Factory Tablet Systems 2026 | Total Cost Is Set by the Update Deadline, Not the Device Price

What decides the 5-year total cost of a factory tablet system is not the device price. It is the day OS and security updates stop. Calculated over 40 devices and 5 years, the consumer tablet with the cheapest hardware comes to THB 3,044,000, while the semi-rugged model at roughly twice the unit price comes to THB 2,302,800. The ranking flips. What creates the gap of THB 741,200 is the fleet-wide replacement of THB 604,000 that lands at the end of year 3, plus the THB 637,200 difference in 5-year failure cost, minus the THB 500,000 saved on the hardware itself.

What expires first on a shop floor tablet is not the screen – it is the update window

The first question at a shop floor device selection meeting is almost always “will it break if we drop it?”

It will. It will break, but a broken unit can simply be replaced. The fact that some of the 40 devices break every year can be booked as an amount of money, as we do later in this article, and if you hold spares, operations do not stop. Breakage is a cost, but it is not an accident.

The accident is a different kind of expiry. It is the day the supply of OS and security updates ends, commonly called EOL (End of Life). The screen is intact, the battery still holds, the housing looks clean. Even so, a device that stops receiving updates loses three things at once.

1. It falls outside MDM management. Most management products publish the range of OS versions they support. A device outside that range no longer gets guaranteed behaviour for policy distribution or remote wipe.
2. It loses guaranteed operation of the business apps. Vendors of electronic form and work instruction apps keep moving the OS versions they validate against. Older versions are dropped one after another.
3. It stops passing internal security standards. It is not unusual for a head office information security department to rule that “devices whose update supply has ended must not be connected to the business network.” If it is flagged in an audit, the device is removed regardless of what the shop floor would prefer.

Once those three line up, only one option remains: fleet-wide replacement. Instead of swapping devices one at a time, you swap them all together on the deadline date. The reason is simple: the update deadline is determined by the model, not by the individual unit. A factory that bought 40 units of the same model reaches the deadline on 40 units at once.

And this fleet-wide replacement lands in the middle of the evaluation period. A factory that made its investment decision on a 5-year horizon takes a lump-sum outlay at the end of year 3. That single hit turns the device-price ranking upside down.

In this article we set up one model of a Japanese-owned factory in Thailand and follow that structure through to the end in numbers. Here is the conclusion up front.

• The ranking by device price is A (consumer) < B (semi-rugged) < C (fully rugged)
• The ranking by 5-year total is B 2,302,800 < A 3,044,000 < C 3,227,600
• On simple payback, A looks best (2.20 years), yet on the 5-year net, B is 1.85 times A

Note that the calculations below are built as “one baseline and three designs.” All three designs are alternatives measured against the same baseline, so their effects cannot be added together. We come back to this point in the summary.

Assumptions | 60 workers, 40 devices, 5 years, a Japanese-owned factory in Thailand (baseline THB 1,035,200 per year)

First, the assumptions. Every number that appears later comes out of this table.

How we set the effective hourly rate of 70 baht

The statutory minimum wage in Thailand is 337 to 400 baht per day, and Bangkok is at 400 baht (revised 2025-07-01). Bangkok is not the only province at the 400 baht ceiling – several eastern provinces where Japanese factories cluster are there too – but the revision date differs by province, so please confirm the effective date for the province you are in. Dividing that daily 400 baht by 8 hours gives 50 baht per hour. We then multiply by 1.4 for statutory contributions such as social security and for allowances, and set 70 baht per hour as the effective hourly rate for shop floor workers in this article.

This 70 baht is not a fact; it is a value we assume for this article. The minimum wage of 337 to 400 baht per day is published fact, but the effective hourly rate varies by factory depending on the mix of allowances, length of service and how overtime is handled. When you run the estimate for your own company, please rebuild it from your own actual labour cost data. If you calculate this line at the minimum wage of 50 baht as-is, the valuation of time saved on the shop floor comes out roughly 29% smaller (1 – 50 / 70 = 28.6%).

The 160 baht per hour for office and IT staff is an assumed value in the same way. We set it as an effective figure covering production control and quality control staff, the clerical staff who compile daily reports, and the systems people.

How we arrive at 40 devices

The deployed count of 40 devices is the peak concurrent use of 35 plus 5 spares. The spare ratio is 5 / 35 = 14.3%.

The important point here is that the number of spares is decided by the duration of downtime, not by the number of breakages. Even a device population that produces 12 breakages a year needs only 1 spare if repairs come back the same day. Conversely, even if only 4 units break a year, you need several spares if repairs take three weeks. The three designs described later differ not only in failure rate but also in repair lead time, but in this model we keep 5 spares common to all three designs. If we varied the spare count by design as well, the comparison would stop being a comparison, so we deliberately fix it.

