OT Tag Naming: The Data Standard Industrial AI Depends On

Why does OT tag naming decide whether industrial AI works?

OT tag naming is the convention that tells people and systems what each plant measurement is and where it belongs. It decides whether industrial AI and analytics work, because a model cannot reason about data that has no consistent names. The fix is to adopt an existing standard such as ISA-95 and IEC 81346, carry it in a historian asset framework, and gate new work against it.

Walk into most Australian processing plants and ask what FIC_1023_PV measures. Someone will know. Ask about FT101.PV, Mill2_FeedRate and AI_4417, and you get three different people, two shrugs and a spreadsheet nobody has updated since the last upgrade. The plant runs fine, so nothing gets fixed.

Then the board asks whether AI can go over the operational data, and the honest answer is usually "not yet". The models are good enough. The data has no names a machine can reason about.

This article is for control systems engineers, plant data leads, reliability engineers and operations managers asked to deliver an AI outcome on an estate that grew one project at a time. Across our consultants' careers, including PI System, AVEVA and Citect work at BHP, Rio Tinto and Senex Energy, the finding has been consistent: the biggest barrier to analytics is rarely the historian or the model. It is the taxonomy nobody owns.

What is a tag name actually doing?

A tag name is a contract. Every time an engineer writes a report, builds a dashboard, configures an alarm or trains a model, they rely on the name meaning what it meant last week and on the panel next door.

When the contract holds, "sum the feed to all three mills" is a one-line query. When it does not, the same query becomes a research project: find every feed tag, confirm each is mass flow and not volumetric, check the units, exclude two decommissioned ones, and hope Mill 3 followed the same logic as Mills 1 and 2. It did not, because a different contractor commissioned it in a different year.

The cost of a weak convention is invisible until you work at scale. One hand-built dashboard is fine. Forty dashboards, a Power BI model, an alarm rationalisation and a condition monitoring project reading the same estate turn the gap into permanent, recurring engineering labour.

	Ad hoc tags	Governed convention	Impact
Find every mill feed tag	Manual search + tribal knowledge	One structured query	Repeatable
Add a new asset	Name it however the contractor likes	Follows the pattern by rule	Consistent
Build a Power BI model	Re-map tags by hand each time	Maps once, inherits structure	Reusable
Onboard an engineer	Months of learning local quirks	Reads the convention and works	Faster
Apply AI or analytics	Data prep dominates the project	Context already present	Lower risk

Which OT data standards already exist for tag naming?

You do not have to invent a convention. Several international standards describe how to structure plant data, and the right move is to adopt and adapt one instead of arguing about it in workshops for six weeks.

ISA-95 (IEC 62264): where a measurement lives

ISA-95, published internationally as IEC 62264, defines the levels of a manufacturing enterprise from the sensor up to the business system. It gives you shared vocabulary for where a measurement lives and what it belongs to: "a Level 2 process value on this equipment, rolling up to this production area". We cover the practical side in our guide to mapping a plant to ISA-95 when there is no MES.

IEC 81346 and RDS-PP: what a thing is

IEC 81346-1 sets out structuring principles and reference designations, a formal method for giving every object an unambiguous identifier tied to its place in the plant hierarchy. In power generation the same idea has a long history as KKS, now carried forward as RDS-PP. The principle is constant: a name should encode what the thing is and where it belongs, by rule.

ISA-5.1: what the letters mean

For instruments, ISA-5.1 governs identification and symbols, so an F is a flow measurement, a T is temperature, and function letters mean the same thing on every drawing. These standards are the accumulated agreement of the industries that have done this longest. Adopting one costs far less than the recurring tax of having none.

1. 1AreaPlant area or unit, by ISA-95 hierarchy
2. 2EquipmentThe asset the measurement belongs to
3. 3FunctionWhat is measured, by ISA-5.1 letters
4. 4AttributeValue, setpoint, status or mode

How do you make a tag naming convention stick?

Choosing a structure takes a fortnight. Making it stick takes governance, and that is where most efforts die. A standard that lives in a PDF on a shared drive is a suggestion the next capital project will ignore under schedule pressure.

