From Plant to Enterprise: sedApta MES Connecting People, Processes, and Data

A manufacturing execution system earns its place when a number measured on the line reaches the boardroom unchanged, and when a decision made in the boardroom reaches the line the same day.

A plant closes a shift with a solid OEE figure on the line dashboard. By the time that figure reaches the monthly operations review, it has passed through three spreadsheets, one email thread, and a manual posting into the ERP. Nobody in the meeting can say which version is correct, so the discussion moves to reconciling numbers instead of acting on them. Plant directors see this as lost time on the floor. Operations directors see it as a gap between reported and actual performance. CIOs see it as an integration backlog, and digital transformation managers see it as the reason pilots stall. A manufacturing execution system (MES) built to connect people, processes, and data addresses all four views at once.

Key takeaways

• Treat the MES as the operational layer between ERP and machine control, where plant events become enterprise information.
• Standardize the data model before adding dashboards, so every site reports OEE, downtime, and scrap on the same definitions.
• Connect machines through open protocols such as OPC-UA and MQTT to keep existing equipment and reduce customization.
• Close the loop between planning and execution, so a schedule change reaches the line and a line event reaches the plan.
• Design for operator adoption from day one, with digital work instructions that replace paper and spreadsheets on the shop floor.
• Scale from one line to the network in stages, packaging what works at the first site for the next one.

Where plant data and enterprise decisions part ways

Most manufacturers run two operating realities. The first lives on the floor: machines, operators, work orders, quality checks, changeovers. The second lives in the enterprise: orders, inventory, costs, commitments to customers. Each reality has its own systems and its own rhythm. ERP works in transactions and planning cycles. The plant works in seconds and shifts.

The gap between the two rarely shows up as a single failure. It shows up as small, repeated costs. A planner schedules against a capacity assumption that the line stopped meeting two weeks ago. A quality deviation is recorded on paper and reaches the batch record the next day. A recall request arrives and the team spends hours rebuilding which components went into which lot. A director asks for performance by plant and receives five answers built five different ways.

Executive ambition for closing this gap is high. The Deloitte 2025 Smart Manufacturing and Operations Survey, which polled 600 executives at large manufacturers, found that 92% believe smart manufacturing will be the main driver of competitiveness over the next three years. Respondents reported average gains of 10% to 20% in production output, 7% to 20% in employee productivity, and 10% to 15% in unlocked capacity since implementation. The same survey shows where the effort is still unfinished: only 54% report having a data standard in place, and 45% have an architecture standard. Technology is being bought faster than the foundations that make it comparable across sites.

The McKinsey analysis of the Global Lighthouse Network describes the same pattern from the opposite side. The network has grown from 16 to 189 Lighthouse sites in six years, and the article identifies a "scaling slump" in which pilot results fail to reach network-wide impact. The sites that escape it build data platforms that connect process, product, supplier, and customer data, invest at the site level first, and then package what works for reuse elsewhere. The leaders treat the connection between plant and enterprise as infrastructure that keeps evolving after go-live, with its own owner, budget, and roadmap.

The layer between ERP and the machines

The industry already has a shared vocabulary for this problem. The ANSI/ISA-95 standard, also published as IEC/ISO 62264, defines models and terminology for integrating enterprise and control systems. As Siemens summarizes the framework, Level 0 covers the physical process, Level 1 the sensing and manipulation of that process, and Level 2 the monitoring and control, typically through PLCs and distributed control systems. Level 4 covers the business-related activities that run a manufacturing operation, with ERP as the usual example.

Level 3 sits between them. It covers the workflow activities that produce finished goods, and ISA-95 Part 3 describes manufacturing operations management as the activities that coordinate personnel, equipment, and material. Production, maintenance, quality, and material handling all live here. This is the layer where a work order becomes a sequence of steps on a machine, and where a sequence of steps on a machine becomes a record the enterprise can trust.

When Level 3 is thin, the enterprise guesses and the plant improvises. Spreadsheets fill the vacuum, and each spreadsheet encodes its author's assumptions. When Level 3 is solid, every event on the floor has one owner, one timestamp, and one definition, and the enterprise reads the same record the operator wrote.

The sedApta MES is designed for this position. It integrates planning and execution, removes spreadsheets from the shop floor, and consolidates production KPIs into a single dashboard. Its core capabilities include digital work orders, automatic machine data collection, in-line quality control, bidirectional batch and component traceability, and role-based KPI dashboards. Because it sits between the ERP layer and machine control, it carries preconfigured connectors for SAP, Microsoft, and Oracle on one side and OPC-UA and MQTT for PLCs and SCADA on the other. The practical effect is shorter go-live times and fewer customizations, which is the part of the project a CIO feels most directly.