The baseline (current state = paper operation)

This is the only baseline in this article. We convert the current state, running on paper forms and paper work instructions, into an annual cost.

Let us verify. 221,760 + 253,440 = 475,200. +180,000 = 655,200. +260,000 = 915,200. +120,000 = 1,035,200.

Some readers will find 12 minutes per person per day hard to accept. That figure is not “time spent writing on paper.” Writing itself does not disappear when you digitise; it is replaced by data entry. What we put into the 12 minutes is only the actions that disappear once you digitise: walking to fetch a form, rewriting a form after a mistake, copying results from the previous process off another sheet of paper, carrying the completed sheet to the office. 60 people x 12 minutes = 720 minutes per day, which comes to 3,168 hours per year.

The 180 minutes per day x 2 people on the office side follows the same logic. It is the time spent retyping the paper that comes up from the floor into Excel, cross-checking it and compiling it. This is the cost item where digitisation has the largest effect, and as described below we assume an 85% reduction here.

Annual savings from digitisation (identical across all three designs)

The three designs assume the same business application. The savings from digitisation are therefore identical across the three designs. The only difference is on the device side.

Let us verify. 133,056 + 215,424 = 348,480. +144,000 = 492,480. +182,000 = 674,480. +108,000 = 782,480.

Over 5 years, 782,480 x 5 = THB 3,912,400.

Please note that we do not set any reduction rate at 100%. Floor writing is 60%, office compilation is 85%, paper is 80%. Switching to form entry on a tablet does not take paper to zero. Stamped forms for customer submission, originals that must be retained by law, and backup operation during power or network outages all remain. We also cap the reduction in transcription errors at 70%. Misreading handwriting disappears, but pressing the wrong option and selecting the wrong target record do not.

The design of the electronic forms themselves, that is, which forms to digitise first and how to reduce the number of input fields, is covered separately in Electronic forms and the paperless factory. This article fixes the software side and narrows the question to how to choose the device.

Three designs | What changes between consumer, semi-rugged and fully rugged

Let us organise the options that line up in a shop floor device selection into three classes.

A, B and C are classes, not specific product names. Nor is this an argument about which vendor is superior. A is the plan of buying retail consumer tablets and handing them out with protective cases, B is a business-oriented model, and C is a rugged device engineered from the outset for drops, dust and water.

The failure rates of 30%, 12% and 4% are model values for this article

These three failure rates are model values for this article. They are not measured values. Published material from rugged device vendors repeatedly indicates a tendency for consumer devices deployed in industrial use to show high annual failure rates, and we set these figures with that tendency in mind. When you run the estimate for your own company, please rebuild it from the actual record of the devices you use today.

Converting the annual failure rates into counts:

Design A produces 1 failure every 22 working days, which is close to one device a month. That is 60 over 5 years. Against a deployment of 40 devices, 60 failures means that over 5 years every device breaks 1.5 times on average. Design B, at 24 failures, works out to 0.6 times per device, and Design C, at 8 failures, to 0.2 times.

That “one a month” matters operationally even more than it matters financially. Pulling a spare out and putting one back, redoing the kitting, updating the asset register, checking the data on the broken device. It recurs as work for the responsible person every single month.

The update commitments of 3, 5 and 5 to 7 years are vendor-published policies

The update commitment years are likewise settings for this article, informed by the policies that vendors in each class publish. They are not periods that we guarantee.

For reference, the following kinds of policies are published in the industry. Some rugged device vendors run their own security patch programmes for Android and offer extensions beyond the standard support period. Major smartphone vendors also publish OS and security update policies such as 7 years for flagship models and 4 years for mid-range series. There are also frameworks such as Android Enterprise Recommended that list devices meeting the requirements needed for business use, including update provision.

Here is the single most important practical check. Ask not “how many years of support” but “until when,” and get it by model name and date.

The convention is that the stated number of years is counted from the model’s release date, not from your purchase date. If you buy a model today that was released two years ago, a model with 5 years of support has 3 years left as far as you are concerned. If you plan on the assumption that the catalogue’s “5 years” starts on your purchase date, the replacement arrives two years earlier than you expected. This happens especially when you have picked up heavily discounted end-of-line stock.

Fixing what does not change across the three designs

To make the comparison valid, let us first set out the items we hold common across the three designs.

Annual licences come to 40 devices x 900 + 40 devices x 1,100 = 36,000 + 44,000 = 80,000 THB per year. That is 2,000 baht per device per year, or 10,000 baht per device over 5 years. Over 5 years in total, 80,000 x 5 = THB 400,000.