A convention holds only when three things are true:

1. The rule is written down and owned by a named person, with authority over new work and the historian.
2. New work is checked against it before commissioning, as a gate, never as a clean-up afterwards.
3. The existing estate is mapped to it over time, instead of being left as a permanent exception.

The third point matters because a brownfield site will never be renamed wholesale. The PLCs, loops and operator habits are too entrenched. What you can do is build a structured layer above the raw tags that carries the standard, which is exactly what an asset framework is for.

Why the asset framework carries the standard

Instead of renaming 40,000 points, you build a hierarchy that references them, applies consistent attribute names, and inherits structure from templates. Adding the fourth identical pump takes minutes and arrives correctly named. Our guide to PI Asset Framework templates that survive a plant change explains how the template turns a one-off naming effort into a standard that propagates itself.

1. Weeks 1-2Choose and adaptPick ISA-95 plus a reference designation scheme, tailor to the site
2. Weeks 3-6Build the top layerModel the area and equipment hierarchy in the historian asset framework
3. OngoingGate new workEvery capital project commissions against the standard, checked before handover
4. OngoingMap the legacyReference existing tags into the model as projects touch each area

Can AI sort out messy plant tags on its own?

On operational data, the idea that a capable model can read any tag soup and work it out is close to backwards. A language model can guess that Mill2_FeedRate is a feed rate. It cannot know that the number is wet tonnes, measured after the scalping screen, double counts recirculating load, or that the tag was repurposed in 2023 and the old meaning still appears in three reports. That context is not in the signal. It lives in people's heads.

Anaconda's State of Data Science surveys have repeatedly found data preparation to be one of the largest calls on analysts' time. On a plant, that preparation is mostly reconstructing context a proper naming standard would have carried for free. Fix the names and structure, and you remove one of the biggest cost items from every future analytics and AI project at once.

This is why a resilient data foundation comes before model choice. An AI agent reading operational data needs the tag context and a read-only boundary far more than a particular model. The discipline that makes a Power BI report trustworthy is what makes an AI answer trustworthy, because both draw from the same well.

What is your most pressing constraint?

• Every report rebuilt by handModel the asset hierarchy first
• New projects keep divergingWrite and gate the convention
• Nobody owns the standardAssign the data owner before anything else
• Legacy tags are chaosReference, do not rename, via the framework

What does carrying on without a standard cost?

The decision gets deferred because the status quo has no line item. The cost is spread across every engineer who spends a day working out what a tag means, every dashboard that breaks when an asset is renamed, every analytics project that loses its first two months to data prep, and every AI initiative that stalls at proof of concept.

For example, consider a single mid-sized processing plant. The figures below are illustrative, framing the recurring tax, and are not a measured result.

• Engineer time rebuilding a dashboard after one asset rename2-3 days
• Share of a typical analytics project spent on data prepLargest single item
• Lead time to add a new identical asset by handHours each time
• Same task once the asset framework template existsMinutes, by rule

None of this is a technology purchase. It is a governance decision wearing a technical hat, which is why it falls between the control systems team, who own the tags, and the business, who own the reporting. On boards it surfaces as a question about AI readiness, which we address in our note on what Australian boards should ask before approving AI.

How should an Australian plant start on tag naming?

The mistake is to scope a full-estate rename, cost it at a number nobody approves, and shelve it. The better path is narrow and compounding:

1. Pick one production area that matters for reporting.
2. Model it in the historian asset framework using an adopted standard.
3. Prove that a report or analytics query over that area is now simple where it used to be laborious.
4. Make the standard a commissioning gate so new work cannot undo the progress.
5. Bring the legacy estate into the model area by area as projects touch it.

We use the same sequencing whether the goal is production reconciliation, where SCADA, the historian and the ERP need to agree on tonnes, or alarm management, where rationalisation depends on historian evidence. Our data, BI and analytics and system integration teams do this work on mining, energy and manufacturing sites.

How Solve8 can help

We help mining, energy and manufacturing operators put a governed data foundation under their reporting and AI work, starting with one production area. See our manufacturing and resources solutions or talk to our team about your site.

Related reading

• ISA-95 on a plant with no MES
• PI Asset Framework templates that survive a plant change
• When SCADA and your ERP disagree on tonnes
• Getting historian data into Power BI