On a metal fabrication floor, cutting, stamping, bending, welding and machining happen in-house, but plating, heat treatment and other surface processes almost always leave the building and come back. That single fact changes everything about production control. Whether your production management system for metal fabrication actually fits the way work flows through those in-house and outsourced steps shows up directly in your on-time delivery rate and in how much you can trust your cost figures. This article works through the problems that are specific to metal fabrication — invisible outsourced operations, setup time, and dual units of measure — then sets out the requirements a system needs to meet and how to run the implementation. It also covers the additional realities faced by plants operating in Thailand and across ASEAN.
Why production management is genuinely harder in metal fabrication
Reading through the feature list in a production management system brochure will not tell you what makes metal fabrication difficult. The difficulty does not come from whether a feature exists or not. It comes from the shape of the process itself. Four issues account for most of the trouble.
The moment a part goes outside, it disappears from the schedule
The deepest problem in metal fabrication production control is the inability to track outsourced operations and in-house operations as one continuous flow. Surface treatment, plating and heat treatment are typically handled by partner shops because the equipment is not economical to own. In most plants, in-house operations are tracked in the production management system or on a whiteboard on the floor, while the outsourced steps live somewhere else entirely — in purchase orders, goods receipt slips, or in one person’s head.
Once that split exists, nobody can answer the two questions that matter most: where is this part right now, and when does it come back. It is common to find plants where every delivery inquiry from sales triggers a phone call to the partner shop to ask how many days are left, followed by a manual update to a spreadsheet. The wasted effort is only part of the cost. When queue time at an outside processor is invisible, you cannot separate a delivery delay caused by insufficient internal capacity from one caused by congestion at a supplier, so you end up applying the wrong fix.
There is also an accounting question. While material is sitting at a partner shop, where does it live in your own books? If it is unclear whether it is work in process or consigned inventory held under separate control, every physical count produces a variance. Building outsourced operations into a production management system does not mean digitizing purchase orders. It means placing each outsourced operation on the same routing as the in-house operations, as one step, with a start date, a planned completion date, and an actual date.
Setup time is never “thirty minutes, once”
Metal fabrication is a business of changeovers. Die changes on the press, tooling setup on the press brake, fixture swaps and program calls on the machining center, welding jigs being exchanged. The further you move toward high-mix, low-volume production, the more changeovers you accumulate in a single day.
This is where the arithmetic of accumulation becomes brutal. If one setup takes thirty minutes and setups occur forty times a day, you lose twenty hours a day to setup alone. Thirty minutes does not feel long to anyone on the floor, and forty changeovers a day is a perfectly realistic number in a high-mix shop. Multiply them together and the loss becomes impossible to ignore relative to available operating hours.
The real problem is that those twenty hours are usually not recorded anywhere. If the daily work report has no line item for setup, setup time gets absorbed into run time, and the shop reaches the wrong conclusion — that a given part number is simply slow to machine. Should you consolidate lots? Invest in quick-change or setup-free tooling? Reschedule so similar parts run back to back? Whichever lever you choose, the prerequisite is the same. Setup time has to be recorded by operation, by part number, and by operator.
Weight-based and piece-based units run in parallel
In metal fabrication, material is bought in kilograms and tonnes, while work instructions and inventory are managed in sheets and pieces. Buy a tonne of steel plate, decide how many blanks come out of it, and determine how many finished pieces those blanks become. The conversion is routine on the floor, but if the system does not automate it, a human has to intervene at every conversion point.
Unit conversion errors rarely produce one dramatic incident. They accumulate quietly as inventory drift. The yield setting does not match reality. Remnant handling varies from person to person. The gap between actual material weight and nominal specification weight is never absorbed anywhere. These small discrepancies compound month over month and reappear at every stock count as an unexplained variance. Once variance becomes normal, the inventory data itself loses credibility, and the floor goes back to walking out and looking at the physical stock. At that point most of the value of the production management system is gone.
One rush order sets off a chain of replans
Metal fabricators are frequently asked for short lead times, and when one urgent order cuts into the schedule, other orders get pushed back. If a pushed-back order has an outsourced operation ahead of it, the drop-off date at the partner shop shifts, the return date shifts, and the customer delivery date shifts with it. None of this is visible at the moment the rush order is accepted.
