Guide 17 min read

When the plant does not answer

Ask the plant a question and three incomplete answers arrive: SCADA, the historian, and the previous shift. None of them is a whole record.

By WizeeMind Team Published on August 16, 2026 Updated on August 16, 2026

Night-shift operator studying three mismatched control-room screens

The question is usually ordinary. Why did the line lose rate after 02:00? Is this batch still inside the hold rule? Did nights already chase the same alarm, or are we about to repeat their work? Nobody is asking the plant for a theory of operations. They want one answer they can use before the next decision hardens.

What comes back is almost never silence. SCADA shows the live state. The historian shows a trend, a tag list, sometimes a long archive. The previous shift leaves a sentence that sounds helpful until you try to act on it. Each answer is true in its own window. Each one is incomplete in a different way. The frustration is not that the plant has no data. It is that the question still has no owner after three systems have spoken.

When people say the plant does not answer, they are not describing an empty control room. They are describing the absence of a single usable record: source, time, asset, and the limit of what the evidence is allowed to decide. Continuity of work is a communication problem, not a storage problem (HSE). Manufacturing control and enterprise systems do not already share one language (ISA-95). This page stays with that feeling. It is not a platform tutorial. The more specific work lives in industrial data context, search cost, and what a handover packet should carry.

Three answers, none of them whole

SCADA is the answer that feels most honest in the moment. The graphic is live. The color is current. The operator can see the state that exists now, which is exactly what a control system is for. Then someone asks what happened two hours ago, or whether the same sequence appeared on the previous product, and the live display has already moved on. The current number is not a memory. It is a present tense. That is useful. It is also why a supervisor can stare at a healthy-looking screen and still not know whether the night crew already bypassed a check, already reset a sequence, or already decided the problem was just the sensor.

The historian looks complete because it is long. Process historians were once simply software packages that stored and presented real-time process data. Wayne Matthews, writing in ISA InTech, says that period is over. Process historians now store data from a wider variety of sources, including control and monitoring, laboratory information management, enterprise resource planning, and asset management systems. Matthews writes that organizations also use historians to transform that data into information for equipment diagnostics, maintenance, safety, alarms, production, and performance. That breadth is real. It still does not make the historian the conversation. A trend can show that a temperature moved. It cannot tell you whether the outgoing supervisor already knew, already dismissed it, or already asked maintenance to look and never heard back.

The previous shift carries the human part. Someone saw the noise, the smell, the workaround, the call that did not get returned. They leave a line in the log or a remark at the console: watch the pump, same as last night, quality is looking at it. The sentence is an act of care. It is also a compression. It drops the time window, the asset alias, the procedure revision, and the point where the outgoing crew stopped being sure. The incoming person inherits a mood more than a record. That is why the same question can be asked again at 07:10 and still feel unanswered.

None of those answers is fake. The damage comes from treating them as interchangeable. The live value, the archive, and the night note are different kinds of record. Mixing them as if they were one proof is how a meeting gets louder without getting clearer. ISA-95, also published as ANSI/ISA-95 or IEC 62264, is an international set of standards aimed at integrating logistics systems with manufacturing control systems (ISA-95). The standard exists because those systems occupy different layers and need an interface, not because one of them already contains the whole plant. NIST describes smart manufacturing systems as having the ability to transform data gathered from manufacturing processes into actionable knowledge for decision making. NIST also reports that manufacturers often obtain actionable analytics results too late to have impact. The late result is often the hour spent noticing that the live screen, the archive, and the night note do not add up to a decision.

The feeling on the floor is smaller than that language. Someone asks, “What do we know?” and people look at different screens. A manager wants a cause and gets a current value, a plot, and a shrug. An incoming supervisor does a private calculation: trust the screen, the trend, or the person who just went home. That calculation is not incompetence. It is what a split record feels like when the line is still running.

