Manufacturing Execution Systems (MES): The Plant Leader's Guide for 2026

Most plants that own a manufacturing execution system can show you its dashboards. Fewer can show you a shift lead who used it this morning to find out why line 3 lost forty minutes, or a quality technician who used it to put the right pallets on hold before they shipped. That gap is the real story of MES. In most plants Joltek has worked in, the MES ended up as a source of management metrics, not as a tool the people running the plant use to solve problems.

Joltek’s position is simple. A manufacturing execution system (MES) earns its keep when it creates information plant teams act on during their shift. Getting there is mostly a data problem, not a software problem. Joltek’s founder, Vladimir Romanov, has rolled out MES at Kraft Heinz, Post Holdings and other manufacturers, on platforms from FactoryTalk View and PlantPAx to Ignition with Sepasoft, Proficy, Wonderware, custom builds and MES functions built directly into SCADA. The hard part was rarely the software. It was getting to the data on the plant floor, and keeping the plan small enough to deliver.

This guide is for the plant or engineering leader who is choosing an MES, replacing one, or trying to get value from the one they have. It covers:

• what an MES is and what it actually does, function by function
• how it differs from ERP, SCADA, a historian and a unified namespace
• how data flows into and out of an MES, with one order traced through an interactive production line
• why connecting the plant floor is the hardest part
• how discrete, batch and hybrid plants need different things, and how industry changes the choice
• who sells MES in 2026, after a year of acquisitions
• the KPIs worth producing, what an MES costs, and why so many projects stall

What is a manufacturing execution system (MES)?

A manufacturing execution system (MES) is the software that runs and records production between the business plan in ERP and the machines on the floor. It turns orders into instructions for the line, tracks what was made, from which materials, on which equipment, by whom and to what quality. The ISA-95 standard calls this layer manufacturing operations management.

The industry’s classic definition comes from MESA International, which wrote in 1997 that an MES delivers “information that enables the optimization of production activities from order launch to finished goods. Using current and accurate data, MES guides, initiates, responds to, and reports on plant activities as they occur” (MESA White Paper 6, 1997). Nearly thirty years later it still holds up. The important words are “as they occur”. An MES that tells you tomorrow what happened yesterday is a reporting tool.

You will also see the term manufacturing operations management (MOM). MOM is what the ISA-95 standard calls the whole layer between business planning and control, covering production, quality, maintenance and inventory. MES is what most people call the software that does the production part, and in practice the two terms are used almost interchangeably. Searchers, vendors and plant teams overwhelmingly say MES, so this guide does too.

An MES is a set of jobs, not a product

The most useful way to think about an MES is as a set of jobs every plant already does somewhere. Someone decides which order runs on which line. Someone checks that the right ingredients or parts were used. Someone records what was made, what was scrapped and why the line stopped. Someone holds a batch that failed a test. In a plant without an MES those jobs still happen, on paper, in the ERP, in the SCADA, in spreadsheets and in people’s heads.

So the question is never really “do we need an MES”. It is “which of these jobs should move into a dedicated system, and which can stay where they are”. The answer depends on the plant, which is why the rest of this guide spends as much time on plant types, data and people as on software.

Who an MES is for

An MES serves more people than the reports suggest:

Who	What they need from the MES
Operators	The right instructions, recipe and materials for the order in front of them, and a fast way to record what happened
Shift leads	What is running, what is behind, why the line stopped, and what to do about it now
Quality	Results, holds and release, and a record that proves what was made and from what
Maintenance	Which machines fail, how often and how, before the failures become downtime
Planners	What actually got made against the plan, so the next schedule is realistic
Continuous improvement	Losses ranked by size and cause, to pick the next project
Plant and site management	Trends, capacity and cost

Most plants Joltek sees serve only the last row. The MES produces a daily OEE figure and a production summary for the morning meeting, and the people who could use it to fix problems during the shift work around it. One integrator described a plant where the MES recorded when each batch started and finished from pallet scans, and nothing in between: “There is no realtime data” (Inductive Automation forum, 2026). That is an MES serving the report, not the plant.

Where MES sits relative to the rest of Joltek’s technology stack is covered on where MES and batch sit in Joltek’s technology stack.

What does an MES actually do?

ISA-95 breaks MES work into four areas (production, quality, maintenance and inventory operations), each with the same eight activities, from scheduling and dispatching to data collection, tracking and performance analysis. On the floor that means work orders, recipes and instructions, material tracking and genealogy, quality checks and holds, downtime and performance, and the records that prove what happened.

The standard model: ISA-95 Level 3

The reference model for MES functions is ISA-95, published by ISA in the United States and as IEC 62264 internationally. Part 3 of the standard defines the activity models of manufacturing operations management. It splits the layer into four categories, production operations, maintenance operations, quality operations and inventory operations, and gives each the same eight activities (IEC 62264-3):

1. Product definition management
2. Resource management
3. Detailed scheduling
4. Dispatching
5. Execution management
6. Data collection
7. Tracking
8. Performance analysis

Four categories times eight activities gives thirty-two boxes. No plant needs software in all of them, and no vendor covers all of them equally well, which is the first useful thing the model tells you. The standard is current: ISA revised Part 1 in April 2025, and IEC published a new edition of Part 2, the object models for the interface with business systems, in March 2026. For how to use the activity model as a requirements checklist, see using the ISA-95 activity model as an MES requirements checklist.

The four areas, in plant terms

The four categories are easier to use once they’re translated into what happens on the floor:

• Production operations follow the order from release to confirmation: what to make, where, in what sequence, with which recipe, and what actually came off the line. This is what most people mean by MES.
• Quality operations cover samples, tests, inspections, holds and release, and the records behind them. In many plants this work is split between the MES, a laboratory information management system (LIMS) and a quality management system (QMS).
• Maintenance operations cover equipment condition, maintenance requests, the work itself and its history. That usually lives in a computerized maintenance management system (CMMS); the MES contributes run hours, fault counts and equipment status.
• Inventory operations cover material movement inside the plant: receiving, staging to the line, work in progress, and finished goods waiting for release. They overlap with the warehouse management system (WMS) and with ERP.

The model helps in two ways. It makes scope explicit: when a vendor says its product is a full MES, the right follow-up is which of the four areas, and which of the eight activities in each. And it shows where the integration work will be, because every area the MES doesn’t own is an interface to another system. A plant with a strong LIMS and CMMS needs an MES that exchanges data with them cleanly, not one that duplicates them.

