Production execution is where an MES proves whether it understands the plant. Operators and supervisors judge the system by a simple standard: does it help work move correctly, visibly, and with less confusion? If work orders are dispatched late, if data capture is unreliable, if WIP disappears between stations, or if labor performance is still reconstructed from manual logs, then the MES has not become the execution backbone it was meant to be.
This chapter focuses on the design choices that make production execution workable in daily operations. It explains how work orders should be dispatched and routed, how shop-floor transactions should be collected, how real-time WIP visibility should be built, and how labor tracking should support productivity management without overwhelming users. The best MES designs reflect how the plant actually runs while tightening the points where discipline creates value, especially during unstable shifts.
9.1 Work Order Dispatch and Routing Design
Work order dispatch is the moment a planned order becomes active work at a specific resource, and it is one of the most important control points in MES. If dispatch is unclear, late, or disconnected from plant reality, every downstream transaction becomes less trustworthy. Operators may start jobs before the system is ready, supervisors may resequence work outside the tool, and material or quality events may be recorded against the wrong order. A strong dispatch design therefore does more than move orders from ERP into a queue. It determines how the plant authorizes work to begin.
Design objective: The dispatch model should tell the plant what can run now, where it can run, in what sequence, and under what prerequisites. Those prerequisites usually include order release status, material availability, line or machine readiness, approved instructions, required tooling, and any quality or engineering conditions that must be met before execution starts. If those checks are missing, the plant may appear productive while actually accumulating hidden rework or inaccurate status data.
The routing design behind dispatch has to mirror the operational complexity of the value stream. Some plants run simple linear flows with minimal branching. Others depend on alternate work centers, parallel operations, split and merge points, temporary holds, rework loops, and inspection gates. MES does not need to reproduce every irregularity, but it does need to represent the real paths the product can take. When routing logic is oversimplified, plants compensate with manual notes, offline approvals, or local databases, and the system quickly loses authority.
Sequencing rules deserve special attention. In theory, ERP or APS may provide the right production sequence. In practice, supervisors often need limited flexibility because actual conditions change by the hour. Material arrives late, a feeder line slips, a qualified operator is unavailable, or a short order becomes urgent. The design challenge is to allow controlled resequencing without destroying visibility. MES should show both the planned sequence and the executed sequence, and it should capture why a meaningful deviation occurred.
Another critical element is operation start and completion logic. The system should define exactly what constitutes the start of an order or operation and what evidence is required before completion can be confirmed. In a highly automated line, start may be triggered by a machine state and product count. In a manual work cell, it may require an operator action tied to a barcode or device log-in. Completion may require produced quantity, scrap declaration, quality confirmation, and the release of the next step.
MES should also present the right instruction, recipe, setup sheet, or parameter set in the exact context of the order and operation being run. If the route calls the correct step but the instruction presented is outdated or ambiguous, the plant has digital routing and analog confusion.
A practical dispatch model therefore needs three characteristics. Clarity: Users can see what is next and why. Control: The system prevents or flags invalid progression. Flexibility: Supervisors can manage real conditions without breaking the execution record. That is where trust starts.
9.2 Shop-Floor Data Collection Framework
Once work is dispatched, the next question is how the MES will know what is actually happening. That is the purpose of the shop-floor data collection framework. In theory, every plant wants real-time production data. In practice, data quality rises or falls based on the design of the transaction experience. If capturing an event is slow, confusing, or poorly matched to the physical rhythm of work, users will delay it, batch it, bypass it, or invent side records. Good data collection is therefore not mainly a technology question. It is an operational design question.
Framework principle: Collect data at the point where the event occurs, using the least burdensome method that still preserves control and traceability. That method may be barcode scanning, RFID, touchscreen entry, mobile devices, weight-scale integration, automated machine signals, or a combination of several. The right answer depends on cycle time, ergonomics, automation maturity, traceability requirements, and the frequency of exceptions.
Barcode scanning remains one of the most effective methods because it is simple, inexpensive, and visible. It works well for order start, unit or lot identification, material consumption, container movement, and completion confirmation. RFID can reduce manual effort where contactless capture is valuable, but it requires more careful physical design and exception handling. Manual entry is still necessary in many plants, yet it should be reserved for information that cannot practically be captured another way.
Machine integration is powerful when it is purposeful. Automated count capture, state changes, test results, torque values, or recipe confirmations can improve speed and data integrity significantly. But automation should not be added merely because the signal exists. Each machine-fed event should answer a real execution question: did the operation complete, was the parameter in range, did the unit pass the test, did the line stop, or did the resource produce output against the active order?
The framework should also define transaction timing. Some events must be captured immediately, such as lot verification before use or a failed quality check that should stop progression. Others can be collected at short intervals, such as labor allocation in a repetitive cell or minor scrap categories reviewed within the hour. The plant should make these timing decisions consciously. Many unstable MES deployments result from assuming everything must be real time or from allowing critical transactions to be delayed until shift ends.
User interface design matters as much as the device itself. Operators need screens that reflect the sequence of physical work, with minimal navigation and obvious error handling. Supervisors need fast ways to correct or approve exceptions without re-entering the entire transaction path. Plants often overcomplicate their data collection forms by asking for information that is nice to have but rarely used. Every extra field increases friction.
Exception handling must be built into the framework from the start. What happens if a barcode is unreadable, a scanner fails, the network drops, or an operator needs a fallback device? If the answer is improvised, data integrity will suffer. Good MES teams define continuity procedures and re-entry rules before go-live so that temporary workarounds remain controlled and traceable.