Common initial cost is 450,000 + 120,000 = THB 570,000.

Here is one fact that always catches people out in a selection meeting. The THB 570,000 of Wi-Fi upgrade plus charging racks is larger than the 40 devices of Design A hardware (13,500 x 40 = 540,000). The gap is THB 30,000. It really does happen that a team holds meeting after meeting on model selection while the larger common infrastructure spend is handled in a single “other” line.

Cost per failure

What you do when a device fails differs by class.

We treat Design A as “replace” because repair costs for consumer tablets come close to the device price, so buying a new one is cheaper than sending it in. Design C carries a high device price and therefore high repair costs, so on a per-failure basis it is the most expensive of the three designs.

Design C still ends up with the lowest annual failure cost, because the number of failures is small.

The annual gap between Designs A and B is 176,400 – 48,960 = THB 127,440 per year. Over 5 years, 127,440 x 5 = THB 637,200. Please keep that figure of 637,200 in mind. It matters later.

What sits inside the THB 604,000 of one replacement round

Only Design A carries the fleet-wide replacement at the end of year 3. Let us open it up.

540,000 + 48,000 + 16,000 = 604,000. Per device, that is 604,000 / 40 = THB 15,100.

Do not forget the THB 16,000 for wiping and disposal. The amount is small, but leaving devices that hold business data stacked in a factory storeroom is the kind of practice that always gets flagged in an audit. Keeping a wipe record, removing the items from the asset register, and processing them as industrial waste are all part of replacement.

Line up the 5-year totals and the ranking flips (device A

This is the heart of the article.

Initial investment

Design	Devices	Kitting	Common initial	Total initial investment
A	13,500×40 = 540,000	48,000	570,000	1,158,000
B	26,000×40 = 1,040,000	48,000	570,000	1,658,000
C	52,000×40 = 2,080,000	48,000	570,000	2,698,000

At the initial investment stage, the ranking follows the device price. A 1,158,000 < B 1,658,000 < C 2,698,000. Design A is THB 500,000 cheaper than Design B and THB 1,540,000 cheaper than Design C.

If the approval document stops here, Design A gets chosen.

The 5-year total

Now we add up all five years.

Design	Initial investment	Failures over 5 years	Licences over 5 years	Replacement	5-year total
A	1,158,000	882,000	400,000	604,000	3,044,000
B	1,658,000	244,800	400,000	0	2,302,800
C	2,698,000	129,600	400,000	0	3,227,600

Let us verify.

A: 1,158,000 + 882,000 = 2,040,000. +400,000 = 2,440,000. +604,000 = 3,044,000.

B: 1,658,000 + 244,800 = 1,902,800. +400,000 = 2,302,800.

C: 2,698,000 + 129,600 = 2,827,600. +400,000 = 3,227,600.

The ranking has flipped.

	1st (cheapest)	2nd	3rd
Device price	A 540,000	B 1,040,000	C 2,080,000
5-year total	B 2,302,800	A 3,044,000	C 3,227,600

Converted to a 5-year total per device, that is THB 76,100 for A, THB 57,570 for B and THB 80,690 for C (3,044,000 / 40, 2,302,800 / 40 and 3,227,600 / 40 respectively). A device with a unit price of THB 12,000 becomes THB 76,100 per device over 5 years. That multiple of 6.3 times is the real shape of a shop floor device purchase.

*Figure 1: 5-year totals for the three designs. Against the device price ranking (A

Breaking down the THB 741,200 of A minus B

The gap between Design A and Design B is 3,044,000 – 2,302,800 = THB 741,200. Let us break down where that 741,200 comes from.

Component	Amount (THB)
Replacement (A only, end of year 3)	+604,000
Difference in 5-year failure cost (882,000 – 244,800)	+637,200
Difference in initial device cost (540,000 – 1,040,000)	-500,000
Total A – B	+741,200

604,000 + 637,200 = 1,241,200. -500,000 = 741,200. It matches the difference in the table.

The initial kitting of 48,000, the common initial cost of 570,000 and the 5-year licences of 400,000 are identical across the three designs, so not one baht of them shows up in the difference. Only three items create the gap: devices, failures and replacement.

Put as a single line, it comes out like this. You think you saved THB 500,000 on hardware, and you pay THB 741,200 more over 5 years. You end up paying 1.48 times what you saved (741,200 / 500,000 = 1.4824).

Break it down a different way and you still get 741,200

For safety, let us verify with a different split. This time we divide it into “money spent on the devices themselves” and “failures.”