The older list: MESA-11

The list most vendor pages still quote is older. In 1997 MESA International described eleven functions of an MES: operations and detail scheduling; resource allocation and status; dispatching production units; document control; product tracking and genealogy; performance analysis; labor management; maintenance management; process management; quality management; and data collection and acquisition. MESA later replaced it with broader models in 2004, 2008 and 2022.

One update matters for anyone citing MESA: MESA International ceased operation on 30 June 2026, and ISA took over its content and programmes (ISA, June 2026). Many links to MESA’s models now lead nowhere. MESA-11 is still a good checklist. It is a 1997 checklist.

Where each MES job usually lives today

The practical version of both models is a map of jobs, who uses each one, and where it lives in a typical plant before an MES arrives. Use it as a self-audit: a plant with half these jobs on paper has a different first step from one with them scattered across five systems.

MES job	What it does	Who uses it on the floor	Where it often lives today
Scheduling and dispatching	Turns ERP orders into a sequence by line and shift, and puts the next job in front of the crew	Planners, shift leads, operators	ERP, a spreadsheet, a whiteboard
Recipes, specs and work instructions	Delivers the approved recipe, settings and instructions for this product to the equipment and the operator	Operators, process engineers	Batch server, HMI screens, binders
Material tracking and genealogy	Records which lots went into which batch, unit, case and pallet	Operators, warehouse, quality	Paper batch sheets, ERP backflush, WMS
Data collection	Captures counts, states, faults and values from equipment and people	Everyone downstream	SCADA, historian, manual logs
Quality checks, holds and release	Records in-process and lab results, holds product that fails, releases it when cleared	Quality, operators	LIMS, QMS, paper, spreadsheets
Labor and training records	Records who did what, and whether they were qualified to	Supervisors, HR, quality	Time clocks, training binders
Downtime and performance	Turns states and reasons into downtime, OEE and losses	Shift leads, maintenance, CI	SCADA reports, manual downtime sheets
Maintenance triggers	Turns run hours and fault patterns into maintenance requests	Maintenance	CMMS, if anyone connects it
Production records	Keeps the record of what was made, how and from what, ready for an auditor or a recall	Quality, management	Paper, scanned PDFs, ERP

Most plants find that two or three of these jobs are well served, three or four are done on paper, and the rest are split across systems that don’t agree. That pattern, not a vendor demo, should decide what the MES does first.

How is an MES different from ERP, SCADA, a historian or a unified namespace?

ERP plans and keeps score in days and dollars. SCADA runs equipment second by second. A historian stores time-series values. A unified namespace shares current state across systems. An MES sits between them and owns the production transaction: which order, which materials, which equipment and people, what quality, and the record of all of it, in minutes and hours.

System	Works in	Owns	The question it answers	Main users
ERP	Days, weeks, dollars	Orders, inventory, cost, customers	What should we make, and what did it cost?	Finance, planning, supply chain
MES	Hours and minutes	The production transaction and its record	What is running, from what, by whom, to what quality, and what happened?	Operations, quality, planners
SCADA and HMI	Seconds	Equipment control and alarms	Is the equipment running, and what is it doing right now?	Operators, controls engineers
Historian	Seconds to milliseconds, kept for years	Time-stamped process values	What was this value at that moment?	Engineers, process and quality
Unified namespace	Real time	Current state, shared	What is the state of everything, right now, in one place?	Every system that subscribes

The boundaries matter because each system is good at its own job and poor at the others. A historian keeps millions of values efficiently but doesn’t know which order they belong to; for the difference in detail, see what a data historian does that an MES doesn’t. A unified namespace shares current state across the plant but doesn’t own orders, genealogy or records; it is a way of moving data, and an MES is one of the systems that publishes to it and subscribes from it. For how that architecture works, see how a unified namespace shares plant data.

Where MES stops and LIMS, QMS, CMMS and WMS start

The harder boundaries are often inside Level 3, between the MES and the other operations systems beside it. Each owns a different record, and the integration has to agree on who owns what:

System	Owns	What the MES sends it	What the MES gets back
LIMS (laboratory)	Samples, tests, specifications, results	Sample requests tied to a batch, lot or time window	Results, and pass or fail against specification
QMS (quality management)	Deviations, investigations, corrective actions, change control, controlled documents	Deviations and exceptions raised during production	Approved procedures and, in some plants, release decisions
CMMS (maintenance)	Work orders, preventive maintenance plans, spare parts	Run hours, fault counts, maintenance requests	Equipment availability and work in progress
WMS (warehouse)	Storage locations, stock movements, picking, shipping	Material requests, consumption, finished goods ready for put-away	Staged lots and their locations

The rule that keeps this manageable is that each record has one owner and the other systems reference it. Plants get into trouble when two systems can both place a hold, or both keep a version of the specification, and nobody decided which one wins. That decision belongs in the design, before any interface is built. The illustration in the next section shows one of these boundaries at work: the quality flow runs from the checkweigher and the lab, through the LIMS, to a hold on the line.

Can ERP do the job?

Sometimes, and honest guides should say so. A simple discrete plant making a stable range of products to stock can often run on the shop-floor modules of its ERP: work orders, labor reporting, backflushed material and a few quality checks. Plants with regulated records, batch recipes, high product mix or high-speed lines usually can’t. ERP doesn’t see what happens on the line minute by minute, can’t enforce how the work is done, and turns every shop-floor transaction into a finance transaction. One practitioner put it plainly: “ERPs are financial focused”, and their vendors “will happily upsell” add-ons such as “MES/BOMs, which are fine for generic manufacturing with limited SKUs” (Hacker News, 2025). Fine for some plants, then, and not for others.

What about OEE software?

Many plants start with an OEE or production-monitoring tool that counts output and records downtime on a few lines. That covers one MES job, performance, very well, and often delivers value in weeks. It isn’t a full MES: it doesn’t dispatch orders, enforce recipes, track materials or keep production records. Inductive Automation, the maker of Ignition, notes that OEE is not technically part of MES, though the two are closely linked. For plants whose biggest problem is lost capacity, starting there is often right; some never need more.

How does data flow into and out of an MES? One order, traced

Orders, product definitions and schedules come down from ERP, and the MES turns them into dispatch lists, recipes and setpoints. Machines send states, counts, faults and process values back up, joined by quality results and material lots. The MES then reports confirmations, consumption and performance to ERP. ISA-95 calls these exchanges product definition, production capability, production schedule and production performance.

