Equipment Maintenance Outsourcing 2026 — Where the Line Sits and the 7 Contract Terms

Equipment maintenance outsourcing is usually considered a way to plug a staffing gap, and just as often it is followed by the complaint that it never actually got cheaper. The cause is rarely the quality of the contractor. It is the way the quotations were compared. This article follows the numbers through a model Japanese-owned plant in Thailand with 80 machines, and shows what changes the moment you stop comparing monthly fees and start comparing annual total cost with downtime losses included.

The real reason people say outsourced maintenance never gets cheaper

Collect quotations from several contractors and you will end up with a stack of very similar sheets. Each one says something like a fixed monthly fee, response started within 24 hours. The line items match, the comparison looks easy, and the cheapest sheet is the obvious winner. That is the first fork in the road.

As long as you line up monthly fees side by side, the option that wins is always the one that wraps the entire plant in a single annual flat-rate contract. It even looks like a rational decision. In the calculations later in this article, that option really does come out cheaper than running everything with three in-house technicians, and cheaper than pushing everything out to spot repairs. The problem is that the comparison stops there.

A maintenance quotation contains direct cost and nothing else. Not a single baht of what your plant currently loses to stopped equipment appears anywhere on the page. Whatever is not on the page is not compared, so the contract with the longer downtime gets selected as the cheap contract. Outsourced maintenance did not fail to save money. The definition of cheap simply never included downtime loss.

The second trap is how the SLA is applied. Most contracts wrap the whole plant in one service level. Conditions such as response started within 24 hours, or attended on the next business day, are applied identically to the main machine that halts the entire line and to a backup machine that has an alternate unit standing next to it. But the stoppage impact of those two machines differs by an order of magnitude. A single SLA is too slow for the equipment that hurts when it stops, and too expensive for the equipment that nobody misses. You lose on both ends, and the monthly fee column shows neither loss.

The argument of this article is simple. Use equipment maintenance outsourcing not to make maintenance cheaper, but to buy back the downtime of the equipment that hurts when it stops. To do that, you stop signing one contract. You sort equipment into three tiers by stoppage impact and change the contract type per tier. The outsourcing spend itself goes up, and the annual total cost goes down. The rest of this article breaks down those numbers.

It is worth noting that unplanned stoppages are not an exceptional event anywhere. In a survey Fluke ran through Censuswide with more than 600 decision makers and maintenance staff in the United States, the United Kingdom and Germany, 55 percent of United States manufacturers reported an unplanned stoppage in the previous year, and the losses across United States manufacturing were reported at up to 207 million United States dollars per week. Unplanned stoppages are not something that would be a problem if they happened. They are a fixed annual quantity that belongs in the budget.

Before you outsource maintenance, sort your equipment into three tiers

There is one job to finish before comparing quotations, and it is sorting the equipment list into tiers. This is not work you hand to a contractor. It can only be decided on the plant side. If you request quotations before the tiers exist, contractors have no choice but to propose an average level of protection across the whole plant, and every sheet you get back will be a single-SLA wrapper.

The model used throughout this article is a Japanese-owned plant in Thailand with 80 machines, running two shifts of 16 hours a day, 25 operating days a month, with 180 unplanned stoppages per year. Your machine count will be different. The method transfers unchanged.

How to separate Tier A, Tier B and Tier C

The test is not asset value and not year of purchase. It is one question only. When that machine stops, how many minutes pass before the line stops.

• Tier A. The line halts the instant the machine stops, or within a few minutes. There is no buffer and no alternate unit.
• Tier B. A stoppage is absorbed for a while by work in process sitting between processes. You have a few hours of grace, and beyond that the downstream processes stop.
• Tier C. There is an alternate unit, or the machine sits in a non-bottleneck process. One machine down and the day’s production plan still holds.

In the model plant, the split came out like this.

By machine count, Tier C is the largest group at 40 units and Tier A holds only 12. But the loss per hour of downtime is 25,000 baht in Tier A and 1,000 baht in Tier C. Machine count and priority of protection are not proportional at all. Sign a per-machine monthly contract while confusing those two things and most of the money flows to the 40 machines in Tier C.