Handling that chain requires more than the ability to drag operations into a new sequence. You need to be able to simulate which operations on which orders move by how many days when one order is pulled forward, and the calculation has to include the outsourced steps. Put the other way round, a system without that capability leaves every rush-order decision resting on the intuition of your most experienced planner.
What a production management system for metal fabrication must provide
Turn those problems inside out and the requirements become clear. On top of the usual evaluation criteria for any production management system — requirements planning, inventory control, progress tracking — a metal fabricator should treat the following five as industry-specific must-haves.
| Requirement | What “in place” actually means | What happens without it |
|---|---|---|
| Outsourced operation control | Outsourced steps sit on the same routing as in-house steps, with ship-out date, planned return date and actuals | Nobody knows where parts are, and delivery answers depend on phone calls |
| Job-level costing | Material, processing and outsourcing costs roll up by order or job number | Profitability per part number is unknown until the monthly spreadsheet close |
| Setup time analysis | Setup is recorded separately from run time and can be aggregated by operation and part | Setup loss hides inside run time and improvement targets cannot be identified |
| Die shot count tracking | Cumulative shot counts per die are recorded and maintenance timing can be forecast | Die failure arrives as unplanned downtime and wrecks the delivery plan |
| Automated unit conversion | Kilograms and tonnes convert to sheets and pieces with yield built in | Inventory drift accumulates until the data is no longer trusted |
Of these five, job-level costing is the one most often included as standard in off-the-shelf packages. Outsourced operation control and setup time analysis vary enormously in depth from product to product, and die shot count tracking and automated unit conversion are frequently offered as options or add-on modules rather than core functionality.
Do not let “outsourced operation control” be redefined as “purchasing”
A common trap during selection is having outsourced operation control explained to you as part of the purchasing module. Issue a purchase order, receive the goods, book the payable. That is a procurement flow, not process control. What process control requires is the ability to see, tied to a specific job, when the material went out, when it is due back, and where it is right now.
The reliable way to test this in a demo is to pick one real job and ask the vendor to display a routing on screen that mixes in-house and outsourced operations. Then ask what happens when the outside processor runs late — do the planned dates for downstream operations recalculate automatically? If somebody has to retype every subsequent operation by hand, it will not be maintained once the system is live.
Job-level costing has to include comparison against the quote
In metal fabrication, the gap between estimated cost at quotation and actual cost is where the margin is decided. Misjudged material yield, more changeovers than planned, a revised outsourcing rate, rework caused by defects — all of these happen after the quote has gone out.
Job-level costing therefore has to do more than total up actual cost. It has to place actual cost alongside estimated cost so the variance is visible. Analyze the jobs with the largest variances after the fact, then feed what you learn back into the next quote. Cost control only starts affecting profit once that loop is turning. A system that merely accumulates actuals can produce a monthly financial pack, but it will not improve quoting accuracy.
Setup analysis and die shot counts are decided by data entry design
As features, setup time analysis and die shot count tracking are simple. The hard part is how the underlying data gets captured. Asking operators to clock the start and the end of every setup separately is accurate but adds effort. Ideally shot counts come straight from the counter on the press, but on older machines there may be no counter at all.
The practical compromise for setup is to treat the transition between operations as the boundary for actual data capture, counting the time between completion of the previous operation and the start of the next as setup. It is not a strict setup measurement, but it is more than sufficient for seeing trends by operation and by part number. For die shot counts, start by accumulating totals from recorded production quantities, then add counter integration later for the specific dies where precision matters. That staged approach is far easier to get moving.
Either way, during selection you should be asking not only whether a feature works, but what the shop floor has to enter in order to make that feature work. Skip the estimate of data entry burden and you end up with a system where the functionality exists but the data never accumulates.
Be honest about industry fit
There is no shortage of production management packages, but most of them were designed around assembly manufacturing — buy components, assemble them according to a bill of materials, produce a finished product. Metal fabrication follows a different logic. Material is cut, blanked on a press, bent, sent out for treatment, and inspected when it returns. It is a process-flow model. The BOM hierarchy is shallow, while the operation sequence and the movement in and out of outside processors are complex.
That difference rarely shows up in a side-by-side feature comparison. In practice, the number of packages that genuinely fit a cut-stamp-bend-outsource-inspect flow is far smaller than the number aimed at assembly operations. Rather than filtering on the label “for manufacturing,” the fastest route is to ask each vendor how many metal fabricators they have implemented, and whether any of those have a process structure resembling yours.