Most explanations draw an MES as a box with one arrow from ERP and one arrow from the machines. The real flow is a conversation in both directions, at very different speeds, between systems owned by different people. ERP thinks in days. The filler thinks in milliseconds. The MES has to translate between them and keep a record that holds up when someone asks, months later, what went into a particular case.

The clearest way to see it is to follow one order through a hybrid food plant: a batch kitchen upstream that makes a sauce, and a packaging line downstream that fills, caps, labels, checks, packs and palletizes it. Use the controls on the illustration to switch flows on and off, change the plant type, or step through the order.

The eight flows shown in the illustration:

1. Orders and schedule: ERP releases production orders; the MES schedules them by line and shift and puts them on the line’s screens.
2. Recipes and setpoints: the MES sends the approved recipe and its parameters to the batch controller, and the target weights and speeds to the filler.
3. States, counts and faults: every machine reports whether it is running, stopped, blocked or faulted, and how many units it made, through SCADA to the MES.
4. Quality checks and holds: lab results, checkweigher rejects and in-process checks reach the MES, which can put a batch or pallet on hold.
5. Materials and genealogy: ingredient lots are linked to the batch, the batch to the filled units, and the units to cases and pallets.
6. Performance and downtime: the MES turns states, counts and reasons into OEE and downtime for the shift lead, maintenance and the plant manager.
7. Confirmations to ERP: quantities made, materials consumed and scrap go back to ERP to close the order.
8. Maintenance requests: repeated faults on a machine become a request in the maintenance system.

One order, step by step

1. ERP releases the order. A production order arrives for a run of sauce: product, quantity, due date, and the version of the formula and packaging bill of materials to use. In ISA-95 terms this is the production schedule, carrying a reference to the product definition. It changes daily or by shift.
2. The MES turns it into work. The MES checks production capability: is the kettle clean and released, is the filler set up for this bottle, are the right ingredients staged, are qualified people on shift. Then it schedules the order to a kettle and a line and dispatches it to their screens.
3. Materials are checked and linked. At weigh and dispense, each ingredient is scanned. The MES checks that the lot is released, not on hold and not expired, and records which lots went into which batch. This is where an MES stops a line with a message an operator may not like. One engineer described being called out at night because “MES is displaying an error message”, and finding that the real problem was “finding out what integration failed and who else to wake up” (Hacker News, 2026).
4. The recipe goes to the line. The MES sends the control recipe to the batch controller: the phases, their order, the setpoints and the limits, following ISA-88. It sends the fill target and line speed to the packaging line, and the work instructions to the operator’s screen.
5. The line reports back. The batch controller reports each phase as it starts and ends, with temperatures, mix times and any deviation. On the packaging line every machine reports its state, its counts and its fault codes, within seconds. The historian keeps the full time series. The MES keeps what matters to this order.
6. Quality joins the record. A viscosity result comes in from the laboratory system. The checkweigher sends its rejects. If anything is out of specification, the MES places the batch, or the pallets made from it, on hold until quality releases them.
7. Genealogy closes. The MES links the ingredient lots to the batch, the batch to the bottles filled during its time window on the filler, and the bottles to cases and pallets. This is the point where a batch plant becomes a discrete one, and where traceability most often breaks.
8. Performance goes back up. The MES confirms the quantity made, the materials consumed and the scrap to ERP, which closes the order and updates inventory. This is the production performance exchange. The same data becomes the shift’s OEE, downtime by reason for the shift lead, and a pattern of faults for maintenance.

Every step above happens somewhere in every plant, whether or not it has an MES. Without one, steps 2, 3 and 6 are usually paper and phone calls, and step 7 is a spreadsheet someone builds after a customer complaint.

How fast each layer moves

The layers work at very different speeds. These are typical ranges, not the standard’s wording:

Layer	Typical systems	Moves in	What it keeps
Business planning (Level 4)	ERP	Days and shifts	Orders, inventory, cost
Manufacturing operations (Level 3)	MES, LIMS, QMS, CMMS	Hours and minutes	The production record: what, from what, by whom, to what quality
Supervisory control (Level 2)	SCADA, HMI, batch server, historian	Seconds	Equipment states, alarms, process values
Control (Level 1)	PLCs, batch controllers	Milliseconds	Logic, interlocks, motion

Why the speed matters: the five-minute MES

The speeds are not a detail. On one P&G line, the MES recorded every tag value from the controllers once every five minutes, as Vladimir Romanov described in a Factory Field Notes Q&A. A fault that starts and clears in ten seconds never appears in data like that. The MES had years of records and could not explain a single stop.

The fix is to capture what changes when it changes. State changes, faults and counts should arrive as events with their own timestamps. Fast-moving process values belong in a historian. The MES should store what the order needs. For how polling rates and controller scan cycles shape what you capture, see how polling and scan cycles shape the data you collect.

Keeping the line running when the network doesn’t

A plant can’t stop because the link to ERP is down. Good MES designs push orders, recipes and master data down to the site ahead of time, buffer production data locally when an upstream connection drops, and send it on when the connection returns. ISA-95 has a standard format for these exchanges, B2MML, an XML implementation of the standard first published by MESA, though most integrations today run through APIs, databases or messaging instead.

Every one of these flows is where an MES creates value, and every one is where the integration effort goes. Which is why the next question matters more than the choice of vendor.

Why is connecting the plant floor the hardest part of an MES?

Because an MES is only as good as the data it can reach, and plant-floor data is hard to reach. Old controllers, incompatible protocols, IT and OT roadblocks, locked-down production areas, and data that has to be extracted, structured and given context before it means anything. In Joltek’s experience, MES projects stall here, well before anyone argues about software.

Across the MES rollouts Joltek’s founder has led, the same seven obstacles appear. None of them is in a vendor demo, and all of them decide the schedule.