Why the next shift inherits a gap

The unanswered question rarely dies at the end of a turn. It just changes form. The outgoing crew still has the last hour in their heads: which alarm cleared, which one they stopped believing, which call they meant to make. The incoming crew receives a current state and a short note. The gap between those two things is the inheritance. It is not that the log is blank. It is that the thing the next person needs, the unresolved question, did not survive the transfer.

HSE defines the goal of handover as the accurate, reliable communication of task-relevant information across shift changes or between teams, thereby ensuring continuity of safe and effective working. HSE describes effective handover as three elements: preparation by outgoing personnel, an exchange of task-relevant information, and cross-checking by incoming personnel as they assume responsibility. That model is stricter than a verbal download at the door. Preparation means the outgoing person has to decide what the next crew must know. Exchange means both sides share the work of making the record usable. Cross-checking means the incoming person is allowed, and expected, to test the story before they own it. When any of those steps is skipped, the next shift inherits a gap even if the logbook has ink in it.

HSE notes that many accidents have occurred because of failure of communication at shift handover, and that the majority of these involved planned maintenance work. That sentence should sit in the room when someone says the missed detail was only operational. A gap in the record is not automatically a safety event. It is also not a clerical inconvenience. The unanswered question is often about work that crosses a boundary: a permit still open, a temporary condition, a check that was postponed, a piece of equipment that is running while someone else owns the repair. Those are the questions that feel too large for a one-line note and too urgent to leave until tomorrow’s meeting.

HSE says spoken and written communication can be critical in maintaining safety, including communications between teams during operations or maintenance work. HSE on communications If a message is important, HSE advises considering two or more methods or media, such as written backup to a verbal communication. The plant already half-knows this. That is why there is a log and a conversation and a screen. The failure is more precise. The channels repeat the same thin sentence instead of carrying different parts of the same question. The verbal update says watch it. The written note says watch it. The live display shows a number that looks calm. Nobody records the disagreement, the missing work order, or the fact that the outgoing crew never saw the laboratory result.

This is why the next shift can feel accused by a question they did not create. They did not lose the information. They received a plant that had already answered in fragments. Our article on what gets lost at shift handover maps the packet that should travel with responsibility. The point here is narrower and more uncomfortable. Even a decent handover can still hand over an unanswered question if SCADA, the historian, and the outgoing note were never made to face one another. The incoming supervisor then spends the first hour reconstructing a conflict that the previous crew already lived through. That reconstruction has a tone. It feels like the plant is refusing to speak, when the plant has been talking the whole time in separate rooms.

The plant is not silent; the record is split

Silence is the wrong word. It also puts the blame on the wrong people. Operators are not withholding the answer. Historians are not empty. SCADA is doing the job it was bought to do. The record is split across layers that were designed for different activities. ISA-95 organizes technology and business processes into layers defined by activities and outlines how an enterprise can set up an interface to communicate among those layers. ISA-95 relies on the Purdue model and describes an equipment hierarchy that can be applied across discrete, continuous, and logistics operations. That is an architecture statement, not a plant insult. It means the live control view, the operations record, and the business record were never supposed to be the same object.

Yan Lu, Katherine C. Morris, and Simon P. Frechette write that standards are fundamental to facilitate the delivery of the right information at the right time. NIST Their NIST review focuses on integration standards across product, production system, and business dimensions. Read that against the floor question and the split becomes easier to see. Product information lives in quality and genealogy. Production-system information lives in control, maintenance, and the historian. Business information lives in orders, inventory, and downtime codes. A supervisor asking whether the line can keep running is standing in the gap between those dimensions. The plant answered in all three. It did not answer in one place.

Matthews notes that older historians often used proprietary databases that limited user access and made it difficult to interface with other plant information systems. ISA on historians Modern historians, he writes, can send and receive data through more open interfaces such as OPC, Ethernet, and SQL. Open interfaces reduce one kind of silence: the kind where a person cannot even see the archive. They do not automatically repair the other kind, where the archive, the live display, and the night note still refuse to name the same asset, the same clock, or the same decision. More pipes can move more fragments. They do not, by themselves, produce a usable answer.