Three things trip people up when they do the tiering for real, so decide them up front. First, a tier belongs to the role the machine currently plays, not to the machine itself. Two identical units can be Tier A in the bottleneck process and Tier C on the spare line. Second, review the tiers once a year. Change the product mix and the bottleneck moves. Third, do not let the maintenance department decide tiers alone. The people who can say accurately how many minutes it takes for the line to stop are production control and manufacturing.

One hour of recovery is worth 6.7 times more in Tier A

Once you have the loss figures per tier, you can convert the value of shortening recovery time by one hour into money. Multiply the annual stoppage count by the loss per hour of downtime.

• Tier A: 24 incidents x 25,000 = 600,000 baht per year
• Tier B: 66 incidents x 6,000 = 396,000 baht per year
• Tier C: 90 incidents x 1,000 = 90,000 baht per year
• Plant-wide total, the loss coefficient: 1,086,000 baht per year

The same one hour of faster recovery is worth roughly 6.7 times more in Tier A than in Tier C, which is 600,000 divided by 90,000. That factor of 6.7 is the entire reason a single contract is the wrong shape.

The plant-wide loss coefficient of 1,086,000 baht per year becomes the foundation for every calculation that follows. Every hour of change in average recovery time moves 1,086,000 baht a year. Whether or not you hold this coefficient decides whether or not you can compare quotations at all. To build the same table for your own plant, you need only two inputs per tier: the annual count of unplanned stoppages and the loss per hour of downtime. Start the loss figure from hourly output on that line multiplied by contribution margin, then add the extra labour of late delivery and restart.

Keeping tiers and the loss coefficient current requires stoppage records held per machine. If you run on paper daily reports and personal spreadsheets, every tier review starts from re-aggregating raw data. How to hold those records is covered in choosing an equipment maintenance management system, which is worth reading alongside this article.

There are five contract types, so line them up by recovery time and annual cost

With tiers settled, the next step is an inventory of contract types. Strip away the product names and maintenance response fits into five shapes. What matters is lining them up on two axes that are not the monthly fee: what is the average recovery time, and what is the annual direct cost.

In-house team of three, spot, annual flat rate, priority contract and resident engineer

The build-up is below. When you read a quotation, always confirm which of these components the contractor’s monthly fee corresponds to.

• Three in-house technicians = maintenance technician at 47,000 baht per person-month including social security and overtime x 3 people x 12 months = 1,692,000, plus training, certification, tools and instruments at 180,000 = 1,872,000 baht
• One spot call = travel charge 2,500 baht plus labour 900 baht per hour x 3 hours = 5,200 baht
• Annual flat rate = 95,000 baht per month x 12 = 1,140,000 baht
• Priority contract = 134,000 baht per month x 12 = 1,608,000 baht
• Resident engineer = 150,000 baht per month x 12 = 1,800,000 baht

Three things should be read off this table.

First. Three in-house technicians deliver an 8-hour average recovery time at 1,872,000 baht a year. That is more than the 1,140,000 baht annual flat rate and more than the 1,608,000 baht priority contract. In-house is unconsciously assumed to be cheaper than outsourcing, but once social security, overtime, training, tools and measuring instruments are stacked on top, it is not. A day-shift-only structure also means anything that stops at night waits until morning, and that is exactly what the 8-hour average is telling you.

Second. Spot outsourcing looks cheapest and is slowest. A unit price of 5,200 baht per incident is attractive, but next-business-day attendance produces a 12-hour average recovery time. This shape is a tool for equipment where slow is acceptable, and nothing else.

Third. The gap between the priority contract at 134,000 baht per month and the annual flat rate at 95,000 baht per month is 39,000 baht a month, which is 468,000 baht a year. For that difference, average recovery time drops from 6 hours to 3 hours. Converted at a loss coefficient of 1,086,000 baht per year, those 3 hours are worth 3,258,000 baht. Once you understand it as paying 468,000 baht to buy back 3,258,000 baht, you stop spending meetings agonising over a 39,000 baht monthly difference.

Separately, reducing the number of unplanned stoppages themselves through condition monitoring and vibration or current trend analysis is a different lever worth pursuing in parallel with these five shapes. The reasoning is set out in this guide to introducing a predictive maintenance system. Fewer incidents means a lower loss coefficient even if the contract type never changes.