1. Old systems. A twenty-year-old controller with no spare communication port, an HMI on an unsupported operating system, a machine that speaks a protocol nothing modern understands. The usual fixes are a protocol gateway, an added communications module, or reading the data from the SCADA or HMI layer that already talks to the machine. Sometimes the right answer is a controller migration that was overdue anyway.
2. Incompatible systems. Three PLC brands on one line, an OEM machine sold as a black box, a vendor that won’t expose its tags. A plant ends up with as many integration methods as machine builders. The fix is a standard: one way of exposing states, counts and faults that every new machine must meet, written into the purchase specification, and a plan for the existing ones.
3. IT and OT roadblocks. Nobody owns the boundary. A firewall change takes three months, a server request goes into a corporate queue, a security review nobody budgeted for appears halfway through. These are rarely bad intentions. They are missing agreements, and they are cheaper to settle before the project starts than during it.
4. Extracting the data. Getting values out of controllers without disturbing control, at the right rate, through a network that can carry it. Polling every tag every few seconds can load an old network and still miss the event that matters. The fix is to collect events where they happen and to design the network for the traffic; see building an OT network that can carry MES traffic.
5. Structuring the data. Tag names that differ from line to line, data types nobody standardized, no model of which machine belongs to which line and area. An MES needs a consistent structure, an asset model, before its reports mean anything across lines.
6. Contextualizing the data. A temperature is just a number until it is tied to an order, a batch, a product, a lot and a shift. Adding that context is the MES’s core job, and it depends on every earlier step working.
7. Locked-down production areas. Cleanrooms, validated systems, sanitation zones and plants where nobody may connect a laptop without a permit. Work takes longer, testing has to be planned around production, and remote access has to be designed rather than improvised; see getting controlled access into locked-down production areas.

These obstacles set the real timeline. In Joltek’s experience, a focused rollout takes roughly one to three months per site. The range depends on the complexity of the lines, the controllers and the network, and on what has to be fixed to reach the data and deploy servers. A plant with modern controllers and a segmented network is at the short end. A plant with mixed generations of equipment and no agreed IT and OT boundary is at the long end, before any MES function goes live.

Take a connectivity inventory before choosing software

The fastest way to turn those obstacles into a plan is an inventory of every machine the MES will touch, done before any vendor demo. For each one, record:

What to record	Why it matters
Controller, model and firmware	Sets which drivers and protocols are possible, and whether a migration is due anyway
Communication options	A spare port or module, the protocols it speaks, and whether the OEM allows access
Network path	Which segment it sits on, and what route exists, or doesn’t, to the MES servers
Data available today	States, counts, fault codes and recipe parameters: what exists in the program and what would have to be added
Access constraints	Validated systems, OEM warranties, work permits, sanitation or cleanroom rules
Fix needed and owner	The gateway, network change, code change or server work, and who will do it

The inventory usually changes the project. It shows which lines can be connected in days and which need weeks of network or controller work, so the first line can be chosen for speed rather than politics. It also gives IT and OT a concrete list to agree on, which settles three questions before they become delays:

1. Who owns the boundary between the plant network and the business network, and who approves changes to it?
2. Where will the MES servers live (at the plant, in a data center or in the cloud), and who patches them?
3. How will vendors and integrators get access, during the project and after it?

Discrete, batch and hybrid: how does plant type change the MES you need?

Discrete plants build countable units through a routing and track serial numbers. Batch plants make lots from ISA-88 recipes and need batch records and lot genealogy; continuous plants define lots by time window. Most food, beverage, CPG and pharma plants are hybrid, and genealogy breaks where batch becomes packaging. Buy for the hardest part of your plant.

	Discrete	Batch and process	Hybrid
Unit of production	Individual units, often serialized	Lots or batches of material	Batches upstream, units downstream
Master data	Bill of materials and routing	Formula and recipe (ISA-88)	Both, linked
Genealogy	Serial number to components	Lot to ingredient lots	Lot to batch to filled units to cases and pallets
Quality model	Inspection at stations, defects per unit	In-process checks, lab results, critical process parameters	Both, plus checkweighers and vision on the packaging line
Records	Device history record, traveler	Batch record, often electronic	Batch record upstream, packaging records downstream
Typical KPIs	Throughput, first pass yield, cycle time, OEE	Batch cycle time, yield, right-first-time batches	Batch KPIs upstream, OEE on the packaging line
Where it gets hard	Routing changes, rework loops, high mix	Recipe management, scaling, hold times, cleaning	The handover between the two

Batch: recipes, phases and records

Batch plants run on recipes, and the standard for them is ISA-88 (ISA-88 series; IEC 61512 internationally, with a new edition of Part 1 published in February 2026). It defines four types of recipe, from the general recipe that describes a product independent of any site, through the site and master recipes, to the control recipe that runs a specific batch on specific equipment. It also defines a procedural model that breaks a recipe into procedures, unit procedures, operations and phases.

That structure matters for MES selection. A batch MES has to manage recipe versions and approvals, send the control recipe to the batch controller, receive each phase’s results, record deviations, and produce a batch record that quality can review. ISA published a technical report in 2008 on how the two standards fit together: ISA-88 runs the batch, ISA-95 handles the exchange with the business. In plants on Rockwell systems, the batch layer is often FactoryTalk Batch alongside PlantPAx; see FactoryTalk Batch and PharmaSuite in the Rockwell stack.

At a greenfield food expansion at Post Holdings, the data system Vladimir Romanov’s team delivered combined SCADA, MES and batching, including recipe management and full automation of a roasting process from raw ingredients. The site’s OEE improved 18% as the data, kaizens and modernization took hold. The lesson from that project and others like it: in a batch plant, the recipe and its record are the heart of the MES, and everything else hangs from them.

Discrete: routings, units and serial numbers

Discrete plants build things you can count: assemblies, devices, components. The MES follows a routing, a sequence of operations at work centers, and tracks each unit or serial number through it. The hard parts are rework loops, engineering changes mid-run, high product mix, and labor reporting that doesn’t make operators walk away from their machines. Medical device and automotive plants add regulated device history records and serial traceability.

Continuous process: time windows and material balances

Not every process plant runs batches. Chemicals, refining, pulp and paper, extrusion and parts of dairy and beverage processing run continuously, sometimes for weeks, without a natural start and end for each lot. The ISA-88 recipe model fits poorly, because there is no batch to step through.

In a continuous plant, lots are defined by time window, by tank or by grade run: everything produced between two times, or everything that went into a given storage tank. The MES work centers on four things:

• Production accounting and material balance: what went in, what came out and where the difference went, by unit and by tank.
• Grade changes: when one product ends and the next begins, and what happens to the transition material in between.
• Quality by sample time: lab results tied to the moment the sample was taken, and so to the material produced around it.
• Process data in context: historian values linked to the lot or grade they belong to.

For selection, that shifts the weight. Integration with the historian and genealogy by time window matter more than routing or recipe phases, and a product designed for discrete assembly or for batch execution may model a continuous plant awkwardly. Many continuous plants also need less dispatching, because the schedule is a sequence of grades rather than a queue of orders.

Hybrid: where the genealogy breaks

Most food, beverage, CPG and pharma plants are both. A kettle or bioreactor makes a batch; a filling and packaging line turns it into thousands of units. The MES has to carry genealogy across that boundary: this batch went into the bottles filled between these times on this filler, which went into these cases and pallets. Partial batches, a tank that feeds two lines, rework blended back in, and a recipe revision mid-campaign all make that link harder.

