Four Criteria for Comparing MES Options for Thai Factories

You set out to compare MES products, lined up the brochures from every vendor, and then stalled. We hear this often from production engineering and IT managers at Japanese-owned factories in Thailand. Every feature list is filled with the same vocabulary, and almost every price is “contact us for a quote.” This article lays out how to put several MES candidates side by side and narrow them down to the one that fits your plant, working through four evaluation criteria, the two broad types of vendor, the choice of deployment model, the checks that matter specifically for a Thai site, and how to run the RFP.

What an MES is, and why comparing products is so hard

MES stands for Manufacturing Execution System. It is the layer that captures what is being made right now, on which machine, by whom, and in what quantity; issues work instructions; records actuals and quality data; and makes the history of each individual product traceable.

The reason so many people stumble at the start of the comparison is that the scope covered by the term MES varies enormously from product to product. Some products are built around actuals collection and OEE measurement. Others put material traceability and quality records at the core. Others again take on the distribution of work instructions and the pass/fail judgement of in-process inspection. The sign over the door says MES in every case, but the centre of gravity is different. If you line up feature lists without understanding that, every product starts to look the same.

The difference between MES and ERP is money versus material

Sorting out the ERP question early moves the whole discussion forward. The relationship is often summarised as ERP managing money while MES manages material. ERP takes a monthly or yearly view and manages people, materials and money as corporate resources, optimised for the company as a whole. MES works on the shop floor with operators and equipment, using real-time information at the level of minutes and seconds to control execution itself.

That difference in time horizon is what matters when you compare. Someone will always ask whether the production module inside your existing ERP could do the job instead. In practice ERP production planning deals with the world of planning and costing, and is not designed to absorb signals arriving from equipment every second. Expecting an MES to handle company-wide accounting is equally misplaced.

If you want to firm up where MES and ERP sit within the wider production management landscape first, we cover that systematically in What a production management system is and how it differs from ERP and MES. This article picks up one step further along, focusing purely on how to compare several products within the single category of MES.

A production scheduler plays a different role again

The other thing that gets confused with MES is the production scheduler, usually called APS. APS is a tool for building a plan that respects constraints, deciding when, on which line, and in what sequence work should run. MES is the tool that executes that plan on the floor and collects the results. If what you actually want to compare is planning software, the evaluation criteria themselves change. We treat the APS criteria separately in Comparing production schedulers for factories in Thailand, so deciding up front whether your target is MES or APS will save you a round of pointless proposal requests.

Three structural reasons MES comparison is difficult

First, pricing is essentially never published. The major MES packages covered in the Japanese market all list their pricing as available on request. Surveys of overseas vendors show charging models split across per-user, per-line, per-plant and enterprise agreements, with no single market rate. In other words, you cannot use price as your first filter.

Second, the granularity of feature names is not consistent between vendors. What one product covers in a single line called quality management, another splits into three lines for in-process inspection, defect analysis and corrective action management. Counting rows tells you nothing about the breadth of the functionality.

Third, industry specialisation does not appear on the surface of the material. A product strong in moulding and die casting, one strong in high-mix low-volume assembly, and one strong in batch production for chemicals and food all have different internal data models. Yet the heading on every catalogue reads “manufacturing execution system.” These three points are exactly why you have to bring your own common yardstick to the comparison.

Four evaluation criteria you can use to compare MES products

The criteria used in practice across MES comparison work fall into four groups: functional coverage, data collection method, multi-site capability, and industry specialisation. Scoring your candidates on these four keeps you from being pulled along by how the brochure looks.

Criterion 1 is functional coverage, measured against MESA-11

When you want a common yardstick for breadth of function, use the MESA-11 model. It is a framework of eleven core functions defined in 1997 by MESA International (Manufacturing Enterprise Solutions Association International), organising the role of an MES from three viewpoints, namely material, people and process. Because products differ in how much of those eleven they cover, the model works directly as a shared axis for comparison.

Some products on the Japanese market state explicitly that they cover ten of the eleven, excluding scheduling. Asking every vendor how many of the eleven they cover lets you rank them by real scope rather than by the number of rows in a feature table.

Attach this table to your request for proposal and have each vendor fill it in, and the distribution of supported, partially supported and not supported becomes obvious. The performance analysis row in particular is worth pressing on down to the definition of each indicator. OEE is a number that changes with how the formula is drawn, so whether your own definition matches the product’s standard definition is something to settle early. We cover the OEE formula itself and how to drive improvement in OEE improvement in practice on the factory floor.

Criterion 2 is data collection, which decides how the system is actually operated

The value of an MES is determined by how little effort it takes to get shop floor data into it. Collection methods split broadly into automatic capture from equipment, barcode or RFID scanning, entry from smart devices, and entry from touch panels installed on the floor.