Item	A	B	Difference (A – B)
Devices (first purchase)	540,000	1,040,000	-500,000
Kitting (first round)	48,000	48,000	0
Replacement (devices 540,000 + kitting 48,000 + disposal 16,000)	604,000	0	+604,000
Device-related subtotal	1,192,000	1,088,000	+104,000
Failures over 5 years	882,000	244,800	+637,200
Licences over 5 years and common initial	970,000	970,000	0
Total	3,044,000	2,302,800	+741,200

104,000 + 637,200 = 741,200. The same figure as before.

The interesting thing about this table is that A is already THB 104,000 more expensive on the device-related subtotal alone. Even though the unit price is less than half, buying twice in 5 years puts it above Design B, which buys once. Compare purely on device hardware and A is 540,000 x 2 rounds = 1,080,000 against B at 1,040,000. A cheap device bought twice costs more than an expensive device bought once.

The THB 924,800 of C minus B

Design C, on the other hand, is the most expensive on the 5-year total. The gap is 3,227,600 – 2,302,800 = THB 924,800.

Component	Amount (THB)
Difference in device cost (2,080,000 – 1,040,000)	+1,040,000
Difference in 5-year failure cost (129,600 – 244,800)	-115,200
Total C – B	+924,800

1,040,000 – 115,200 = 924,800.

Design C has few failures and recovers THB 115,200 over 5 years. But it has paid THB 1,040,000 more on hardware, so it cannot recover the whole gap. Ruggedness is working exactly as it should, but what it earns does not cover the device price difference. That is where Design C sits in this particular estimate.

That said, the assumptions of this model weigh heavily here. This model assumes an ordinary indoor assembly plant without air conditioning, and does not include washdown, chemicals, heavy dust environments, sub-zero cold stores or outdoor yards. In those environments, the very assumption of a 12% failure rate for Design B does not hold. If the assumptions break, so does the conclusion. Sites that genuinely need Design C certainly exist.

Simple payback does not reflect the replacement (what happens at the end of year 3 for A)

Looking only at the numbers so far, the discussion ends with “let us go with Design B.” In practice, though, it sometimes does not, because the metric that goes into the approval document is simple payback.

Calculating simple payback

Annual running cost = licences 80,000 + annual failure cost

Annual net effect = annual savings 782,480 – annual running cost

Simple payback = initial investment / annual net effect

Design	Annual running cost	Annual net effect	Initial investment	Simple payback
A	80,000 + 176,400 = 256,400	782,480 – 256,400 = 526,080	1,158,000	1,158,000 / 526,080 = 2.20 years
B	80,000 + 48,960 = 128,960	782,480 – 128,960 = 653,520	1,658,000	1,658,000 / 653,520 = 2.54 years
C	80,000 + 25,920 = 105,920	782,480 – 105,920 = 676,560	2,698,000	2,698,000 / 676,560 = 3.99 years

Design A looks best. 2.20 years. Design B is 2.54 years and Design C is 3.99 years.

Decide on this table alone and Design A gets approved. And “payback in two years” carries real weight in an approval meeting. Yet we already know that the cheapest option on the 5-year total is Design B, and that Design A is THB 741,200 more expensive than it.

The simple payback metric does not reflect one baht of the end-of-year-3 replacement. The reason lies in the shape of the formula. The numerator is initial investment and the denominator is annual net effect. The THB 604,000 that occurs at the end of year 3 enters neither of them. Since payback is complete at 2.20 years, whatever happens in year 3 is, as far as this metric is concerned, something that never happened.

The picture changes when you look at cumulative cash

Let us redraw this as cumulative cash by year. You pay the initial investment at the start, accumulate the net effect each year, and for Design A only, subtract the replacement of 604,000 at the end of year 3.

Design	Start	End yr 1	End yr 2	End yr 3	End yr 4	End yr 5
A	-1,158,000	-631,920	-105,840	-183,760	342,320	868,400
B	-1,658,000	-1,004,480	-350,960	302,560	956,080	1,609,600
C	-2,698,000	-2,021,440	-1,344,880	-668,320	8,240	684,800

Following the arithmetic:

Design A: -1,158,000 + 526,080 = -631,920. +526,080 = -105,840. Now year 3. From -105,840 + 526,080 = 420,240 you pay the replacement of 604,000, giving 420,240 – 604,000 = -183,760. +526,080 = 342,320. +526,080 = 868,400.

Design B: -1,658,000 + 653,520 = -1,004,480. -350,960. +302,560. 956,080. 1,609,600.

Design C: -2,698,000 + 676,560 = -2,021,440. -1,344,880. -668,320. 8,240. 684,800.