Revisions deserve special attention. One engineer comparing MES platforms on Ignition described what happens when a production model changes: existing work orders stay on the old revision, “This creates a split dataset which can complicate reporting and traceability queries across the revision boundary” (Inductive Automation forum, 2026). That kind of detail decides whether a recall takes an hour or a week.

The rule for selection follows from the table: buy for the hardest part of your plant. A hybrid plant that buys a discrete MES because the packaging line is easier to demo will rebuild the batch side in spreadsheets.

What to test about plant type during selection

Vendor demos almost always show the easy case. Ask instead to see your plant’s hardest situations, built on a sample of your own data:

1. Genealogy across your own batch-to-packaging boundary, forward and backward, including a partial batch.
2. One tank or kettle feeding two lines at the same time.
3. Rework blended back into a later batch.
4. A recipe or routing revision in the middle of a campaign, and what happens to orders already open.
5. A grade change or product changeover, and how the transition material is recorded.
6. An operator correcting a mistake, and what the record shows afterwards.

A product that handles these cleanly in a demo will usually handle the rest. One that needs a workaround for two of them will need workarounds in production.

How do industries differ, and how does that change MES selection?

Regulation and product risk decide most of it. Pharma needs validated electronic batch records under Part 11 and Annex 11. Food needs lot traceability that can answer the FDA within 24 hours under FSMA 204. Automotive and aerospace need serial traceability for their quality standards. High-speed CPG needs performance data more than paperwork. Set those requirements before looking at vendors.

Industry	What drives the MES	Must-have functions	Regulation and standards (as of October 2026)	Selection implication
Food and beverage	Food safety, traceability, allergens, cleaning, high-speed packaging	Lot genealogy across batch and pack, holds and release, allergen changeover control, downtime	FSMA 204 traceability rule; 21 CFR Part 117 records	Hybrid capability and fast trace queries matter more than paperless records
CPG and packaging	Throughput and cost on high-speed lines	Performance and downtime, changeovers, labor	Customer and retailer requirements	Often starts with performance; a full suite may not be needed
Pharma and biopharma	Patient safety, validated records	Electronic batch records, review by exception, weigh and dispense, equipment status, deviations	21 CFR Part 11 and 211; EU GMP Annex 11; GAMP 5	Validation effort and record design dominate cost and time
Dietary supplements	Batch records and identity testing	Master and batch production records, holds	21 CFR Part 111	Batch-record discipline without full pharma cost
Medical device	Device history and quality system	Device history records, serial traceability, nonconformance	FDA QMSR (effective February 2026, aligned with ISO 13485)	Discrete MES with strong quality integration
Automotive	Just-in-time supply, part traceability	Serial and lot traceability, sequencing, error proofing	IATF 16949 (8.5.2.1 identification and traceability)	Integration with customer schedules and error proofing
Aerospace	Configuration control, long-lived records	Serial traceability, as-built records, work instructions	AS9100 Rev D	Configuration and record retention over speed

Pharma: records you can defend

Regulated pharmaceutical manufacturing is where MES requirements are most precise. Electronic records must be “trustworthy, reliable, and generally equivalent to paper records” under 21 CFR Part 11, with “secure, computer-generated, time-stamped audit trails”, and the FDA’s 2003 guidance on the scope and application of Part 11 still governs how narrowly it is applied. Batch production records must be prepared for each batch (21 CFR 211.188) and reviewed by the quality unit before release (211.192).

Two points from regulators are worth knowing before talking to any vendor. The FDA has warned that “qualifying the Manufacturing Execution System (MES) platform … does not demonstrate that a given [master production and control record] generated by the MES contains the correct calculations” (FDA Data Integrity Q&A, 2018). A validated platform is not a validated record. And the Pharmaceutical Inspection Co-operation Scheme (PIC/S) allows audit trail review by exception, “focusing on anomalous or unauthorised activities” (PIC/S PI 041-1), which is what makes review by exception, the main payoff of electronic batch records, defensible.

The rules are moving. The FDA’s computer software assurance guidance was superseded in February 2026 by a version covering production and quality management system software. In Europe, Annex 11 dates from 2011; a revised Annex 11, a revised Chapter 4 and a new Annex 22 on artificial intelligence went to consultation from July to October 2025 and were not final as of October 2026. ISPE’s GAMP 5, in its second edition since 2022, remains the common framework for validation.

Food: traceability on a deadline

For food plants the driver is traceability. The FDA’s Food Traceability Rule under FSMA Section 204 requires firms handling foods on the Food Traceability List to keep key data elements, starting with a traceability lot code, for critical tracking events such as receiving, transformation and shipping, to maintain a traceability plan, and to provide records within 24 hours as an electronic sortable spreadsheet when asked (FDA, FSMA 204).

The compliance date is often misquoted. The rule set January 20, 2026. Congress then barred the FDA from enforcing it before July 20, 2028 in the 2026 appropriations act, and the FDA has proposed moving the date formally; as of October 2026 that proposal was not final. For an MES, the practical requirement is the same either way: lot genealogy across the batch and packaging boundary, queryable in hours rather than days. For how an MES supports that in practice, see how an MES supports traceability and compliance, and for a broader view of where a food plant stands, a digital maturity assessment for food and beverage plants.

How industry changes the selection process

Industry changes the order in which the questions get asked, not just the answers:

• Pharma and biopharma: the vendor’s own quality system, validation documentation and supplier assessment come early, and the implementation plan includes validation from the start. After go-live, configuration changes go through change control, so a product that needs heavy configuration keeps costing more to change for years. Agree the electronic batch record design, and what review by exception will cover, with quality before the shortlist.
• Food and beverage: start with traceability. Ask each vendor to run a mock trace on a sample of the plant’s own data, forward from an ingredient lot and backward from a pallet, and time it. Then look at allergen changeovers, sanitation records and holds.
• CPG and high-speed packaging: start with performance. How fast can a line be connected, how good is the downtime capture, and how much effort does the operator’s screen take? A full suite may not be needed at all.
• Medical device and aerospace: start with records: device history or as-built records, nonconformance links to the QMS, and how long records are kept and how fast they can be retrieved.
• Automotive: start with sequencing, error proofing and part traceability against customer requirements.

In every case, write the regulatory and customer requirements as pass or fail criteria before the first demo. A demo that impresses on dashboards and fails a criterion is a no, however good the dashboards look.