MESA’s manufacturing analytics resources include work on linking plant-floor metrics to financial business drivers and correlating plant performance to business performance. MESA That work is a reminder that the split is not only a controls problem. Operations wants a cause. Finance wants a code. Quality wants a hold status. Maintenance wants a work-order history. Each function is asking the plant a slightly different question and then being surprised that the answers do not snap together. The feeling on the floor is simpler. Someone needed one sentence they could defend, and they received systems that would each be right if the others did not exist.

Neighboring articles should stay in their lanes. Industrial data context is about the relationships a record needs before it can support a decision. This page is about the experience of asking before those relationships are visible. The plant is noisy with evidence. The silence is in the join. Until source, time, and asset travel together, the separate answers keep sounding like a refusal.

What a single usable answer would require

A usable answer is not a prettier dashboard. It is also not a master database that swallows SCADA, the historian, and the night log. Those fantasies collapse the first time two clocks disagree or a tag name does not match a work-order asset. What the floor needs is more modest and harder: one response to a named question that a second person can still use after the first person has gone home.

Collecting and analyzing data to inform changes that improve system performance is a hallmark of smart manufacturing, but NIST says full use of that data is impeded by a lack of well-defined methods and standards. NIST NIST’s operations-driven project develops methods, tools, and standards for identifying and analyzing performance issues through systems characterization and data analysis. System characterization, whether of an individual process or the broader manufacturing system, defines the frame of reference needed to evaluate and improve performance against a norm. In plant language, that frame is the missing piece when separate answers arrive. Without the asset, the operating state, the time window, and the decision being prepared, each system is characterizing a different object.

NIST says much time is spent identifying target performance objectives, selecting or developing the analytics tool, and managing input data, including acquisition and formatting. NIST analytics That inventory is painfully close to the hour after a plant question. People spend the time deciding what they are even asking, then hunting the right screen, then cleaning the identifiers so the historian point and the shift note might refer to the same thing. The delay is not a lack of intelligence. It is the cost of assembling a frame that no system kept.

A usable answer would therefore carry a short, visible set of limits. Which source produced each statement. Which clock it used. Which asset or batch it refers to, including local aliases. Which procedure or approval boundary applies, and where the evidence stops. Yan Lu, Katherine C. Morris, and Simon P. Frechette write that standards are also meant to enable actions based on that information and to reduce the risk of technology adoption. NIST standards landscape The action here is usually small: inspect, hold, escalate, or keep watching under a named condition. The answer is usable when the next person can see why that action is supported and why a larger action is not.

That is the only honest place to mention WizeeMind. WizeeMind is a plant assistant that keeps source, time, asset, and approval limits attached to evidence. It does not approve work. It does not invent a single source of truth by merging incomplete answers into a fluent paragraph. If the live display, the historian, and the previous shift still disagree, the useful output is the disagreement with its citations still attached. A cleaned story that hides the split is another way for the plant not to answer.

The decision test is blunt. Could the incoming supervisor act on this without calling several people and still explain the limit of the evidence to a quality lead or a plant manager? If not, the plant has not answered yet. It has only spoken in parts. Keep the fragments in one packet, name the missing join, and give that packet to the role that owns the decision.

What this costs before anyone opens a project

The cost shows up before anyone writes a charter: another loop around the control room, a second call to the person who just left, a meeting that starts with “we still do not really know.” Those minutes do not look like a project. They look like professionalism. People are checking. They are being careful. They are also paying a tax that never appears as its own line.

According to APQC research published by Lauren Trees, the average knowledge worker spends 8.2 hours each week looking for, recreating, and duplicating information and expertise. APQC That figure is not a plant measurement, and it should not be converted into a local money total. It is a warning about the shape of the work. Lauren Trees is a former principal research lead for knowledge management at APQC. APQC profile Trees writes that a good knowledge-management strategy should increase the chance that people find what they are looking for and trust that it is right. On the floor, trust is the part that fails first. The historian can be found and still not believed. The shift note can be found and still not be enough. The live screen can be trusted for the current state and still be the wrong answer to last night’s question.