One resident engineer can only absorb 90 incidents a year

At this point the natural thought is to cover every machine with a resident engineer and get recovery down to one hour everywhere. Physically, that does not work.

Each unplanned incident takes 6 hours once travel and record keeping are added to the hands-on work. A resident engineer works 1,900 hours a year, but not all of it can go to unplanned response. Preventive maintenance and improvement activity need their share, which leaves roughly 540 hours, or about 28 percent of working hours, for unplanned response.

540 divided by 6 = a ceiling of 90 incidents a year.

The model plant has 180 unplanned stoppages a year. One resident engineer covers exactly half. The next idea is usually to put two residents on site. Kill that idea with numbers rather than opinion. A second resident brings the remaining 90 Tier C incidents to one-hour recovery as well, cutting Tier C from 12 hours to 1 hour, a saving of 11 hours. One hour of Tier C recovery is worth 90,000 baht a year, so the benefit is 11 hours x 90,000 = 990,000 baht. On the cost side you pay 1,800,000 baht a year for the second resident and stop paying the 468,000 baht of Tier C spot fees, a net increase of 1,332,000 baht a year. A benefit of 990,000 baht against a cost of 1,332,000 baht leaves you 342,000 baht short. Against the tiered option in the next section, which lands at an annual total cost of 4,344,000 baht, two residents come to 3,600,000 plus 1,086,000 of downtime loss, or 4,686,000 baht, and that same 342,000 baht gap shows up. It is close, but it does not overtake.

So the fastest shape available, the resident engineer, has to be handled as an allocation problem: which machines get the 90 incidents you can absorb. That allocation is the practical purpose of tiering. Tier A at 24 incidents plus Tier B at 66 incidents is exactly 90, matching one resident engineer’s annual capacity. The remaining 90 Tier C incidents can tolerate slow, so they go out on spot. That combination is the one that appears as Pattern 6 in the next section.

Annual total cost by pattern for the 80-machine Thailand model

Combining everything so far, here are the options ranked by annual total cost, which is direct cost plus downtime loss. Downtime loss is the loss coefficient of 1,086,000 baht multiplied by average recovery time. Only the tiered pattern is calculated tier by tier.

The basis of the calculations is as follows.

• Pattern 2 direct cost = 180 incidents x 5,200 = 936,000
• Pattern 6 direct cost = resident 1,800,000 plus Tier C 90 incidents x 5,200, which is 468,000, giving 2,268,000
• Pattern 6 downtime loss = Tier A 24 incidents x 1 hour x 25,000, which is 600,000, plus Tier B 66 incidents x 1 hour x 6,000, which is 396,000, plus Tier C 90 incidents x 12 hours x 1,000, which is 1,080,000, giving 2,076,000

Pattern 5 is missing on purpose. Covering every machine with residents for one-hour recovery everywhere exceeds the 90-incident ceiling explained in the previous section, so it cannot be executed. On paper it looks the most attractive of all, and putting an unexecutable option into a comparison table distorts the decision, so it stays out. If a proposal in front of you says one resident engineer covering all equipment, bring out the annual incident count and the hours per incident and ask how many incidents that structure can actually absorb.

The first line that catches the eye is Pattern 2. The lowest direct cost in the table at 936,000 baht produces the highest annual total cost at 13,968,000 baht. That single row is a complete description of what happens when selection is driven by the monthly fee. Pattern 1, all in-house, lands at 10,560,000 baht, which shows that zero outsourcing is neither the safe option nor the cheap one.

Now for the main event. Compare Pattern 3, the whole plant wrapped in a single annual flat rate, against Pattern 6, split by tier.

• Annual total cost: 7,656,000 down to 4,344,000, a difference of 3,312,000 baht a year

Most material stops right here and announces that tiering saves 3,312,000 baht a year. Delivering the conclusion that way is dangerous, because in Pattern 6 the outsourcing spend itself has gone up. Broken out, it looks like this.

• Direct cost, the money paid out every month as contractor fees and labour: 1,140,000 up to 2,268,000, an increase of 1,128,000 baht
• Downtime loss, the money that appears in no ledger anywhere but is definitely being lost: 6,516,000 down to 2,076,000, a reduction of 4,440,000 baht
• Net: 4,440,000 minus 1,128,000 = an improvement of 3,312,000 baht

Subtract the 1,128,000 baht of added direct cost from the 4,440,000 baht of removed downtime loss and you land exactly on the 3,312,000 baht total cost difference. Pattern 6 is a transaction in which paying an extra 1,128,000 baht a year buys back 4,440,000 baht a year of stopped time. Call it buying downtime at a return of roughly 3.9 times, if that framing helps.

