MES is becoming manufacturing’s business system
Manufacturing execution used to describe a bounded factory discipline. It managed activity on the line, captured production events, supported traceability and told the plant what had happened. That remains important, but it no longer covers the work manufacturers now need execution platforms to perform.
Tom Forster, regional vice president, enterprise software sales EMEA at Rockwell Automation, sees the old MES category being stretched by supply chain volatility, smaller lot sizes, cloud deployment and AI. “Given the need that exists in the manufacturing environment to not look at MES as a manufacturing execution system, but rather as a business system, coupled with cloud scalability and the dawn of AI, that is what made Plex so attractive,” he says. Rockwell acquired Plex five years ago, adding a cloud-native, SaaS-based enterprise MES platform that Forster says is now used by more than 800 organizations in around 45 countries.
The shift is not about making production control less important. It is about recognizing that production can no longer be separated from the business conditions around it. A machine fault, material shortage, quality issue or scheduling change affects labor, cost, inventory, delivery and customer commitments, so the value of MES increasingly depends on how quickly it can connect the factory event to the business decision.
If that connection still must be rebuilt through spreadsheets, local workarounds or after-the-fact reporting, the business is not operating from production truth. It is reconstructing that truth later, often after the most useful moment to act has already passed. That is why MES is moving from a plant-level system of record toward a broader layer for operational intelligence and business performance.
The factory cannot work from isolated truth
Forster traces the change to the way manufacturers now have to operate. “When you’re just operating on a domain of manufacturing data, you have less flexibility, less speed in your ability to operate when business data isn’t tied in,” he says. “Production information needs to connect horizontally with quality, scheduling, inventory and workforce activity, while also connecting vertically with ERP, finance, costing and planning.”
Manufacturers do not only need to know that OEE has fallen or a line has missed its target. They need to know whether that affects overtime, whether the schedule must change, whether a shipment is at risk and whether a quality issue is beginning to appear elsewhere. Those connections are where execution data begins to become business intelligence.
“For manufacturers to be able to adapt and react more in real time than ever before, they need to have a connected environment where data isn’t locked in Excel spreadsheets on different people’s desktops,” Forster says. “It has to be connected and derive context not only from its operational context but from the business context that exists behind it.”
That redefines what an execution platform is expected to do. “Operational data isn’t this separate team, separate set of data owned by a separate set of people,” Forster says. “You start to make decisions on how you plan, on how you schedule and on how the organization derives priorities based upon the operational truth. The old factory silo may still collect useful information, but it cannot support faster business decisions if the context arrives too late.”
Cloud changes the standardization argument
Cloud MES remains sensitive because factories are physical, local and time critical. Manufacturers are right to worry about resilience when production depends on digital systems. Forster argues that the better way to view Plex is not cloud versus plant, but edge-to-cloud, with the benefits of SaaS combined with local continuity.
“It is true multi-tenant, single-instance SaaS, so you get the advantages of being able to take in updates, not having to incur the heavy cost penalties of managing infrastructure and taking in continuous innovation,” he says. “But we also have an edge component that allows for 24/7 operation if there are challenges with internet connectivity. The point is the ability to standardize without freezing every plant into the same rigid template.”
For manufacturers that have grown through acquisition or allowed sites to choose their own systems, that flexibility matters. A common model can be defined, site differences can be understood, and a blueprint can be rolled out without returning to the heavy customization of older MES projects. “There’s flexibility and extensibility that allows some degree of customization, and then it becomes easier to deploy across multiple sites in an enterprise,” Forster says.
West-Bake shows why these matters in a business that had already outgrown its old ways of working. The Galway-based bakery had grown from 20 employees in 2007 to 150, with turnover reaching €35 million, but that expansion had created data silos, paper-based processes and fragmented communication. Information existed, but it was not collaborative enough to support fast decisions as the company prepared for expansion into Europe and the Middle East.
Working with Rockwell Automation and Cumulus Consulting, West-Bake implemented the Plex Smart Manufacturing Platform, combining MES, QMS and ERP components in a single environment. Real-time production, inventory and quality data gave teams a shared view of whether they were ahead or behind schedule, while serialized barcodes and batch management improved traceability and gave the business the ability to access recall records almost instantly. For a company with capacity to double throughput and ambitions to triple in size over the next decade, execution became part of the growth infrastructure.
Quality has become execution data
Quality is often treated as a separate discipline, but in practice it is inseparable from production. If quality evidence is collected manually, stored separately or compiled after the event, the organization loses speed and confidence just when it needs both. Audits, recalls and compliance checks may expose the weakness, but the same issue also limits everyday improvement.
Forster describes a different model, where the data required to trace products, SKUs and lots is captured in the flow of production. “It’s not something that gets manually tracked, prone to individual user and operator error,” he says. “It is collected in real time through your manufacturing process, stored in a single place and has relation to the products, SKUs and lots that it is related to.”
Once that information is connected to production history, manufacturers can start to identify the patterns behind defects, recalls or process failures. “You can go back and understand what some of the different patterns are that lead to quality challenges or recalls or incidents, and engineer them out of the manufacturing process,” Forster says. “The value is no longer limited to proving what happened.”
AI needs execution context
Every manufacturing technology conversation now reaches AI, but Forster does not see AI reducing the need for execution systems. He sees it making them more important because models need context, not just data. Without an accurate operational record of what was produced, which process was running, which lot was involved and what conditions surrounded the event, AI has a weak foundation.
“For lack of a better term, MES becomes more interesting in the world of AI because the data becomes so important,” he says. “I see AI as something that goes hand in hand with a platform like Plex. It creates data and meaning that AI can train on and build on to then provide inference into other areas.”
The early applications are practical rather than speculative. AI can help with quality patterns, embedded copilots, operator alerts and integration mapping between MES and other enterprise systems. It can prompt people to inspect something before a problem travels further down the production process or help users interrogate MES information more quickly.
That does not remove people from manufacturing decisions. Safety, quality, compliance and customer commitments still require judgment, particularly where the consequences of a wrong decision are physical. “Given the criticality, you’re always going to have a person in the loop,” Forster says. “I see AI as an enabler, not as a full replacer, of what we do in our systems.”
The warning signs of a fragmented execution environment are often visible long before the business names them as an MES problem. Replanning takes too long when supply is disrupted. Quality audits take weeks to prepare. Critical process knowledge sits inside experienced workers’ heads or in what Forster calls “Excel masterpieces,” creating risk as those people retire or move on.
The starting point is not to buy technology first. It is to map how processes work today, which systems support them, where manual workarounds exist and what risks those workarounds create. Only then can the business decide what an execution platform should standardize, what it should connect and where it should leave room for local difference.
The name MES may not describe that future forever. Rockwell has already used the language of Elastic MES because the old category can feel too narrow. Whatever the label becomes, execution is moving from a plant-floor control layer into the connective tissue between production reality and business performance.
