must own the data. That combination is what drives sustainable data quality, accountability, and trust across the organisation.”
Measure the impact of doing nothing
Governance investment can be difficult to justify when its benefits are framed primarily in terms of frameworks completed, roles assigned or policies published. Manufacturers need to connect governance to the operational costs it can remove and the business capabilities it makes possible.
“When discussing the value of data governance, I often ask a simple question: ‘What is the impact of doing nothing?’” Samuel says. Poor information already creates measurable consequences through rework, reporting errors, delayed processes, regulatory exposure and decisions based on incomplete evidence. Governance provides a way to reduce those costs systematically rather than accepting them as an unavoidable feature of complex operations.
“I’ve seen organisations achieve measurable benefits such as right-first-time data creation, reduced manual corrections, faster access to trusted information, improved regulatory reporting, and lower operational costs,” she adds. “It also creates the foundation for AI, automation, and advanced analytics by ensuring data can be trusted and used consistently across the business.”
This changes the investment case. The value does not sit within the governance framework itself, but in improved process performance, faster access to information and the ability to scale digital capabilities without repeatedly correcting the same underlying problems.
Literacy turns trusted data into value
Reliable information will still deliver limited value when employees do not understand how to interpret, use or challenge it. Data literacy must therefore sit alongside ownership, standards and quality controls as part of the wider governance model.
“Data literacy is essential and should be a core component of any data governance framework,” Samuel continues. “It also helps address the common misconception that data governance is simply about policies and controls. Governance creates trusted data, but data literacy is what enables people to understand it, use it effectively, and ultimately unlock its value.
“This is particularly important as manufacturers invest in AI, advanced analytics, and digital technologies. Building trust in AI starts with building confidence in the underlying data. If employees understand how data is created, managed, and used, they are far more likely to trust and adopt data-driven insights.”
Employees do not need to become data specialists, but they increasingly need the confidence to understand where information comes from and question results that do not reflect operational reality. Business glossaries, data catalogues, targeted training, communities of practice and cross-functional workshops can help develop this capability, particularly when they are embedded into everyday working practices rather than presented as a standalone programme.
Join the conversation at the Manufacturing Data Summit
Malgorzata Samuel, Operations Data Standards Manager at BAT, will appear on the panel, “Data governance, quality and literacy: From frameworks to real business outcomes,” at the Manufacturing Data Summit UK 2026. “I’ll be focusing on three closely connected topics: data governance at source, data literacy, and their role in enabling AI and automation,” she says. Her contribution will explore why data quality must be addressed where information is created, how manufacturers can strengthen ownership and accountability, and why employees throughout the business need the confidence to understand and challenge data.
Taking place in London on October 6, 2026, the Manufacturing Data Summit will bring together manufacturing, operations, data and digital leaders to examine how data, analytics and AI can deliver practical operational impact. The agenda will include expert speakers, panel discussions, case studies and roundtables covering data quality and governance, AI deployment, digital twins, predictive maintenance, cybersecurity, architecture and the skills required to scale digital transformation.
Register to attend the Manufacturing Data Summit and hear directly from Malgorzata Samuel and other manufacturing leaders developing the data foundations required to turn AI, automation and analytics into sustainable business value.