*Figure 2: Cumulative cash over 5 years. Design A turns positive once during year 3, then falls back to -183,760 with the fleet-wide replacement of THB 604,000 at the end of year 3*

Design A turns positive during year 3, then goes back into the red

Let us look at Design A month by month. The balance at the end of year 2 is -105,840. Dividing the annual net effect of 526,080 by 12 months gives roughly 43,840 per month. Since 105,840 / 526,080 = 0.20 years, it crosses into positive at around month 2.4 of year 3.

At that point, a “payback complete” report goes up the chain. It matches the forecast of 2.20 years almost exactly.

Then, at the end of year 3, the replacement of 604,000 arrives. The balance is -183,760. Three years of accumulation drops below where it stood at the end of year 2 (77,920 lower than the -105,840 of the end of year 2). It returns to positive 183,760 / 526,080 = 0.35 years later, that is, at around month 4.2 of year 4.

Lining up the effective payback timing for the three designs:

Design	Calculated simple payback	When cumulative cash finally turns positive for good
A	2.20 years	Year 4 (turns positive once in year 3, then reverses on the replacement)
B	2.54 years	Year 3 (around month 6.4)
C	3.99 years	Year 4 (around month 11.9)

A was the fastest on simple payback, yet B is the first to settle into positive cash. The ranking flips here as well.

On the 5-year net, B is 1.85 times A

Finally, the 5-year net. From the 5-year savings of THB 3,912,400 we subtract each design’s 5-year total.

Design	5-year savings	5-year total	5-year net
A	3,912,400	3,044,000	868,400
B	3,912,400	2,302,800	1,609,600
C	3,912,400	3,227,600	684,800

These three figures match the end-of-year-5 cumulative cash exactly (A 868,400, B 1,609,600, C 684,800). Naturally so, since it is the same calculation performed in a different order. Still, confirming that they agree is worth doing. When the two do not agree in a real quotation, some cost item is either double counted or missing.

Now the ratio of B to A.

1,609,600 / 868,400 = 1.85 times

On simple payback, 2.20 years against 2.54 years made A look better; on the 5-year net, B’s net benefit is 1.85 times A’s. The difference is 1,609,600 – 868,400 = THB 741,200. That is the same as the difference in 5-year totals (it always will be, because the savings are identical across the three designs).

C against B works the same way. 1,609,600 – 684,800 = THB 924,800, which again matches the difference in 5-year totals.

Which metrics belong in the approval document

From all of the above, the metrics to use in a shop floor device selection come in this order.

1. Cumulative cash by year. It is the only format in which you can see where replacements and major overhauls land within the period
2. The 5-year net over the evaluation period. Compare on the total across the whole period
3. Simple payback as reference only. Attach it with an explicit note that it cannot reflect events occurring outside the numerator and denominator

There is no need to ban simple payback. But on investments where a lump-sum outlay falls in the middle of the period, this metric will always be wrong. Device replacement, large licence renewals, equipment overhauls. All of them produce the same shape of error.

The device price is 15.8% of the total | Where is the other 84.2%

Here is another angle: the share of the 5-year total taken by the device price.

Design	Devices only	5-year total	Share
A	480,000 (12,000×40, excluding cases)	3,044,000	15.8%
B	1,040,000	2,302,800	45.2%
C	2,080,000	3,227,600	64.4%

Design A’s hardware is THB 480,000 and its 5-year total is THB 3,044,000. The share is 15.8%. Even if you include the protective cases (1,500 x 40 = 60,000) and look at 540,000, it only reaches 17.7%.

In other words, when you compare device prices alone across competing quotations, in Design A you are looking at only 15.8% of the total. The remaining 84.2%, THB 2,564,000 in money, sits outside the comparison table.

By contrast, in Design C the devices account for 64.4%. You could say that a rugged device quotation has a high “visible share.” The amount is larger, so the pushback is larger, but at least the invisible part is small. A Design A quotation looks cheap because 84.2% of it is invisible.

Opening up the 84.2%

Let us lay out Design A’s 5-year total of THB 3,044,000 by cost item.

Cost item	Amount (THB)	Share
Devices (12,000 x 40)	480,000	15.8%
Protective cases (1,500 x 40)	60,000	2.0%
First kitting (1,200 x 40)	48,000	1.6%
In-plant Wi-Fi upgrade	450,000	14.8%
Charging and storage racks	120,000	3.9%
Failures over 5 years (60 x 14,700)	882,000	29.0%
Licences over 5 years (80,000 x 5)	400,000	13.1%
One replacement round (end of year 3)	604,000	19.8%
Total	3,044,000	100.0%

Adding the amounts: 480,000 + 60,000 = 540,000. +48,000 = 588,000. +450,000 = 1,038,000. +120,000 = 1,158,000 (matching the initial investment). +882,000 = 2,040,000. +400,000 = 2,440,000. +604,000 = 3,044,000. It matches.