Who sells MES in 2026? A vendor-neutral map

The market splits into enterprise suites, process and pharma specialists, discrete and electronics specialists, platforms with MES modules, low-code and composable tools, ERP vendors’ shop-floor modules, and OEE or production-monitoring tools that cover part of the job. Ownership has moved a lot since 2023, so check who owns a product, and its roadmap, before you commit to it.

This map is vendor-neutral and as of October 2026. It names representative products in each group; it doesn’t rank them and it doesn’t list prices, which few vendors publish.

Category	Examples (2026 names)	Strengths	Watch for
Enterprise suites	Siemens Opcenter; Rockwell Plex and FactoryTalk ProductionCentre; AVEVA Manufacturing Execution System (formerly Wonderware MES); Proficy (Velotic); Dassault DELMIA Apriso; SAP Digital Manufacturing	Breadth, multi-site standardization, vendor support	Cost, configuration effort, roadmaps tied to the owner’s strategy
Process and pharma specialists	Körber PAS-X; Emerson DeltaV MES (formerly Syncade); Rockwell FactoryTalk PharmaSuite; Honeywell TrackWise Manufacturing; POMSnet	Electronic batch records, validation support, ISA-88 depth	Heavy validation and configuration; best where regulation demands it
Discrete and electronics specialists	Critical Manufacturing; iTAC; Aegis; Siemens Opcenter for electronics; Applied Materials for semiconductors	Serial traceability, routing complexity, equipment integration	Less suited to batch or hybrid plants
Platform with MES modules	Inductive Automation Ignition with Sepasoft MES; Parsec TrakSYS	Flexibility, fast start, strong integration with SCADA	More of the design falls to the plant or its integrator
Low-code and composable	Tulip and similar app-based tools	Fast to deploy, operator-friendly apps, start with one problem	Scale, data structure and governance as apps multiply
ERP shop-floor modules	SAP, Oracle, Microsoft Dynamics, Epicor, Infor modules	One system for planning and execution in simple plants	Limited view of the line; operator burden
OEE and production monitoring	Many vendors	Quick value on performance and downtime	Covers one MES job, not the rest

What changed between 2023 and 2026

Most of what ranks online about MES vendors predates these changes:

Change	What happened	Source
MESA International closed	Ceased operation on 30 June 2026; ISA took over its content	ISA
No Gartner Magic Quadrant	Gartner’s last Magic Quadrant for MES was published in April 2023; since then it has published a Market Guide (May 2025, March 2026) that lists representative vendors without ranking them	Critical Manufacturing on the 2026 Market Guide
Proficy changed hands	GE Vernova sold Proficy to TPG for $600,000,000, completed in March 2026; TPG combined it with PTC’s Kepware and ThingWorx into a new company, Velotic, with Proficy kept as a product line	GE Vernova; Velotic
Schneider Electric to buy PTC	Agreed in October 2026, with closing expected by the third quarter of 2027; Schneider already owns AVEVA	SEC filing
SAP ME and MII retiring	SAP says version 15.5 is the final release of both and they are being retired, with SAP Digital Manufacturing as the successor; maintenance of 15.5 runs to the end of 2030 according to SAP’s community guidance	SAP Digital Manufacturing
Renames	Emerson now presents Syncade as DeltaV MES; Honeywell markets its life-sciences MES as TrackWise Manufacturing; Sepasoft MES 4.0 requires Ignition 8.3	Emerson; Sepasoft
New money	Tulip raised $120,000,000 led by Mitsubishi Electric in January 2026; Körber launched a SaaS edition of PAS-X in May 2026	Tulip

Market size estimates vary with who makes them: MarketsandMarkets put the market at about $15,950,000,000 in 2025, Mordor Intelligence at about $17,270,000,000. Both expect growth of around 9 to 10% a year. The more useful point for a buyer is the churn above: a product’s owner, name and roadmap can change within a single implementation.

What first-hand experience says about the platforms

Joltek’s founder has worked with FactoryTalk View and PlantPAx, Ignition with Sepasoft, Proficy, Wonderware, custom-built MES and MES functions built directly into SCADA screens in FactoryTalk View ME and SE. A few patterns hold across them:

• MES functions built into SCADA are where many plants start, and they work for a line or two: downtime screens, a recipe selector, a production counter. They get hard to maintain when they spread across sites with no common data model.
• Platform-based MES, such as Ignition with Sepasoft, gives a plant control and speed, and leaves more of the design to the plant or its integrator. The database structure and object model are the main thing you buy. As one engineer put it, “You’ll still do a lot of scripting” (Inductive Automation forum, 2023).
• Enterprise suites, such as Proficy and the Wonderware lineage now in AVEVA, standardize well across many sites and carry real configuration effort. They pay off where the same model runs everywhere.
• Custom-built MES fits unusual processes and ages with the people who built it. Its risk is the day those people leave.

The choice between them is less about features than about who will support the system in five years, which the cost section covers.

How to shortlist when ownership keeps moving

Given how much changed in three years, a shortlist should test the vendor as well as the product:

1. Who owns it today, and what has the owner said publicly about its roadmap?
2. Which version would you install, and what is the announced end of support for it?
3. Who implements it in your region, and how many integrators know it well enough to support it if the first one leaves?
4. Where it runs today in your industry and your plant type, ideally at a site you can visit.
5. How it is licensed: per site, per user, per connected machine or by subscription, and what the cost does as you add lines and sites.
6. How you get your data out, in a usable form, if you ever change systems.

None of these appear in a feature matrix, and over ten years each of them matters more than any single feature.

Which KPIs should an MES produce, and who uses each one?

The KPIs that matter most are the ones a team can act on this shift: OEE and its losses, throughput against plan, first pass yield, scrap and rework, changeover time, batch cycle time and right-first-time batch records. ISO 22400 defines 34 of them. Start with five your teams will use, not thirty for a dashboard.

The international reference for MES KPIs is ISO 22400-2, which defines 34 key performance indicators for manufacturing operations management, with a second edition in development (ISO 22400-2). The names below follow the standard where it has one.