The question to compare is not whether a method is supported, but how much of your specific equipment base can realistically be automated. In a plant with a lot of older machines, only a limited number of them can give up a signal at all. In that case, however rich the automatic capture functionality is, only part of it will ever be used and the rest falls back on people. Once manual entry is a given, practical factors matter more: how easily entry screens can be built, whether buttons are large enough to press with gloves on, and how the system behaves when it goes offline.

Criterion 3 is multi-site capability, which drives future integration cost

Even a company with a single site in Thailand cannot ignore this one. If other group sites in the region or the parent company in Japan run a different system, the day will come when the data has to be reconciled. Three things to check: whether master data can be managed across sites, whether processes that differ by site can be expressed within the same product, and whether headquarters can view actuals through the cloud.

Criterion 4 is industry specialisation, which shows its effect last

Moulding and die casting need mould management, chemicals and food need batch and formulation management, assembly lines need traceability at the level of the individual unit. The data model each industry requires is different. Bolting these onto a general-purpose package after the fact tends to inflate customisation cost and create a structure where every version upgrade brings additional charges. Keeping products with reference installations close to your own industry on the shortlist makes the quotation comparison that follows far easier.

Global vendors versus domestic packages

At the stage of assembling candidates, the first thing most people agonise over is whether to look at the large global names or at Japanese packages. The short answer is not to commit to either camp, but to put a few products from each into the same comparison on the same criteria.

A map of the global vendors

A comparative study of overseas MES vendors evaluated thirteen suppliers. In that study the Leaders group contains SAP, TrakSYS, Solumina and Wonderware MES; the Contenders group contains Critical Manufacturing, DELMIA Apriso and Tulip; and the Challengers group contains FactoryTalk ProductionCentre, GE Proficy, iTAC.MOM.Suite and Aegis FactoryLogix.

The point to watch is that these groupings are not a ranking of good to bad. The same study notes that vendor strengths diverge by industry. GE Proficy and AVEVA are described as strong in process and pharmaceutical manufacturing, Critical Manufacturing and Aegis FactoryLogix in discrete and electronics manufacturing, and Solumina in aerospace. Working backwards from your own industry, it is entirely normal for a product outside the Leaders group to be the better fit.

Representative domestic packages

Comparison coverage of the Japanese market introduces fourteen products, sorted into packages for large-scale deployments and packages aimed at small and mid-sized plants. Here are representative examples, together with the characteristic worth spotting during comparison.

None of these products publishes a price; all are quoted on request. What you should read out of this table is therefore not cost but the picture of the shop floor each product was designed around. A product that carries mould management as standard has its data model built on the assumptions of a moulding operation. A product that leads with unit-level traceability assumes serial management on an assembly line. How closely your own products and processes match those assumptions is what determines the volume of customisation later.

Which camp to build around

There are three inputs to that decision. The first is whether standardisation across the group is planned. If headquarters or another country site has already settled on a particular global product, aligning with it lowers future integration cost. The second is the operating language on the floor. If operational design and documentation in Japanese are a precondition, a domestic package will get you live faster. The third is how much customisation you are willing to accept. Whether you can choose a product that covers roughly eighty percent of the requirement as standard and absorb the rest through working practice changes which product you should pick.

Choosing between cloud, on-premises and hybrid

MES deployment models split into on-premises, cloud and hybrid. For a site in Thailand, the decision becomes much faster once you treat it not as a question of which is more modern, but as a question of operational capability, namely who can restore service and how quickly when it stops.

One circumstance specific to companies based in Thailand is local hardware maintenance. If you choose on-premises, you need to pin down at contract level who physically comes to site when a server or network device fails. An arrangement that would deliver a replacement unit the same day in Japan can take a different number of days in Thailand because of customs clearance and local stock.

Cloud, on the other hand, keeps up-front investment low and makes access from multiple sites including headquarters straightforward. But if you hand the parts that touch equipment control directly over to the cloud, the state of your line becomes a direct input to production output. In practice most projects settle on a hybrid configuration, where data collection from equipment and display on shop floor terminals stay inside the plant, while aggregation, analysis and cross-site access run in the cloud.

Extra angles to check when comparing MES for a factory in Thailand

From here on are the checks specific to a Thai site that rarely come up in a purely domestic Japanese selection. Whether or not you write this section into your request for proposal changes the quality of the proposals you get back.

How the Thai MES market is moving

Market research covering Thailand’s manufacturing execution system market for the period 2025 to 2031 has been published. The growth drivers it identifies are adoption of Industry 4.0 technology under the Thailand 4.0 policy, the need for operational efficiency and cost reduction, and government measures promoting digitalisation and automation.