How you explain this internally matters enormously. Take it to a budget meeting and the only line anyone sees is that outsourced maintenance nearly doubles, from 1,140,000 baht a year to 2,268,000 baht a year. Downtime loss is journalled nowhere, so it never reaches the table. Get the explanation wrong here and a well-designed tiering plan is rejected as a cost increase proposal.

Three practical suggestions for the delivery.

First, always show the increase and the decrease together. Put the three lines on one page, contractor cost plus 1,128,000, downtime loss minus 4,440,000, net minus 3,312,000, and lead with the increase rather than hiding it. Hide it and the credibility of the whole proposal collapses the moment somebody notices.

Second, ground the downtime loss in the Tier A hourly figure. A total of 4,440,000 baht is too large to feel real, but anyone can verify the claim that Tier A loses 25,000 baht for every hour it stands still. The number to compare is not the monthly fee difference. It is the 25,000 baht you are already paying for every hour of Tier A downtime.

Third, put Pattern 4 on the same page. Pattern 4, the whole plant on a priority contract, comes to 4,866,000 baht, surprisingly close to Pattern 6 at 4,344,000 baht. The gap is only 522,000 baht. If internal alignment or the administrative load makes tiered operation unrealistic in year one, choosing Pattern 4 as an interim step and migrating to Pattern 6 the following year is a perfectly workable sequence. What has to be avoided is the decision to leave Pattern 3 in place. The gap between Pattern 3 and Pattern 4 is 2,790,000 baht.

Tying calculations like these to the order of investment, meaning which process and which machines to work on first, follows the same prioritisation logic covered in how to use factory automation consulting.

What to outsource and what to keep in-house

Even with the contract type decided, leaving the scope of delegated work vague ends up, a few years later, with a plant where only the contractor understands maintenance. Most of what gets described as outsourcing failure is actually a failure of scope. The axis for the decision is not whether the work is difficult. It is whether the information can be handed to the next contractor if this one leaves.

Work that is safe to outsource

The following work can go outside without concern as long as procedures and record formats exist. In most cases quality per hour is more stable outside than inside.

• Routine preventive maintenance. Daily, weekly and monthly inspection, lubrication, filter changes, anything that a checklist completes
• Parts replacement. Bearings, seals, belts, sensors, anything with an established replacement procedure
• Instrument calibration. Periodic calibration of torque wrenches, pressure gauges and thermometers, plus the evidence trail
• Hydraulic and pneumatic inspection and servicing. Leak checks, drain handling, regulator adjustment
• Control panel cleaning and inspection. Dust removal, terminal retightening, cooling fan condition checks
• Maintenance record entry. Registering work performed, parts used and hours taken

In labour hours these items are the bulk of maintenance work, which is precisely why outsourcing them pays. Hold on to routine work while pushing judgement work outside and you lose both the money and the skill. Deciding what to do with the in-house hours freed up by moving routine work out can be planned with the same framework as the labour-hour reduction logic in labour-saving automation case studies and return on investment.

The equipment memory that stays in-house

Four categories of information have to stay inside the company no matter how short-staffed you are. These are not tasks but memories, and once they go outside they disappear the day the contract ends.

1. Modification history. Changes made on the shop floor after installation appear in no manufacturer drawing and no operating manual. Lose the record that one sensor was added or that a stopper was shifted by five millimetres, and at the next breakdown nobody can describe the machine as it currently stands.

2. Version control of PLC programs. This is the most serious one. A PLC where nobody knows who changed what, when, and why is a black box in practice. If a contractor’s engineer tunes something on site and it never comes back into the company, the program becomes untouchable within a few years. The master program and the change history must be held by the plant.

3. Criteria for judging failure modes. The pattern knowledge that this noise means a bearing, that this vibration is the coupling and not the bearing housing, is rarely written down. Require contractor work reports in a format with four fields, symptom, suspected cause, confirmed cause and action taken, and the knowledge accumulates internally over a few years.