KPI	ISO 22400 name	Plain formula	Data the MES needs	Who uses it	Cadence
OEE	OEE index	Availability × effectiveness × quality ratio	States, counts, rejects, ideal rate	Shift lead, CI, management	Shift, day
Availability	Availability	Run time ÷ planned production time	States and planned time	Shift lead, maintenance	Shift
Performance	Effectiveness	Actual output ÷ output at rated speed in run time	Counts, rated speed	Shift lead, process engineers	Shift
Quality	Quality ratio	Good units ÷ total units	Good and reject counts	Quality, operators	Shift
Throughput against plan	Throughput rate (plus schedule attainment)	Quantity made ÷ quantity planned	Orders, counts	Planners, shift lead	Shift, day
First pass yield	First pass yield	Units passing first time ÷ units started	Inspection results	Quality, operators	Shift, day
Scrap and rework	Scrap ratio, rework ratio	Scrapped or reworked ÷ produced	Reject reasons, rework records	Quality, CI	Day, week
Changeover time	Setup ratio	Setup time ÷ production time	Order changes, states	Shift lead, CI	Changeover
Batch cycle time	(batch practice)	Batch end − batch start, by phase	Phase start and end events	Process engineers, planners	Batch
Right-first-time batch records	(batch practice)	Batches released without deviation or correction ÷ batches made	Deviations, record corrections	Quality	Batch, week
Reliability	MTBF, MTTR	Run time ÷ failures; repair time ÷ failures	Fault events, repair times	Maintenance	Week, month

The order of rows is deliberate. The first ones are what a shift lead can act on today; the last are for the weekly meeting. A plant that builds the management roll-up first and the shift view never is the plant from the start of this guide.

OEE is the KPI most plants start with, and the one most often misread. For how OEE works from the machine to the management meeting, including why it should be measured per machine and not only per line, see how OEE works from the machine to the management meeting. Before founding Joltek, Vladimir Romanov worked at a Kraft Heinz plant where OEE rose from about 40% to about 48% through data systems, root cause analysis and operating discipline. None of it started with a new dashboard. It started with knowing which machines were losing the time, and why. For the cost side of the same losses, see the causes and real cost of downtime.

Downtime reasons: the data behind most of these KPIs

Most of the KPIs in the table depend on two pieces of data: what state each machine was in, and why. Getting those right matters more than any dashboard:

• Take states from the controller, not from people: running, stopped, starved, blocked, faulted. A machine knows its own state far better than an operator remembers it at the end of a shift.
• Take reasons from fault codes where the machine knows them, and ask operators only for what it can’t know, such as waiting for materials or for a quality check.
• Keep the reason list short, two levels at most, so an operator can pick one in seconds. Long lists end up with most stops filed under “other”.
• Separate planned from unplanned stops, so cleaning and changeovers don’t hide inside availability losses.
• Measure per machine, not only per line, so a blocked filler isn’t blamed for a stopped capper downstream of it.
• Set the ideal rate per product, because performance is only as honest as the rate it is measured against.

Plants that do this get KPIs a shift lead trusts. Plants that don’t get numbers the morning meeting argues about.

What does an MES cost, and should you build, buy or compose?

Licences are usually the smaller part. Integration, master data clean-up, infrastructure, validation in regulated plants, training and years of support make up most of the cost, and they’re the parts proposals understate. The options run from extending SCADA and building locally, to a platform with MES modules, to low-code tools, to an enterprise suite. Each fits a different plant.

There is no honest single price for an MES, and figures that circulate online rarely trace to a source. What can be described is where the money goes:

Cost element	What drives it	Usually underestimated by
Software licences or subscriptions	Sites, users, modules, connected equipment	Rarely; it’s the number everyone sees
Integration	ERP, controls, LIMS, QMS, WMS, CMMS; how many interfaces and how standard they are	Proposals that assume clean, documented interfaces
Data access and infrastructure	Network upgrades, gateways, servers, controller changes, OT security	Everyone, because it’s discovered on site
Master data	Products, recipes, routings, equipment model, reason codes, all cleaned and agreed	Plants, which assume ERP master data is ready
Validation	Regulated plants: specifications, testing, documentation	Teams new to regulated software
Training and change	Operators, supervisors, quality; new ways of working	Projects that treat go-live as the finish
Sustaining support	The people, upgrades and vendor support needed for the next ten years	Almost everyone

The last row decides more than the first. A system nobody can support after the integrator leaves becomes the next project.

Build, buy or compose

Option	Fits when	Watch for	Who supports it
MES functions in existing SCADA (for example FactoryTalk View ME/SE or Ignition screens)	One or two lines, a few functions, a strong in-house controls team	Spreading across sites without a data model	In-house controls engineers
Custom build	Unusual processes no product fits	Dependence on the builders; documentation	The builders, for as long as they stay
Platform with MES modules (for example Ignition with Sepasoft)	Plants that want control, speed and good SCADA integration	More design work falls to the plant or integrator	Plant team plus an integrator
Low-code or composable apps	Starting with one problem, fast; operator-friendly screens	Data structure and governance as apps multiply	Plant team, with vendor help
Enterprise suite	Many sites, one standard, regulated records	Configuration effort, cost, the owner’s roadmap	Vendor plus a specialist integrator
ERP shop-floor modules	Simple discrete plants already on that ERP	Operator burden, limited view of the line	IT and the ERP partner

The build-versus-buy arithmetic is clearer from people who’ve done both. One integrator wrote that “you need to be brutally honest in your cost analysis”, and that even at a professional day rate, “I doubt I could produce a product to the quality levels … of Sepasoft before I hit the Sepasoft purchase price”. Another described a custom build the customer loved, “Until they wanted upgrades ($) and expansions ($$) and custom reports ($$$ !!)” (Inductive Automation forum, 2021). Both are right. Building is cheap to start and expensive to own; buying is the reverse.

On-premises, cloud or hybrid

Where the MES runs is now part of the choice, and it changes both cost and risk. Even products sold as cloud services usually keep some functions at the plant, because a line can’t wait for a round trip to a distant data center and has to keep running when the connection drops.

The common pattern is a split:

• At the site, on plant servers or edge devices: dispatching, recipe download, data collection and genealogy capture, the functions the line depends on minute by minute.
• In a data center or the cloud: reporting across sites, analytics, master data and the long-term archive.

Regulated manufacturers are moving the same way; Körber’s SaaS edition of PAS-X is one sign. Four questions sort the options quickly:

1. What keeps running at the plant when the link to the data center or cloud drops, and for how long?
2. Where does the production record of truth live, and how is it backed up?
3. Who patches and upgrades the plant servers, and how is that scheduled around production?
4. In a validated environment, how are vendor-pushed updates controlled?

Subscription pricing also changes the build-versus-buy arithmetic above: it lowers the cost to start, and it turns sustaining support into a line item that never ends.