At its best, the data collection framework creates confidence that the digital record reflects the physical truth of the line. The plant no longer reconstructs what happened after the fact. It captures execution as it happens, in the right context, with enough discipline to make the data dependable and usable for action.
9.3 Real-Time WIP Visibility Design
Most factories carry more WIP uncertainty than leaders admit. Product waits between operations, sits in carts, queues for inspection, moves to rework, or simply becomes hard to find. The ERP order may still look open and the line may still look busy, yet no one can say with confidence where the work actually is or what condition it is in. Real-time WIP visibility is the MES capability that closes that gap.
Visibility goal: The system should show where each relevant unit, lot, batch, or container is in the process, what status it currently holds, what has already happened to it, and what must happen next. That depends on disciplined status modeling, transaction timing, and location logic that fit the plant’s real flow. If those pieces are weak, WIP visibility becomes approximate rather than actionable.
Status design is the foundation. A useful model usually distinguishes queued, in process, complete at operation, awaiting inspection, on hold, in rework, and ready for next step. Some plants need finer distinctions, especially when curing, staging, cleaning, or intermediate storage materially affect throughput. The key is that each status should trigger a management meaning. A supervisor should know whether the work can move, needs attention, or is blocked for a reason that requires escalation.
Location design is equally important. In a simple flow line, work-center visibility may be enough. In high-mix, batch, or campus-style plants, WIP often needs to be visible by rack, room, oven, tank, lane, cart, or quarantine area. The MES model should reflect the level of physical granularity required to control the product, not merely the level that is easy to configure.
Split and merge behavior is where many WIP designs break down. A parent lot may be divided across several work areas and later recombined conceptually in finished-goods reporting. Several subassemblies may feed one final serial number. A tank batch may fill multiple packaging orders. MES must preserve these relationships so that flow visibility and genealogy remain coherent.
Real-time visibility is not only about location. It is also about flow. The plant should be able to see where queue time is growing, where inspection waits are accumulating, which resources are starving, and where rework is consuming hidden capacity. Good MES design therefore presents WIP in a way that supports action, not just inquiry. A supervisor should be able to tell which orders are aging in queue, which holds are preventing release, and which bottleneck is setting the pace for the value stream.
One of the most practical uses of WIP visibility is handoff control between functions. Production, quality, and materials often work from different signals. When MES makes WIP states explicit, those handoffs become tighter. Quality can see what is awaiting review. Materials can see what is staged versus actually consumed. Planning can see which orders are physically close to completion rather than simply open.
Strong WIP visibility improves plant management by replacing search time, guesswork, and verbal reconciliation with a live execution picture. That is especially valuable in unstable conditions, where the plant needs to know not just what should be moving, but what is actually stuck and why across the entire shift.
9.4 Labor Tracking and Productivity Monitoring
Labor tracking is often one of the most sensitive parts of MES because it touches cost, performance, and culture at the same time. If it is designed poorly, operators experience it as surveillance and supervisors experience it as administrative overhead. If it is designed well, it becomes a practical way to understand effort, support scheduling, validate standards, and identify where productivity is being lost.
Purpose: Labor tracking should create an accurate picture of who performed work, on which order or operation, for how long, and under what production conditions. That picture supports costing, staffing analysis, line balancing, and productivity improvement. It also strengthens traceability in environments where operator qualification or sign-off is important. The mistake is to treat labor capture as purely a finance requirement.
The first design choice is the unit of capture. Some plants need direct labor booking by order and operation, especially where work is manual, variable, or high mix. Others need labor captured at team, line, or shift level because individual booking would create too much friction. Repetitive environments may use crew assignment combined with automated production context. Complex assembly sites may need operator-level start and stop records tied to specific route steps. There is no universal model, but there is a universal rule: the method must match how labor is actually deployed.
The second choice is timing. Real-time labor sign-on and sign-off provides precision, but it may be unnecessary in stable, repetitive processes. Periodic allocation can work where labor rotates predictably and costing does not require minute-level detail. The important point is consistency. If the plant changes methods by area without clear rules, productivity metrics become incomparable and disputes over performance data begin almost immediately.
Productivity monitoring should combine labor data with execution context. Hours alone do not explain much. The plant needs to see labor against output, mix, scrap, downtime, rework, and planned staffing assumptions. A line may appear labor-inefficient when the real problem is waiting inspection or chronic microstops. Another area may appear efficient only because rework effort is hidden outside the main route.
Standard rates and target hours are useful only if they are treated as living controls rather than permanent truths. MES can help compare actual labor content to standards at a much higher frequency than traditional reporting allows. That creates an opportunity to refine standards, balance work across stations, and identify where new product introductions or engineering changes have altered the real labor requirement.
Cultural design matters here more than in many other modules. Leadership should be explicit that the intent is better operational control, fairer planning assumptions, and stronger line support, not punitive monitoring of individuals. Where unions or employee councils are involved, the program should define clearly what labor data is collected, how it is used, who can see it, and how long it is retained.
When labor tracking and productivity monitoring are designed with the plant, not just for the plant, they become a source of discipline rather than friction. Supervisors can allocate people with better evidence, planners can understand realistic capacity, finance can trust labor postings more, and operations leaders can see whether productivity loss originates in staffing, process design, equipment loss, or quality disruption on every production day.