Connecting data: from machine signal to enterprise KPI

A plant produces data continuously. Turning it into something an enterprise can compare requires three steps: collect it close to the machine, give it a common structure, and calculate indicators from it on agreed definitions.

Collection is the first obstacle, because plants run equipment from many generations and vendors. Replacing machines to get data is rarely an option. An edge layer solves this by connecting to the equipment already installed. sedApta Edge IIoT connects through open protocols such as OPC UA, MQTT, and REST API, processes data locally, converts it into a common data model, and passes it in real time to MES, planning, and analytics systems. Local processing has a second benefit for plant and IT leaders alike: data collection continues when the connection to the cloud drops, so the plant record has no gaps.

The common data model is where much of the enterprise value is created, and it is the step most often skipped. Without it, line A reports OEE using one treatment of planned downtime and line B uses another, and the average across both means nothing. The MES enforces one definition at the point where data is captured. It calculates OEE from availability, performance, and quality using machine and operator data, and it tracks cycle times, downtime, scrap, MTTR, and rework rate against the same logic on every line.

Indicator definitions matter at the enterprise level as well. ISO 22400-2 specifies a set of KPIs for manufacturing operations management, with each indicator described by its formula, elements, time behavior, and unit. Using an international reference for KPI definitions removes a recurring argument in operations reviews, which is whether two plants are measuring the same thing. sedApta Analytics builds on that reference by offering a predefined catalog of ISO 22400-compliant KPIs, pulling data from ERP, MES, WMS, and APS systems into one store, and giving each role a dashboard with the data relevant to its decisions.

For an operations director, the result is a comparison that holds up. Five plants, one definition of OEE, one definition of scrap, one calendar. For a digital transformation manager, it is the first credible baseline against which every later initiative can be measured.

Connecting processes: closing the loop between planning and execution

Data alone does not make an operation responsive. Processes have to carry it in both directions, from plan to floor and from floor to plan.

In many plants the loop is open. The plan is built in the planning system, exported, adjusted by hand, and printed. Execution deviates from it within hours, and the deviation is reported back through a daily meeting or a morning spreadsheet. The planner then reschedules using information that is already a day old.

A closed loop works differently. The sedApta MES synchronizes with Factory Scheduling and Supply Chain Planning, so a schedule reaches the line as digital work orders and a line event returns to the schedule as a fact. When a machine goes down, the scheduler sees it as it happens and can resequence the remaining orders. When a material shortage appears, the planning layer sees the impact on committed orders while there is still room to respond.

The same logic extends to longer horizons. An Elisa Industriq article on continuous scenario planning with sedApta DDM+ describes planning cycles of three to four weeks set against market conditions that change in days, and argues that connecting demand signals, shop-floor events, and supplier data in one system is a planning architecture question. The shop floor is the source of the events that make that architecture work. A planning model fed by accurate execution data can run scenarios that reflect what the plant can do this week, whatever the master data said when it was last updated. Planners stop spending the first half of the day correcting inputs and spend it comparing options.

Consider a concrete scenario, offered as an illustration. A packaging line on the second shift loses 40 minutes to a recurring jam. With paper records, the loss shows up in a weekly summary and is attributed to "mechanical issues." With an MES, the stop is captured with its duration and reason at the moment it happens, the scheduler pushes the next order to another line with spare capacity, and maintenance sees the third occurrence of the same jam within a week. The plant fixes the cause during the next planned stop. The enterprise sees the effect on delivery dates the same afternoon, and the customer service team can adjust a promise before the customer calls. One event, captured once, serves four functions.

Quality follows the same path. Checks inside the production flow stop the process when a parameter falls outside its specified range and raise a non-conformity at that moment. The deviation reaches quality, maintenance, and planning together. Traceability completes the picture: each order carries a full history, so when a recall request arrives the affected batch can be identified quickly instead of reconstructed from paper.

Governance: who owns the connection

Connected operations raise an ownership question that technology alone does not settle. Plant data now serves operations, quality, finance, planning, and IT, and each group has a legitimate claim on how it is defined and used. Projects that leave this open tend to drift into either a plant-owned system that the enterprise cannot read or an IT-owned system that the plant does not trust.

Three decisions keep the connection healthy. The first is a named owner for the data model, usually a joint role between operations and IT, with authority to approve new indicators and retire old ones. The second is a change process for master data, so that a new product, routing, or work center enters the MES and the ERP once and propagates to both. The third is a review rhythm in which operations and IT look at integration health together, covering failed messages, data latency, and open requests from the plants. Deloitte's finding that operations leaders own smart manufacturing in 51% of companies and technology executives in 38% shows how often the two groups must share the work. A shared owner turns that split from a source of friction into a working agreement.