The largest cost item is not the devices but failures at THB 882,000 (29.0%), followed by the replacement at THB 604,000 (19.8%). Those two make up 48.8%. Together they come to THB 1,486,000, which is 3.1 times the device price of THB 480,000.

Here is the same table for Designs B and C.

Cost item	B amount	B share	C amount	C share
Devices	1,040,000	45.2%	2,080,000	64.4%
First kitting	48,000	2.1%	48,000	1.5%
In-plant Wi-Fi upgrade	450,000	19.5%	450,000	13.9%
Charging and storage racks	120,000	5.2%	120,000	3.7%
Failures over 5 years	244,800	10.6%	129,600	4.0%
Licences over 5 years	400,000	17.4%	400,000	12.4%
Total	2,302,800	100.0%	3,227,600	100.0%

(Shares are rounded, so the columns may not sum to exactly 100.0%.)

*Figure 3: Composition of the 5-year totals for the three designs. In Design A the devices are only 15.8%, while failures at 29.0% and replacement at 19.8% make up about half the total*

The THB 1,018,000 that is common to all three designs

There is one more structure that gets overlooked: the part of the 5-year total that is the same amount whichever design you choose.

Common cost item	Amount (THB)
In-plant Wi-Fi upgrade	450,000
Charging and storage racks	120,000
First kitting	48,000
Licences over 5 years	400,000
Common subtotal	1,018,000

450,000 + 120,000 + 48,000 + 400,000 = 1,018,000.

Subtract that, and the design-dependent part looks like this.

Design	5-year total	Common part	Design-dependent part
A	3,044,000	1,018,000	2,026,000 (devices 540,000 + failures 882,000 + replacement 604,000)
B	2,302,800	1,018,000	1,284,800 (devices 1,040,000 + failures 244,800)
C	3,227,600	1,018,000	2,209,600 (devices 2,080,000 + failures 129,600)

Let us verify. A: 540,000 + 882,000 + 604,000 = 2,026,000. +1,018,000 = 3,044,000 ✓

B: 1,040,000 + 244,800 = 1,284,800. +1,018,000 = 2,302,800 ✓

C: 2,080,000 + 129,600 = 2,209,600. +1,018,000 = 3,227,600 ✓

Seen this way, it becomes clear what the selection meeting should actually debate. The common THB 1,018,000 is money you pay whichever device you choose. Arguments about cutting it (spend less on Wi-Fi, drop the MDM) are a different discussion from device selection, and they consist almost entirely of items that break operations when cut. The only thing device selection can move is the design-dependent part, THB 1,284,800 to THB 2,209,600.

Four things to decide before you decide the device (charging and storage, spares, accounts, connectivity)

There are four things to settle before you get into the model discussion. Every one of them is an item that adds cost later if you hand out devices without deciding it.

1. Charging and storage | Where do 40 devices live

We book THB 120,000 of charging and storage racks in the initial investment, common to all three designs.

In a 2-shift factory the devices run 24 hours. The day shift uses them, hands over at shift change, and the night shift uses them. A practice of “charging everything after work ends” simply cannot work here.

So there are three things to decide.

• When charging happens. Do you swap the whole fleet at shift change, or top up during breaks? Whether you choose a model with swappable batteries is decided here
• Where devices are stored. Do you distribute racks near the processes, or consolidate them in one place? Consolidation is easier to manage, but it creates walking time for the workers. That walking time creates new “extra motion” of exactly the kind the baseline counted as reducible
• Visibility of the count. Being able to tell on the spot how many of the 40 devices are in the rack, how many are on the floor, and how many are in for repair

On sites that decide “we can buy charging racks later,” devices typically end up left on workbenches or inside toolboxes, and by morning the batteries are flat. An uncharged device is the same as a broken one. Unusable time occurs and the paper practice comes back.

2. Spares | Operating rules for the 5 spare devices

Five of the 40 deployed devices are spares. Spares do not work simply by being kept somewhere.

• Who issues them. Nominate the person who takes the failure report and issues a spare
• In what state they are kept. Store them kitted, charged and with no account assigned. If you start the initial setup after the failure occurs, that day is lost
• How returned devices are handled. A device back from repair does not go straight back to the floor; it goes back into the spare pool

Recall the pace of Design A: 12 failures a year, one every 22 working days. This procedure fires almost every month. In a factory where the procedure is not defined, someone goes hunting around every single time. That time is not included in the estimates in this article. Include it and Design A’s disadvantage widens further.