Comparing packaged software, custom development, and staying on spreadsheets
Once the requirements are settled, the next question is how to deliver them. There are broadly three options — implement an off-the-shelf package, build custom software, or continue with spreadsheets while improving them.
| Option | When it fits | Main risk |
|---|---|---|
| Packaged software | Process structure close to industry norms, short time to launch required | Your outsourcing practices or unit conversions do not fit standard functionality |
| Custom development | Many unique operations, complex integration with existing systems | Cost and schedule expand, and maintenance depends on specific individuals |
| Continue with spreadsheets | Few part numbers and orders, with a stable, dedicated administrator | Growing key-person dependency, with outsourcing and cost permanently invisible |
In practice, rather than treating this as a three-way either-or, most companies land on a middle path — take a package as the base and cover only the metal-fabrication-specific gaps with add-ons. The critical discipline there is to size the add-on scope during selection, not after. If add-ons cover half the requirements, you are effectively building custom software, and both the budget and the timeline need to be re-estimated on that basis.
For a broader framework on how to compare options, our comparison of production management systems sets out the functional, cost and implementation-timeline axes. Establishing those general selection criteria first, then layering the five metal fabrication requirements on top, produces a much more stable decision.
Choosing to stay on spreadsheets does not mean doing nothing. Centralizing at minimum the ship-out and expected return dates for outsourced operations, or adding setup time as an explicit line on the daily work report, are partial improvements that pay off on their own and also serve as preparation for a later system rollout. Analyses of digital adoption among small and mid-sized manufacturers report cases where simply moving away from fragmented paper-and-spreadsheet control produced substantial labor savings in routine work such as incoming inspection. Start where the return is easiest to demonstrate rather than attempting whole-plant optimization at once — that principle applies just as well to metal fabrication.
High-mix, low-volume and make-to-order issues specific to metal fabrication
Most metal fabricators run repeat volume parts and one-off custom orders through the same plant, and that mixture is precisely what makes production control difficult. Volume parts repeat the same routing, so standard times accumulate naturally. One-off parts follow a different routing every time, and there is no historical data from which a standard time could be derived.
In shops where make-to-order work dominates, three problems stand out. First, there is no consistent set of unit rates on which to build a quote. Second, capacity planning is impossible when nobody can predict how long a one-off part will take. Third, material sometimes has to be ordered before the drawing is final, so a later specification change turns the purchase into waste.
The workable response is to stop treating every one-off as entirely unique. Group them by similarity of work content and hold a standard time at the group level. Classify by material thickness, material grade and cut length, and hold an approximate setup time and run time for each class. Precision drops, but you gain numbers you can use for both quoting and load planning. On the system side, verify whether it can hold these “similar part groups” as master data.
For the planning and costing concepts specific to make-to-order work, we cover the subject in detail in our article on production management for make-to-order manufacturing. For metal fabricators, add the movement in and out of outside processors on top of that, and assume the routing is one level more complex again.
What happens on the floor when cost is invisible
When cost stays invisible, the first thing that breaks is pricing. Part numbers that are actually losing money get judged by feel as “doing fine” and continue to be accepted. Metal fabrication carries a high material cost ratio, and steel price movements hit cost directly, so a pricing error works its way into the profit and loss quickly.
The second failure is misdirected improvement. Without cost visibility at job level, all that remains is a general sense of being busy without making money, and there is no way to separate whether the cause is material yield, changeover frequency, outsourcing rates or rework on defects. The result is a shop floor working hard on things that barely move the number.
The third failure is delayed investment decisions. Equipment renewal and automation projects only get a credible payback plan once the current cost structure is expressed in numbers. If you cannot state how many hours a year go into setup, and what those hours are worth in baht or in your reporting currency, you cannot put a quick-changeover investment in front of the board.
We have summarized how cost management systems work and how to evaluate them in our article on manufacturing cost management systems. For metal fabricators, add two checks — whether the system can allocate cost across the boundary between weight-based material units and piece-based product units, and whether outsourcing cost can be attached directly to a job.
What is different for plants operating in Thailand and ASEAN
Everything above applies to metal fabrication as an industry. Plants operating in Thailand and the wider ASEAN region carry an additional layer of considerations.