4. Parts interchangeability information. Genuine parts can take time to source in Thailand, and shop floors have a long history of getting by with substitutes. A list stating that this part number can be replaced by this locally available item is an asset that directly shortens procurement lead time.

Whether you have a mechanism to keep these four inside decides whether you can switch contractors five years from now. Being unable to switch is the same condition as having no negotiating power on price.

Seven items the contract must settle

At the quotation stage, put the following seven questions to every contractor in identical wording. The quality of the answers predicts the quality of execution.

1. The definition of response. Make explicit what within 24 hours refers to. Acknowledgement of the call, arrival and start of work on site, or completed recovery. Sign with this unclear and the contractor will measure from acknowledgement while the plant measures from recovery, and the mismatch is guaranteed. Definitions may differ by tier. Managing Tier A on start of work and Tier C on acknowledgement is a realistic design.

2. How the four timestamps are recorded. Decide who records occurrence, detection, start of work and recovery, and in what medium. Without those four timestamps there is no performance evaluation and no leverage at renewal. The gap between occurrence and detection is a particularly important indicator of weak monitoring. As noted in choosing an equipment maintenance management system, having the destination for those records ready before the contract starts is the preferred sequence.

3. The equipment list and tier designation. Attach the equipment list as an annex to the contract and state each machine’s tier. A contract with no tiers written in it is, in the end, the same single-SLA wrapper you were trying to escape.

4. Excluded work and the rates for extra charges. Fix in advance what the flat fee covers and what is billed separately. Night and holiday call-outs, large component replacement, overhauls and equipment from other makers are the usual sources of extra charges. Put the rate card in the contract and you never negotiate case by case.

5. Parts procurement responsibility and stock ownership. Decide whether the contractor or the plant holds the parts, and who owns the stock. Depend on contractor-held stock and you may lose the spares along with the contract. Keeping only the critical Tier A parts in plant ownership is a realistic compromise.

6. Ownership of maintenance records. State that every record, including work reports, inspection sheets, vibration data and photographs, is an asset of the plant. Never accept a situation where the data exists only inside the contractor’s system. Specifying export format and frequency is better still.

7. Handover scope at contract end. Decide at the start what will be handed over, in what format, and within how many days of termination. The mere presence of that clause changes the quality of records during the contract. A contract with no handover promise is effectively a contract that forces renewal.

Of the seven, items 1, 2 and 3 should be issued as conditions at the request-for-quotation stage. Specify those three and every proposal arrives on the same footing, which is what finally makes the monthly fee comparison meaningful.

Issues specific to outsourcing maintenance in Thailand

Designing a maintenance structure in Thailand comes with four considerations that differ from the assumptions used in Japan.

Labour cost keeps rising. The minimum wage has been raised to 400 baht per day, applied in Bangkok from 1 July 2025 and in four provinces and one district including Chonburi and Rayong from January 2025. Maintenance technicians already sit above the minimum wage band, but lifting the floor pushes the whole scale up. The assumption that three in-house technicians can be retained at 47,000 baht per person-month needs revisiting every year. Put the other way, the in-house annual figure of 1,872,000 baht is more likely to drift upward, which moves the comparison in favour of outsourcing as the years pass.

Hiring technicians is structurally hard. In the JETRO survey of Japanese-affiliated companies operating overseas for fiscal 2023, 40.4 percent of Japanese-affiliated companies in Thailand cited a shortage of human resources, with difficulty securing managers at 79.8 percent and difficulty securing IT and engineering staff at 56.7 percent. Securing people who can handle automation technology is especially hard. And this is not a Thailand-specific condition. Research by The Manufacturing Institute and Deloitte points to a need for as many as 3.8 million new employees by 2033, of which as many as 1.9 million may go unfilled, with 65 percent of companies naming attracting and retaining talent as their top business challenge. A plan that assumes hiring will solve the problem is not a safe plan.