The challenges it names just as clearly are high initial implementation cost, resistance and a skills gap on the shop floor, and concerns over data security. Those same three points are what you will be asked to explain in your internal approval process. Named major vendors include ABB Ltd. and Rockwell Automation (Thailand) Ltd. Cited adopters include Thai Union Group, the seafood processing major, which introduced a cloud-based MES to improve production efficiency, along with Delta Electronics and Thai Beverage. The simple fact that vendors with local reference installations exist is a precondition for comparing support arrangements at all.

On multilingual UI, the question is not whether it translates but who maintains it

Plenty of products advertise multilingual support, but the number of supported languages is not the thing to ask about. What to ask is where the boundary sits between the wording supplied by the product and the wording you register yourself as master data.

Most of the text actually read on the floor is not the product’s own menu names but data you register: process names, defect code names, notes attached to work instructions. If none of that is entered in Thai, it does not matter how thoroughly the product menus have been localised, because operators still cannot read it. Confirm during comparison who will add and amend Thai wording once you are live, and whether that work requires help from the development vendor.

Connectivity with existing PLCs and sensors

Factories in Thailand commonly run a mix of equipment relocated from Japan, equipment procured locally, and equipment assembled to a machine builder’s own specification. When you take MES proposals, hand over the equipment list together with a document setting out what each communication interface actually is, whether industrial Ethernet, serial, or nothing more than a dry contact output. The precision of the proposals improves substantially.

Where a lot of equipment offers only contact output, the cost driver is not the MES itself but how you build the mechanism that picks up those signals and passes them upward. Whether the MES vendor takes that on in-house or assumes it will be carved out to another company changes the scope of the quotation, so always confirm it.

Whether local support exists on the ground

An MES is a system that stops production when it stops. We recommend keeping first-line incident ownership, the hours of coverage in Thai time, and whether engineers can attend site as a separate axis alongside the product comparison itself. Even a strong global product can take several days just to complete first-line triage if there is no direct contact point in Thailand and the route runs through a distributor and a trading company in Japan.

How far to push OT security

As the Thai market research flags data security concerns as a challenge, introducing an MES is also a conversation about changing the architecture of your factory network. Equipment that has been running on a closed network will now connect to the information network through the MES.

Four points are worth putting on the table from an OT security perspective: how the design separates the information network from the control network, whether the MES server sits on the equipment side or the information side, how the route is built for the vendor to reach the factory network for remote maintenance, and how account management is handled. The remote maintenance route in particular is too late to design once all the operational decisions have been made; it should be confirmed with your IT department before contract.

Cost expectations and how to run the RFP

How to think about the range

MES cost is driven less by the product itself than by the deployment model and the volume of bespoke build. As a reference point, quotation benchmarks for MES built on large-scale custom development, such as for semiconductor manufacturing, put initial cost per site at roughly JPY 30 million to JPY 150 million with an implementation period of around six to twelve months. That figure covers current-state analysis, consulting, software, customisation, training, hardware and environment build.

That number represents the level for large-scale custom development in sectors such as semiconductor manufacturing. It does not mean every MES implementation lands in that range. Products offering a cloud-based modular approach with a small start, or deployment in as little as one month, leave room to begin from a far smaller initial investment. Since actual pricing is quoted on request from every vendor as noted above, all that can be said from public information is the difference in order of magnitude, namely that the volume of bespoke build changes the number of digits in the initial cost.

Not mixing those two levels together matters when you explain the budget internally. Presenting large-scale custom development figures as the general market rate for MES will not only sink your approval request, it will also remove the small-start option from consideration entirely.

Send the same RFP to several vendors

The most reliable way to sharpen the comparison is to write a request for proposal and submit identical content to several system integrators. Because every proposal comes back against the same requirements, it is far easier to compare them on aligned evaluation criteria. Do the opposite, and explain things slightly differently to each vendor in individual conversations, and the assumptions behind each proposal diverge until neither the price nor the functional scope can be compared.

The timeline for the selection process

Research on MES selection practice offers rough guidance on the timeline. It puts RFP preparation at two to three weeks, the shortlist at three to seven vendors, and the whole selection process at around six to ten weeks. The same research lists twenty-one evaluation items forming the core of the selection criteria, including production execution, quality management, material traceability, equipment connectivity, real-time visibility, regulatory compliance, ERP integration, and user interface and mobile support.

Note that this timeline does not include the time needed to build internal consensus. For a site in Thailand, approval processes at the Japanese parent company and the constraints of the budget year are added on top, so in practice it usually runs longer. Put the other way round, planning on the comparison work itself finishing within roughly two months keeps the evaluation from sitting idle while approvals are pending.

Pulling the criteria and check points together

Here is the whole discussion in a form you can transfer straight into a request for proposal.

The last two rows of this table tend to be treated lightly at comparison time, but they are what bites a few years after go-live. Cost structure in particular is not a real comparison unless you line the products up on a total of roughly five years including annual maintenance and the cost of adding sites, not just the initial cost.