NIST reports that manufacturers often obtain actionable analytics results too late to have impact. NIST The late result has a human texture. By the time the separate answers have been compared, the batch has moved, the crew has changed, and the person who saw the event is asleep. The plant then pays twice: once in the search, and again in a colder reconstruction. Our guide to the hidden cost of plant information search is the place to measure those hours. This page is about the cost that appears before anyone agrees to measure anything: asking the same question again, and the way a room grows careful when a manager wants a cause.

There is also the cost that sits next to lost runtime. APQC publishes an open-standards measure titled unplanned machine/equipment downtime as a percentage of scheduled run time. APQC downtime measure The measure does not tell a plant what last night’s stop was worth. It does name the thing the unanswered question is usually circling. While people reconcile SCADA, the historian, and a night note, scheduled time continues. The project can wait. The question cannot. Treating the split record as a later IT problem is how a plant spends real operating attention on a join that should have traveled with the evidence.

NIST’s operations-driven work says smart manufacturing systems rely on the ability to quickly measure and collect a wide variety and volume of data from manufacturing operations, while full use of that data is impeded by missing methods and standards. NIST measurement That is the cost in official language. On the floor, it sounds shorter. We have the data. We still cannot answer. The delay is not only the stop itself. It is the hour spent deciding which fragment to believe before anyone can assign the next check.

None of this requires a new program this week. It requires a refusal to pretend that partial answers are one answer. Keep the disagreement visible. Keep the source attached. Ask whether the next shift can inherit the question without inheriting a scavenger hunt. If they cannot, the plant has not answered, and the cost is already being paid in the only currency the floor uses: attention that should have gone to the process.

Frequently asked questions

These questions come back after the meeting, when the screens are still open and the night note still does not settle the argument. They are operational, not a substitute for local procedures or a formal investigation.

Why does the plant feel silent when every system already has data?

The plant is producing fragments. SCADA, the historian, and the previous shift each answer a different part of the question, so no one record is usable alone. The live display answers the present state. The historian answers a stored interval. The outgoing crew answers what a person noticed. ISA-95 exists because those layers need an interface. Feeling ignored is what a split record feels like from the chair.

Which of the three answers is the source of truth?

None of them is. Live control data, stored history, and human observation are different record types, and a usable answer has to keep those types visible. Collapsing them into one truth usually means discarding the conflict that mattered. Matthews’ account of historians receiving control, laboratory, ERP, and asset data is a description of more sources, not of one court of appeal. ISA historian article A source of truth that erases the disagreement is not truth. It is a cleaner gap.

What should a supervisor do when SCADA, the historian, and the previous shift disagree?

Name the disagreement, keep each source attached to time and asset, and assign a bounded check. Do not flatten the three answers into one story before the conflict is visible. HSE’s handover model asks for outgoing preparation, an exchange, and incoming cross-checking; the disagreement is exactly the material those steps should preserve. HSE shift handover A bounded check has an owner, a window, and a stop condition. It does not authorize a process change.

Can software create a single source of truth on the plant floor?

No. Software can keep source, time, asset, and approval limits attached to evidence. It cannot invent one authoritative record or approve the next action. NIST’s warning about analytics arriving too late is partly a warning about time spent shaping inputs; a tool that hides those inputs repeats the delay in a smoother voice. NIST data analytics

Why does the unanswered question survive the next shift change?

Handover often transfers the current number and a short note, not the unresolved question, the missing record, or the conflict the outgoing crew already saw. HSE says the goal is continuity of safe and effective working through accurate, reliable communication of task-relevant information. HSE A number without the open question is not continuity. It is a clean-looking gap.

If the answers still do not make one record, leave the split visible and give the next check to the role that can close it. That is a better close than another summary of systems the plant already owns.