3. Accounts | Tied to the device, or tied to the person

We book an MDM licence at 900 baht per device per year, but deploying MDM and designing your account model are two different things.

Shop floor device operation broadly splits into two patterns.

Model	Content	Where it fits	Watch out for
Shared device	The device belongs to a process and workers log in each time	Processes where many people work in shifts, or where one device is shared by several people	Login effort recurs every time. Unless you provide simple login by ID card or barcode, people start sharing one account
Personally assigned	The device is issued to an individual	Patrol inspection, maintenance, team leaders	You need as many devices as people. A retrieval procedure at resignation is mandatory

This model has 40 devices for 60 workers, so it is a shared device model with 1.5 people per device (60 / 40 = 1.5).

The thing you must decide in a shared device model is how you record who entered the data. Run it on a shared account without deciding this, and the worker’s name is not left in the electronic form record. Paper forms had a signature, and after digitisation nobody can tell who wrote it: this genuinely happens. From an audit perspective, that is a step backwards from digitisation.

4. Connectivity | Wi-Fi is decided before the devices

The in-plant Wi-Fi upgrade of THB 450,000 is also common to all three designs. And it is an item that should be decided before device selection.

There are three reasons.

1. One single spot without coverage means that spot goes back to paper. And a process that goes back to paper spreads a “paper is fine” practice to the other processes
2. Offline requirements determine the specification of the business app. Where there are areas with unstable coverage, the app needs to buffer entries offline and sync on reconnection. This requirement is hard to retrofit and should be checked at the package selection stage
3. AP installation work takes longer than device delivery. Site survey, equipment procurement, cabling, mounting on ceilings and beams. If you start the Wi-Fi work after the devices arrive, the waiting time becomes deployment delay

Factory Wi-Fi is designed differently from office Wi-Fi. Reflection off metal fixtures, blocking by moving objects such as forklifts, ceiling height, disconnection during roaming. Our design guidance in this area is collected in Factory wireless LAN and industrial networks. Before you decide the device model, get a quotation for a radio site survey.

How to read the standards | MIL-STD-810H and IP65 are saying different things

A rugged device catalogue always carries two kinds of notation: an IP code such as IP65 or IP67, and MIL-STD-810H. These two measure different things. If you conflate them during selection, some form of resistance you need will be missing.

IP codes cover dust and water

An IP code is a protection rating against the ingress of dust and water. It is written as two digits, the first for dust and the second for water.

Notation	First digit (dust)	Second digit (water)
IP54	5 = dust ingress is not fully prevented, but not to a degree that interferes with operation	4 = splashing water from any direction
IP65	6 = dust ingress is fully prevented	5 = water jets from any direction
IP67	6 = same as above	7 = temporary immersion at a defined depth and duration

That difference is why we set Design B at IP54 class and Design C at IP65 in this article. IP65 can take a hose; IP54 cannot. In dusty processes and washdown processes, that one character decides how the operation runs.

One caution: a higher depth and duration for water does not always mean a superset. It is entirely possible for a product to support IPX7 (temporary immersion) but not IPX5 (water jets). If your operation involves spraying water jets, check that a 5 is listed, not just a 7.

MIL-STD-810H is a test framework, not a pass/fail certification

This is the most misunderstood point.

MIL-STD-810H is a document issued by the US Department of Defense that defines methods for environmental testing. It defines a large number of test methods, covering drop, vibration, temperature, humidity, sand and dust, low pressure and more, and for each product it is a question of which methods were run under which conditions.

The notation “compliant with MIL-STD-810H” on its own does not tell you what was tested. That is because the standard does not set a uniform pass line. 810H is a framework that defines how to test; it is not a pass/fail certification saying “pass this test and you may call yourself rugged.”

So there are three things to check.

Check item	What to ask
Methods performed	Which of drop (Method 516), vibration (Method 514) and temperature (Methods 501/502) were performed
Test conditions	For drops, the height, the number of faces and the material of the impact surface. For temperature, whether it is the operating range or the storage range
Who tested	In-house testing or a third-party laboratory, and whether the report can be disclosed

Also, none of the methods in MIL-STD-810H is a waterproof rating. 810H does include water-related tests such as rain, but that is a different thing from an IP water rating. Reading “it is MIL-STD-810H compliant, so it is strong against water” is incorrect. If you need dust and water protection, check the IP code; if you need drop, vibration and temperature resistance, check the MIL-STD-810H methods performed. Check them separately.

On the drop test, note that the height specified as 122 cm in the older MIL-STD-810G has been raised to 152 cm in 810H. The significance of that change is not simply “it got higher.” It means that even with the same wording of “passed the drop test,” the test conditions differ between an 810G-compliant product and an 810H-compliant one. Check the revision letter in the catalogue too.