There is spare capacity in the equipment base. According to the Office of Industrial Economics in Thailand, the manufacturing production index for June 2026 fell 3.10 percent year on year, the second quarter fell 1.79 percent, and average capacity utilisation was 57.47 percent. Spare capacity means that when a Tier C machine stops, decisions such as diverting to an alternate unit or catching up the next day are easy to justify. In a plant running flat out even Tier C would need immediate attention, but at Thailand’s current average utilisation, placing Tier C on a spot contract with next-business-day attendance is a comfortable choice. Conversely, if your own utilisation is above 90 percent, the definition of Tier C itself needs revisiting.

The distinction between a service contract and labour dispatch. If you are considering a resident engineer, the legal review is not optional. If the plant gives the resident direct day-to-day work instructions, the arrangement may be treated as labour dispatch in substance even though it is a service contract in form. Confirm operating rules internally before signing, such as routing all work requests through the contractor’s supervisor and having the contractor prepare the work plan, and communicate them to the shop floor. This is judged on daily operating reality, not only on contract wording. Share it early with your legal and human resources functions on the assumption that a specialist will be consulted.

A 90-day transition plan

Tiered contract design collapses every time somebody tries to switch everything at once, because it means signing expensive contracts with no performance data. Split 90 days into three blocks.

Days 0 to 30. Fix the equipment list and the tiers, and start recording the four timestamps

Build the equipment list and assign a tier to every machine. Judge on the single question of how many minutes pass before the line stops, and agree it across production control, manufacturing and maintenance. At the same time, start recording occurrence, detection, start of work and recovery. Change no contract during this period. Concentrate purely on collecting data. Even 30 days of records changes the quality of the conversation with contractors, because you now hold initial values for stoppage counts and actual recovery times per tier.

Days 31 to 60. Send Tier C out on spot work and gather real data on rates and response

Start outsourcing with Tier C, where failure is cheap. Tier C costs 1,000 baht per hour of downtime, so even a disappointing contractor produces limited losses. What you are checking here is not only price. How work reports are written, in what format records are submitted, the lead time for parts procurement, the skill level of the technicians, and whether the emergency contact route actually connects. None of that is visible in a proposal document. Ideally, place small volumes with several contractors and compare.

Days 61 to 90. Decide the contract type for Tier A and Tier B

Only after Tier C has produced real data do you move on to the Tier A and Tier B contracts. By now you should hold measured stoppage counts per tier, actual response times per contractor, and the pattern of extra charges. In that state, the choice between a resident engineer and a priority contract can be made on numbers.

There is one principle to protect. Do not sign a resident or priority contract without performance data. A resident engineer is 1,800,000 baht a year and a priority contract is 1,608,000 baht a year. A contract signed on the spot to skip a three-month run-up cannot easily be changed later, because the contract term locks you in. Ninety days of delay is a far smaller cost than a full year of losses from a wrong decision.

Frequently asked questions

How much does equipment maintenance outsourcing cost?

It is decided by speed of response. In the model plant used here, with 80 machines and 180 unplanned stoppages a year, spot outsourcing runs at 5,200 baht per incident, an annual flat rate with response started within 24 hours is 1,140,000 baht a year, a priority contract with response started within 4 hours is 1,608,000 baht a year, and one resident engineer plus backup is 1,800,000 baht a year. Comparing those direct costs alone will not produce a decision. On annual total cost with downtime loss added, wrapping the whole plant in an annual flat rate comes to 7,656,000 baht and splitting by tier comes to 4,344,000 baht, so the option with the higher direct cost is the cheaper one overall.

Will outsourcing maintenance leave no engineering capability inside the company?

Sending work outside does not remove capability. What disappears is memory. Modification history, version control of PLC programs, criteria for judging failure modes, and parts interchangeability information. Name those four as plant assets in the contract and specify the record format and submission frequency, and knowledge keeps accumulating internally even with outsourcing in place. Capability is lost not because work was outsourced, but because it was outsourced without deciding who owns the records and in what format.

Which is better value, a spot contract or an annual contract?

The answer changes with the tier. Tier C, meaning non-bottleneck machines with an alternate unit and a downtime cost of 1,000 baht per hour, is fine on spot. Even at 12 hours to recovery the annual loss stays at 1,080,000 baht. Put Tier A on spot, at 25,000 baht per hour of downtime, and the same 12-hour delay produces losses of a completely different order. The right question is not which is better value, but which machines get which. Choose one type uniformly across the plant and you are guaranteed to be losing somewhere.

How does production engineering outsourcing differ from equipment maintenance outsourcing?