Deep supply chain tiers mean a lot of outside processors
Thailand’s automotive parts supply chain contains more than 2,200 parts manufacturers and generates more than 420,000 jobs. Broken down by tier, there are 525 Tier 1 suppliers, of which roughly 300 are foreign-affiliated, and 1,760 companies across Tier 2 and Tier 3, of which roughly 500 are foreign-affiliated. Japanese-affiliated suppliers are estimated at roughly 700 to 800 companies.
The composition of the Tier 2 and Tier 3 layer is the part worth noting. In that layer, wholly Thai-owned local companies account for about half of all firms, and the concentration is reported to be particularly heavy in metal fabrication, injection molding and die and mold work. In other words, the partner shops a Japanese-affiliated fabricator uses for surface treatment or heat treatment, and the subcontractors handling portions of internal operations, are largely local companies.
This feeds straight into system requirements. Communication with outside processors is not going to happen in a single language. Purchase orders and delivery confirmations move in Thai and English. The counterparty may have no system at all, tracking progress by phone and chat app. To make outsourced operations visible under those conditions, either the system needs master data covering the contact person, working language and communication channel for each supplier, or — at minimum — you need to lock down an internal routine that reliably records planned and actual ship-out and return dates on your own side first.
The industry as a whole is moving toward smart manufacturing
Interest in smart manufacturing across Thailand’s metalworking sector is clearly rising. Metal Thailand 2026, scheduled for 14 to 16 October 2026 at BITEC in Bangkok, is a good marker of that shift. It is co-organized by Chinese foundry associations together with the Thai Foundry Association, with full backing from the Federation of Thai Industries, and its content reflects cross-industry initiatives aligned with the FTI’s 5I strategy.
Exhibitions like this are a place to look at equipment, but they are equally a place to check where the surrounding software and data integration landscape has moved. If you are in the middle of evaluating a production management system, spending time on the show floor confirming how far machine builders have gone in exposing data externally will make the design of automated operation data capture considerably easier later on.
Build the investment plan around BOI incentives and current investment flows
When investing in equipment or systems in Thailand, the promotion scheme run by Thailand’s Board of Investment (BOI) is a genuine factor in the decision. BOI-promoted companies can receive corporate income tax exemption for up to eight years depending on activity and zone, along with exemptions from import duty on machinery and on raw materials. A production management system on its own will not necessarily qualify, but when it is filed as part of a business plan alongside capital equipment, it affects how the total investment package is structured.
Japanese investment into Thailand has continued at scale. Investment promotion applications from Japanese-affiliated companies reached a cumulative 1,380 projects worth 396 billion baht from 2021 through the first half of 2026. Investment at that level means new line launches and expansions of existing plants are still very much underway. Timing a production management system rollout to coincide with capital investment lightens the load considerably, because the routing and master data build can run alongside the new line commissioning work.
Design on the assumption that people will turn over
Shop floor turnover in Thai plants tends to be faster than in Japan. Any outsourcing control or changeover judgment that depends on one veteran’s memory breaks the moment that person moves on. This happens in plants everywhere, but the shorter cycle in Thailand means the consequences arrive sooner.
The system requirements that follow are concrete. Entry screens must display in Thai. The number of fields must be small enough that entry can be completed on the floor without escalation. Decision criteria must be held as master data rather than in someone’s judgment. Conversely, a system with Japanese-only screens, or one where data entry requires a supervisor’s decision, may run at go-live but has a high probability of being abandoned within six months.
How to run the implementation
A metal fabrication production management rollout is best run in three stages — requirements definition, proof of concept with a small start, and full implementation.
Stage one — start requirements definition from a real routing
The instinct in requirements workshops is to start from a feature checklist. For metal fabrication, it is far more reliable to pick three to five real jobs and write out their routings on paper. Follow each one from cutting onward, marking exactly when material goes out to a partner shop, when it returns for inspection, and where rework occurs.
Doing this always surfaces practices that were never shared internally — one particular customer requires a different inspection record format, while one particular outside processor only accepts deliveries twice a week. Discovered during requirements definition, these details can be assessed against standard functionality. Discovered after go-live, they become change requests and additional development.
At the same time, measure your current numbers. Changeover frequency and time per changeover, average lead time and variability at each outside processor, and the value of your stock count variances. These become the baseline for measuring benefits after implementation, and they double as the material for the business case you present to management.