Security belongs in the same conversation. The Deloitte survey reports that 65% of manufacturers rank operational risk as their first or second smart manufacturing concern, and it lists unauthorized access as the most frequently cited OT risk. Connecting more systems enlarges the surface to protect. An integration design with defined interfaces, role-based access, and local data processing at the edge gives the CIO a clear map of what talks to what, which is a stronger position than defending a web of undocumented point-to-point exports.

Connecting people: why adoption decides the outcome

Many MES projects that technically work never deliver their return, because the people who generate the data on the floor find the system slower than their old method. Paper and spreadsheets return within weeks, the data quality drops, and the enterprise layer built on top of it inherits the gaps.

The Deloitte survey is direct about where the industry stands on this point. A Deloitte analysis of the same survey calls human capital the least mature area of smart manufacturing. Of the executives surveyed, 35% cite adapting workers to the factory of the future as a top workforce challenge, and only 48% have a training and adoption standard in place. Leadership alignment and change management rank among the most-cited implementation challenges. The analysis concludes that real transformation happens when operations and technology come together, which matches the survey finding that 51% of companies place smart manufacturing ownership with operations leaders and 38% with technology executives.

That shared ownership has a design consequence. The system has to serve the operator at the terminal as well as the director reviewing the dashboard. In the sedApta MES, operators receive digital work instructions on a terminal or panel PC, with step-by-step progress tracking in place of paper. Plant managers see OEE, MTTR, and rework rate by line, and operators see their own instructions and order status. Each role gets the view relevant to its job, drawn from the same record.

The McKinsey Lighthouse research points the same way. In that group, 75% of sites have deployed solutions in each of five frontline talent areas: safety, skill development, skill augmentation, work augmentation, and worker experience. Frontline people are part of the technology plan from the start in the factories that scale.

Digital tools can also extend what experienced people know. Elisa Industriq's piece on AI-driven operational intelligence frames AI as a way to enhance and extend human expertise, with AI supplying scale and speed while people supply context and judgment. That division of labor depends on clean, connected execution data, which is what the MES layer provides.

What each leader gets from the same system

The same connected foundation answers different questions depending on the seat.

A plant director gets a line view without the manual work: OEE by line, downtime reasons captured at the source, and orders that arrive with instructions attached. Shift handovers rely on the system record instead of verbal summaries.

An operations director gets comparability across sites. When definitions are identical, the review shifts from reconciling figures to deciding where to intervene. Capacity utilization, scrap, and cycle times can be ranked across plants with confidence.

A CIO gets an integration pattern instead of a custom project per site. Preconfigured ERP connectors and open protocols for the shop floor shorten implementation and limit customization, which lowers the long-term maintenance load. The architecture can start where it matters and extend as the business case develops, since the MES is described as scalable for small, medium, and large companies.

A digital transformation manager gets the baseline and the reference model. One site with reliable data becomes the template, and the Lighthouse practice of packaging site-level assets for reuse gives a proven route to the network.

How to start: a seven-step roadmap

1. Pick the first line with intent. Choose a line where the pain is measurable, such as frequent unplanned stops, paper-based quality checks, or slow traceability. A clear baseline makes the return visible.
2. Define the indicators before the dashboards. Agree on OEE, downtime categories, scrap, and rework definitions, using ISO 22400-2 as the reference, and have operations and finance sign off on them.
3. Map the ISA-95 levels you already have. List which systems cover Levels 1 to 4 today, where data is re-keyed by hand, and which handoffs rely on spreadsheets.
4. Connect machines with open protocols. Use an edge layer and standard protocols so existing equipment feeds the MES without replacement, and verify that data collection survives a connection loss.
5. Integrate ERP and planning in both directions. Confirm that work orders flow down and confirmations, consumption, and quality results flow up, with Factory Scheduling and planning kept in step.
6. Involve operators in the workflow design. Test digital work instructions with the people who will use them, collect their feedback, and adjust screens before rollout.
7. Package and replicate. Document the data model, the connectors, the training approach, and the results from the first site, then use them as the template for the next one.

Conclusion

Connecting a plant to the enterprise is an operating discipline that has a technology layer, and the MES is where the two meet. The evidence from large surveys and from the factories recognized for scaling points to the same ingredients: shared data definitions, open integration, a loop between planning and execution, and systems that frontline teams choose to use. A plant that has them can answer a question from the boardroom with a number from the line, and can move a decision from the boardroom to the line within the same shift.

Next step

To see how this layer fits with planning and scheduling, read the article on continuous scenario planning with sedApta DDM+, or talk to the Elisa Industriq team about a first-line assessment.