What happens after the award

Once you have compared and narrowed down to one vendor, a different body of work is waiting: negotiating contract terms, designing FAT and SAT, and acceptance judgement after go-live. That stage demands different preparation from comparison, so we treat it separately in How to run ordering, FAT and SAT and acceptance for a successful MES implementation. Even at the comparison stage, knowing what you will be examining at acceptance makes it much easier to settle the level of detail your request for proposal needs.

What to keep in view for MES selection from here

2026 is described as the year in which falling inference costs let AI agents start spreading onto the shop floor. Implementation concepts presented for manufacturing describe an AI agent receiving a signal of equipment abnormality, working with the MES and core systems to analyse the order backlog, delivery dates and the utilisation of alternative lines, then proposing a process change on its own initiative and autonomously placing orders for the materials required.

You do not have to decide today how much of that your own plant needs. From the standpoint of MES selection, however, whether a product has the data structure and API connectivity to work with such mechanisms in future is becoming part of the comparison. Concretely, that is three things: whether actuals data can be extracted mechanically from outside, whether work instructions can be written in from an external system, and whether the meaning of the data, that is which field represents what, is documented.

Even a product that makes no claim about AI today can be connected later if its data interfaces are properly organised. Choose a product where data can only be pulled off the screen, and that decision becomes a constraint a few years out. In the comparison meeting, asking “in what format can you export actuals data” will get you a far more useful answer than asking “do you support AI.”

Frequently asked questions

Should we implement MES or ERP first

There is no general right answer; it depends on what you currently cannot see. If the problem is that cost and inventory figures only appear after the books close, ERP comes first. If the problem is that nobody knows what happened on the floor until the next day or later, MES comes first. We would encourage caution on plans to introduce both at once at a Thai site, because the load on the shop floor rises sharply all at the same time.

Do small and mid-sized plants need an MES

Judge on the level of traceability demanded of you and the complexity of your processes rather than on size. If customers are asking for per-unit history, or you can only identify in-process defects after the fact, implementation makes sense even at a small scale. Products on the Japanese market include cloud-based modular offerings that allow a small start and others that promise short deployment, so starting with a narrowed set of functions is a realistic option.

What should we do first to start comparing MES for a Thai factory

Before collecting product brochures, write down your equipment list with communication protocols, how records are kept today, and no more than three specific problems you want the system to solve. With those three things in hand, the skeleton of your request for proposal is essentially finished. Listen to vendor presentations while those points are still vague and the product with the most features will simply look best, which is not a comparison.

Which is safer, an industry-specific package or a general-purpose one

If a specialised package exists close to your industry, it is worth putting on the shortlist first. When industry-specific concepts are already standard functionality, customisation volume drops and additional charges at version upgrades are contained. That said, specialised packages can become constricting if the products or processes you handle change substantially later. Add to your inputs whether your product mix is likely to shift significantly over the next five years or so.

How long should we allow for the comparison

Overseas research puts request for proposal preparation at two to three weeks and the whole selection process at around six to ten weeks. Approval at the Japanese parent company and budget allocation are added on top, so in practice allowing three to six months keeps the plan from breaking down.

Summary

MES comparison is difficult because of three structural conditions: prices are not published, feature naming granularity is inconsistent, and industry specialisation does not surface in the catalogue. What works in that situation is bringing your own common yardstick.

Concretely, measure functional coverage against the eleven MESA-11 functions, measure data collection by how much of your own equipment can be captured automatically, measure multi-site capability against the possibility of adding sites later, and measure specialisation by reference installations in your own industry. For a site in Thailand, add four more items to those four criteria: multilingual operation, connectivity with existing equipment, local support arrangements, and OT security. Then issue an identical request for proposal to three to seven vendors and narrow down within a six to ten week window. On cost, keep the large-scale custom development level and the small-start level apart, and line the candidates up on a five-year total. Follow that sequence and you get a comparison that is not swayed by how the brochure looks.

Comparing MES products looks like the work of choosing a product, but it is really the work of describing your own shop floor. If you are already stuck at the stage of organising the equipment list or putting the problems you want solved into words, we are happy to work through that with you from the beginning. TOMAS TECH works on the floor with Japanese-owned manufacturers in Thailand, and takes on consultations covering everything up to connectivity with existing equipment and multilingual operation. It is entirely fine if you have only just begun looking internally and have neither candidates nor a budget fixed. If anything here raises a question, please get in touch through our contact page.

References

• The 11 functions of an MES and an explanation of the MESA-11 model
• Manufacturing Execution Systems vendor comparison and RFP guide
• A comparison of 14 manufacturing execution system products
• The difference between ERP and MES
• How to write an RFP and submit it to multiple companies
• Thailand Manufacturing Execution Systems Market 2025-2031
• Cost benchmarks for MES development in semiconductor manufacturing
• A 2026 guide to using AI agents in manufacturing