Whichever option a plant chooses, the integration work usually goes to a systems integrator, and the scope decides how much of the risk the plant keeps. For how to write that scope, see writing an integrator scope of work for MES.

Why do MES projects fail, and how do you start one that works?

Most MES programs don’t fail on software. They fail on an overly ambitious plan that never gets past the roadmap, on plant-floor data nobody can reach, on master data that isn’t ready, and on systems operators have to work around. Start with one line and one problem, deliver something the shift uses within weeks, then expand by function.

The overly ambitious plan

The most common failure Joltek sees is not a failed go-live. It is a plan that never materializes. A multi-site roadmap covering every MES function, written before anyone has connected a single line, sized to impress a steering committee, and abandoned somewhere in year two when the budget or the sponsor moves on. Compare that with the realistic pace from the connectivity section: roughly one to three months per site for a focused rollout, set by the lines, the controllers and the network. A plan built on that pace delivers. A plan built on the steering committee’s ambition usually doesn’t.

The numbers everyone quotes

Many MES business cases still cite benefits such as a 45% cut in manufacturing cycle time or a 75% cut in data entry. Those figures come from MESA International’s surveys of the early and mid-1990s, sent by vendor members to their own customers. MESA’s own paper said the data “cannot reflect a statistically meaningful average or consensus”, and several figures in the summary everyone repeats don’t match the surveys they came from (MESA White Paper 6, 1997). A later MESA survey in 2006, sponsored by vendors and answered by 135 manufacturers, found that 40% of plants with an MES did not measure its return at all. And no verifiable MES failure rate exists: the percentages that circulate don’t trace to a study. A business case should rest on the plant’s own losses, not on 1990s averages.

The other ways projects stall

• The data. Everything in the connectivity section above. If a project plan doesn’t name the controllers, the network changes and the server work, it hasn’t started yet.
• Master data. Recipes, routings, bills of materials, equipment models and downtime reasons that differ between ERP, the line and the people. One practitioner summed it up: “Would love to utilize MES, but that’s gonna require changing the BOMs” (EpiUsers, 2020).
• Operator burden. An MES that makes operators “leave the machine, walk to a data collector, end activity, start activity and then back to the machine” (EpiUsers, 2022) will be worked around, and data that people keep on paper alongside the system means it “isn’t really “truth” anymore” (EpiUsers, 2026).
• Ownership. Nobody decides between operations, quality, IT and engineering, so every decision goes to a committee.
• The sustaining team. The system goes live and the people who understood it move on. Five years later nobody can change a report.

How to start one that works

1. Pick one line and one problem the shift has. Unexplained downtime, paper batch records that take days to review, holds that miss pallets. Not “visibility”.
2. Map where each MES job lives today on that line, using the table earlier in this guide.
3. Fix the data access for that line: the network, the controllers, the servers. Budget the one to three months it realistically takes.
4. Clean the master data that line needs: products, recipes or routings, the equipment model, downtime reasons.
5. Deliver one function the shift uses every day within weeks: downtime by reason on the line, an electronic batch record for one product, genealogy for one product family.
6. Expand by function, then by line, and standardize what worked before taking it to another site.

How Joltek assesses the MES layer

When Joltek assesses a plant’s MES layer, it scores nine dimensions, each from 1 to 5. They make a useful self-check before any selection:

• Production visibility: can the people running the line see what is running, against plan, right now?
• Performance measurement: are downtime, OEE and losses measured consistently, per machine and per line?
• Quality operations: are checks, holds and release recorded and enforced, not just written down?
• Traceability: can the plant trace a lot or unit forward and backward in hours?
• Recipe and specification control: are recipes and settings controlled, versioned and delivered to the equipment?
• Digitization of operations: how much of the shift still runs on paper?
• Data model and naming: is there one structure for equipment and data across lines?
• Integration architecture: are ERP, controls and quality systems connected by design or by workaround?
• Data quality and governance: does someone own the data, and is it trusted?

A plant that scores low on the last four will get little from any MES until those are addressed. That is the work behind Joltek’s independent MES selection work: start from what the plant already does and where its data really lives, then choose.

Frequently asked questions

What does MES stand for in manufacturing?

MES stands for manufacturing execution system: the software layer that runs and records production between the business plan in ERP and the machines on the floor. It turns orders into work for the line, tracks materials, quality and results, and keeps the production record. The ISA-95 standard calls the same layer manufacturing operations management, or MOM.

Can ERP replace an MES?

Sometimes. A simple discrete plant making a stable range of products to stock can often run on its ERP’s shop-floor modules. Plants with regulated records, batch recipes, high product mix or high-speed lines usually can’t, because ERP doesn’t see what happens on the line minute by minute and can’t enforce how the work is done.

What is the difference between MES and SCADA?

SCADA monitors and controls equipment second by second: it starts and stops machines, shows alarms and records process values. An MES works at the level of the production order: which order runs where, with which materials and people, to what quality, and the record of all of it. Most plants need both, connected.

Do small manufacturers need an MES?

Every manufacturer does the jobs an MES does: scheduling, tracking materials, checking quality and recording results. Whether those jobs need a dedicated system depends on regulation, product complexity and how much still runs on paper. Many smaller plants start with one function, such as downtime tracking or electronic batch records, before buying a full suite.

How long does an MES implementation take?

In Joltek’s experience, a focused rollout takes roughly one to three months per site. The range depends on the complexity of the lines, the age of the controllers, the state of the network, and how much work it takes to reach the data and deploy servers. Plans that try to cover every function at every site at once take far longer, and often never finish.

Can an MES connect to old equipment?

Usually, yes, through protocol gateways, drivers, added sensors or remote I/O, or by reading the HMI or SCADA layer that already talks to the machine. It is rarely the software that limits this. Old controllers, incompatible protocols, network gaps and locked-down production areas are where most of the effort in an MES project goes.

Start small, and start on the floor

A manufacturing execution system is worth what the people running the plant do with it. The plants that get value start with one line and one problem the shift actually has, connect the data properly, and expand by function once the first piece is in daily use. The plants that don’t usually started with a roadmap for every site and a dashboard for every executive.

Three things decide most outcomes, and none of them is the vendor: whether the plant can reach its own data, whether the master data is clean and agreed, and whether the first function delivered is one the shift actually uses. Get those right on one line, and the choice of software becomes a smaller decision than it looked at the start.

If you’re choosing an MES, replacing one, or trying to get value from the one you have, Joltek’s independent MES selection work starts with what your plant already does and where its data really lives. Send Joltek the situation.