Stage two — restrict the proof of concept to one line or one product family
Targeting every operation and every part number from the start means several months of master data preparation before anything visible happens, and the shop floor loses interest. Choose one representative product family that includes outsourced operations, build the routing master and supplier master for that scope only, and run real orders through it.
What the proof of concept has to demonstrate is not whether the features work, but whether the shop floor can sustain the data entry. Are setup start and end times actually being clocked? Are outsourced ship-outs and returns registered the same day? Wherever entry stalls, the cause is either screen design or an operating rule, and either way it should be fixed before full rollout.
Stage three — run full implementation through a complete cost close
In full implementation, widen the scope and simultaneously run one or two monthly cost closes end to end. This is where unit conversion misconfigurations and unlinked outsourcing costs typically surface for the first time. Only after a complete close can you say the system works as a business process rather than as software.
After go-live, establish a routine for reviewing setup time totals and actual outsourcing lead times on a regular cadence. Without a forum that looks at the numbers, hard-won data simply accumulates unread. Collecting data is no longer the difficult part. The difference is made by whether the collected data is fed back into planning and cost.
Frequently asked questions
Does a metal fabricator really need a production management system?
If you have few part numbers, almost no outsourced operations and a limited volume of orders, spreadsheet control can still work. The two most useful indicators are the presence of outsourced operations and the number of part numbers. Once you have several operations going outside and part numbers running into the hundreds, tracking location through memory and spreadsheets is close to its limit. If delivery answers now require a phone call, or if stock count variances can no longer be explained, it is time to evaluate.
What does a production management system for metal fabrication cost?
Cost varies so much with deployment model and scope that quoting a single market rate would be misleading. The concept worth holding onto is that comparison must be on total cost — not just the initial license or implementation fee, but master data preparation effort, add-on development, maintenance fees, and shop floor training time. For metal fabricators in particular, the total swings on whether outsourced operation control and unit conversion are standard or add-on. When requesting quotes, ask each vendor to state explicitly, for each of the five industry-specific requirements, whether it is standard functionality or an add-on. That is what makes cross-product comparison meaningful.
Should we choose custom development or a package?
The practical approach is to first establish how much of your requirements a package can satisfy, and decide based on the proportion that is left over. If standard functionality covers 70 to 80 percent and the remainder fits within add-ons, the package is the stronger option. If more than half your requirements cannot be met as standard, custom development is worth evaluating seriously. Be aware, though, that custom development carries key-person risk in maintenance, so who maintains the system after delivery needs to be settled at contract stage.
How long does implementation take?
One to two months for requirements definition, two to three months for the proof of concept, and three to six months for full rollout and adoption is a reasonable guide, but the determining factor is not product functionality — it is the volume of master data. Routing master, part master, supplier master, and yield settings for unit conversion. Starting without these in place can easily double the schedule. The corollary is useful: preparing master data in parallel with system selection can shorten the implementation window dramatically.
Conclusion
Production control in metal fabrication carries four structural problems — invisible outsourced operations, unmeasured setup time, conversion between weight-based and piece-based units, and the chain reaction that follows any schedule change. When selecting a production management system, add five requirements to your general functional comparison: outsourced operation control, job-level costing, setup time analysis, die shot count tracking, and automated unit conversion. Split the rollout into requirements definition, proof of concept and full implementation, and confirm the result by running a complete monthly cost close. For plants in Thailand and ASEAN, assume local outside processors and continuous workforce turnover, and prioritize data entry that can be completed on the floor — that alone removes a large share of the failure modes.
TOMAS TECH supports manufacturers operating in Thailand and ASEAN across the full arc of a production management system project, from selection through to sustained adoption on the shop floor. It is entirely fine if you are still at the exploratory stage with requirements undefined, or weighing whether to migrate off an existing system. We are happy to start from your actual routings and work through the structure together — please get in touch with our team to open the conversation.
References
- Structural challenges in metal fabrication production control and a system selection checklist
- Functional overview of production management systems for metal fabrication
- Ascot production management product information for manufacturing
- Hitachi Systems solutions for the metalworking industry
- Challenges facing small and mid-sized manufacturers in the 2026 SME white paper
- Comparison and selection criteria for cost management systems
- Report analysis on the structure of Thailand’s automotive parts supply chain
- Metal Thailand 2026 event information
- JETRO overview of Thailand’s investment promotion scheme
- Trends in Japanese investment promotion applications in Thailand