They address different time horizons. Equipment maintenance outsourcing is about keeping machines that are already running from stopping, and getting them back quickly when they do. Production engineering outsourcing covers new line start-up, process design, jig design and equipment modification, which is work on things not yet built. Mix the two into one contract and modification projects end up running on a maintenance SLA, and neither can be measured. Separate the contracts and separate the metrics. Maintenance is measured on recovery time, production engineering on start-up duration and yield.

How is a service contract distinguished from labour dispatch in Thailand?

It turns on whether direction and supervision are exercised in substance. Even if the contract is titled as a service agreement, a plant that gives a resident engineer direct day-to-day work instructions may find the arrangement assessed as labour dispatch. Before signing, organise the operating rules, such as routing work requests through the contractor’s supervisor, having the contractor prepare work plans and procedures, and leaving attendance management with the contractor, and brief the shop floor supervisors as well. Because the assessment rests on operating reality, we recommend confirming with your internal legal and human resources functions and with a specialist before signing.

Summary

Outsourced maintenance underperforms because of how it is compared, not because of which contractor was chosen. The essentials are these.

• Compare monthly fees alone and the winner is always the whole plant wrapped in a single annual flat rate, at an annual total cost of 7,656,000 baht. Downtime loss never reaches the comparison because it is not on the quotation
• Stoppage impact differs by an order of magnitude between machines. One hour of recovery is worth 600,000 baht a year in Tier A against 90,000 baht a year in Tier C, a gap of roughly 6.7 times
• The plant-wide loss coefficient is 1,086,000 baht a year. Every hour of change in average recovery time moves that amount
• One resident engineer tops out at 90 incidents a year. Protecting every machine at maximum speed is physically impossible, so the design question is which tiers receive those 90 incidents
• Put Tier A and Tier B on a resident engineer and Tier C on spot, and annual total cost is 4,344,000 baht. The gap against the single-contract option is 3,312,000 baht
• The outsourcing spend itself rises. Direct cost plus 1,128,000 baht, downtime loss minus 4,440,000 baht, and 4,440,000 minus 1,128,000 = 3,312,000. You are buying back roughly 3.9 times the added spend in stopped time
• Settle seven items in the contract, being the definition of response, recording of the four timestamps, equipment list and tiers, excluded work and rates, parts and stock, ownership of records, and handover at termination
• Run the transition over 90 days. Fix tiers and start recording, trial on Tier C, then contract Tier A and Tier B, and never sign a large contract without performance data

What deserves comparison is not the difference between the monthly fees on the table. It is the 25,000 baht lost for every hour Tier A stands still. Re-sort the quotations while holding that number and the right contract becomes obvious.

If you are still at the stage of working out how to split your own equipment list into three tiers, or how to estimate the hourly cost of downtime, that is a fine place to start a conversation. TOMAS TECH has worked on maintenance and equipment on the floor of Japanese-owned plants in Thailand, so we can help with the tiering approach or simply with organising the points worth checking before a contract is signed. If you already hold quotations from other companies, there is no need to show them to us. Aligning the axis of the decision is a good enough starting point. Please get in touch through our contact page.

References

1. Fluke, Unplanned downtime costs United States manufacturers up to 207M USD weekly, 30 October 2025 https://pressroom.fluke.com/unplanned-downtime-costs-united-states-manufacturers-up-to-207m-weekly-exposing-critical-vulnerabilities-in-industrial-resilience/
2. The Manufacturing Institute and Deloitte, Taking charge, April 2024 https://themanufacturinginstitute.org/manufacturers-need-as-many-as-3-8-million-new-employees-by-2033/
3. JETRO, Minimum wage in Bangkok raised to 400 baht per day, 4 July 2025 https://www.jetro.go.jp/biznews/2025/07/b21007a1ac8f7fca.html
4. JETRO, Labour shortage and responses, with attention on future minimum wage trends in Thailand, 16 May 2024 https://www.jetro.go.jp/biz/areareports/special/2024/0303/f5b4d6344434b2a9.html
5. Xinhua, Thailand’s industrial output slips 3.1 pct in June, 27 July 2026 https://english.news.cn/20260727/f8b7ceb202d04b1f8facbcbf44ed60ce/c.html