Tools can enforce rules about tag definitions, but a rule only exists if a person decided it, and a definition only stays correct if someone is accountable for keeping it that way. The data steward is that person: the named human who owns the tag definitions for an area, approves changes to their naming and scaling, and answers for their quality. This guide explains what a data steward actually does, how the role differs from a data owner, and why tags without a steward inevitably rot - framing SCADA governance as a matter of people and process, not just software.
Data Steward in one line: A data steward in SCADA governance is the person accountable for the quality of tag definitions in a given area or facility - owning the naming conventions, approving changes to tag scaling and limits, and ensuring definitions stay accurate and documented. The steward is distinct from the data owner, who holds ultimate authority and accountability, and their central task is running the approval workflow that keeps tag metadata correct so that unowned, ungoverned tags do not degrade over time.
A data steward is the hands-on custodian of tag definitions within their scope, which is usually an area, a facility, or a class of equipment. They maintain and apply the naming conventions, so a new tag is named consistently rather than however the person adding it felt like naming it. They review and approve changes to a tag's scaling, engineering units, and alarm or control limits, so a re-ranged transmitter is reflected correctly and deliberately rather than casually. And they keep the definitions documented, so the description of what a tag measures stays accurate and meaningful to the next person who reads it.
The steward is also the person who watches for and corrects data quality problems in their domain. When a tag's definition looks wrong, when duplicates appear, when a description is stale, or when a value seems inconsistent with its stated range, the steward is the one expected to notice and act - or to be the point of contact when someone else notices. This is ongoing, unglamorous work: it is the continual tending that keeps a tag database from decaying. The steward does not necessarily do all the technical work themselves, but they are accountable that it gets done and done correctly.
Because stewardship is scoped, a large operation typically has several stewards, each responsible for their own area, rather than one person for everything. A steward for the compression facilities, another for the tank batteries, another for the gathering system - each close enough to their equipment to know whether a definition makes sense. This distribution matters: a steward who understands the process behind the tags catches errors that a distant central team would miss, because they know what a plausible range or limit for their equipment actually is. Good stewardship depends on domain knowledge, not just administrative diligence.
Governance distinguishes the data owner from the data steward, and the distinction is about authority versus execution. The data owner is typically a manager or senior figure who holds ultimate accountability for a body of data and the authority to set policy about it - who decides that tag definitions must be governed, sets the standards, and answers at the highest level for the data's integrity. The owner usually does not do the day-to-day tending; their role is authority and accountability at the top.
The data steward operates under that authority as the person who carries out the governance in practice. The owner says definitions will be consistent and controlled; the steward is the one who actually enforces the naming convention, reviews the change requests, and keeps the documentation current. In smaller operations one person may wear both hats, but the two responsibilities are conceptually separate, and keeping them clear prevents the common failure where everyone assumes someone else is accountable. A useful way to make this explicit is a responsibility matrix that records, for each data domain, who owns it, who stewards it, who must be consulted, and who is merely informed.
Much of the steward's authority is exercised through the approval workflow. A proposed change to a tag definition - a rename, a re-scale, a new alarm limit, a new tag added - is not simply made; it is requested and routed to the steward, who reviews it against the standards and the likely impact and either approves it or sends it back. This workflow is what turns governance from a policy on paper into a working control: because a definition can only change through the steward's approval, the steward is genuinely able to keep the definitions consistent, and every change carries a record of who requested it and who approved it. Without such a workflow, stewardship is a title with no teeth.
The clearest argument for stewardship is what happens without it. A tag that nobody owns is a tag nobody maintains: when the equipment it measured is removed, no one decommissions the tag; when a range changes, no one updates the definition; when the description becomes wrong, no one corrects it. Over time an ungoverned tag database fills with orphans, duplicates, stale descriptions, and mismatched limits, and the data as a whole becomes less trustworthy because no one can vouch for any given tag. Rot is the default outcome of an absence of accountability, not a rare failure.
This is why SCADA governance is fundamentally a people-and-process problem rather than a software one. A tool can hold definitions and enforce a change workflow, but it cannot decide what a correct definition is, judge whether a change makes sense, or care that a description is meaningful - those are human judgments that require someone accountable for making them. Assigning a steward puts a name against every part of the tag database, so that for any tag there is a person who is supposed to know it is right, which is the thing that actually prevents rot.
Software's proper role is to make stewardship enforceable and auditable, and this is where a cloud SCADA platform such as Merobix supports governance rather than replacing it. The platform lets ownership and stewardship be assigned to real users against areas and tags, routes change requests to the responsible steward for approval, and records the full history of who changed what and who approved it, so accountability is not just a policy but a tracked fact. When definitions and their change history live in one central system, a steward can actually see the state of their domain, and an owner can audit whether governance is being followed. The people set the standards and make the judgments; the platform ensures those judgments are the only way definitions change and keeps a durable record that they were made.
The data owner holds ultimate authority and accountability for a body of data - usually a manager who sets policy and answers at the highest level for its integrity. The data steward operates under that authority and does the practical work: enforcing naming conventions, reviewing and approving change requests, and keeping definitions accurate. The owner decides the data will be governed; the steward makes governance happen day to day.
A tag that nobody owns is a tag nobody maintains, so when equipment is removed the tag is not decommissioned, when a range changes the definition is not updated, and when a description becomes wrong it is not corrected. Over time the database fills with orphans, duplicates, and stale or mismatched metadata, and the data becomes less trustworthy because no one can vouch for any given tag. Assigning a steward puts accountability against every tag and prevents that decay.
No. Software can hold definitions and enforce a change workflow, but it cannot decide what a correct definition is, judge whether a change makes sense, or care that a description is meaningful - those require an accountable human. Governance is fundamentally people and process, with software playing the supporting role of making stewardship enforceable and auditable by routing approvals and recording who changed what.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.