Write down what your factory actually needs first

Starting from the standards makes the discussion diverge. Come at it from the other end. Write down what the device is actually exposed to on your own shop floor.

Situation on the floor	Corresponding standard perspective
Falls from a workbench (about 90 cm) to the floor	Drop test height and impact surface material
Placed where vibration from forklifts or conveyors is transmitted	Vibration test
Cutting oil or cleaning agents get splashed	Water side of the IP code plus chemical resistance of the housing material
Used in a process handling powders	Dust side of the IP code (5 or 6)
Moves between an outdoor yard and indoors	Operating temperature range, condensation, screen visibility in direct sunlight
Operated while wearing gloves	Glove-capable touch (a specification, not a standard)

We put that last line in because there are always requirements that do not appear in the standards. Glove operation, operation with wet hands, screen brightness outdoors, a strap for one-handed holding. These sit outside the standards notation, and on the floor they are often more pressing than the standards. At the final stage of selection, always let the actual workers handle real units. If the devices do not respond to gloved hands, all 40 of them will stop being used.

Additional issues on a Thai shop floor (heat, dust, languages, turnover and device binding)

The estimates so far hold in Japan as well as in Thailand, but Thai factories have additional issues that come into play.

Heat, and how to read operating temperature

Thai factories commonly have buildings without air conditioning. Buildings that trap heat in the roof space, areas near moulding machines and furnaces, shipping yards close to outdoors. What you need to check here is the operating temperature range, not the storage temperature range.

The typical symptom in a hot environment is not failure but performance throttling and charging cut-off. Many devices reduce processing performance once the internal temperature exceeds a threshold, and stop charging if it rises further. Nothing is broken, so the warranty does not apply, but on the floor it shows up as “it gets slow in the afternoon” and “it is not charged even though we put it back in the rack.”

The countermeasure is not complete at the device selection stage. Place the charging and storage racks away from direct sunlight and heat sources. That decision belongs to the “charging and storage” item in the previous section.

Dust, and ease of cleaning

In dusty processes the first digit of the IP code (dust) matters. In practice, though, you need one more perspective: ease of cleaning.

On a device fitted with a protective case, powder gets into the gap between the case and the body. Unless you establish a practice of removing the case to clean it, the gap cakes solid within six months. Design A (consumer device plus protective case) is cheap on hardware, but this cleaning effort is structurally built in. A rugged device’s “few seams” design pays off less in dust protection performance itself than in making the cleaning routine lighter.

Languages | Look at both the UI and the master data

In a Japanese-owned factory in Thailand, Thai, Japanese and English sit side by side. In some processes there are also Burmese and Khmer speakers.

There are two layers to check for multilingual support.

1. The language of the UI. Buttons, menus, error messages. This is the layer usually covered by a package’s standard features
2. The language of the master data. Item names, process names, defect category names, equipment names. This is the one that matters.

If the defect category master is registered in English only, Thai-speaking workers start pressing “roughly the second button.” The entry completes, but the data loses its meaning. Cases where you switched to tablet form entry and the collected data turns out to be unusable are mostly caused by the master data language, not the UI.

Translating the master data takes operational effort even when the package supports it. With several thousand item records, translation and verification alone become substantial work. Please build this effort into your deployment plan. It is not included in the estimates in this article.

Turnover, and binding devices and accounts

On Thai manufacturing floors, staff turnover can be faster than in Japan. This is where the “account design” from the previous section pays off.

• When and by whom a leaver’s account is disabled. If HR’s exit process and the system-side account suspension are not linked, accounts survive that should not
• The retrieval procedure for issued devices. Devices issued under the personally assigned model are collected at resignation. A device that is not collected exists on the asset register with no physical counterpart
• Preventing shared-account reuse. If logging in is a nuisance, the floor will always start sharing a single account. Providing ID card tap or barcode login so that identifying the individual is the easier option is the only countermeasure that works

The 900 baht per device per year of MDM is the cost of running this. Remote lock, remote wipe, bulk app distribution, location check when a device goes missing. On a site with 40 devices and annual turnover, operating without MDM is not viable. Conversely, continuing to use devices on an OS version the MDM does not support means losing that control. This is exactly why the update deadline described at the start turns into a fleet-wide replacement.

Where handheld terminals fit alongside tablets

A common occurrence in Thai factories is double investment in tablets and handheld terminals.

The broad division of roles looks like this.

Use	Suited device
Form entry, checking work instructions, viewing drawings and procedures, photo records	Tablet (large screen, many input fields)
Receiving and issuing, picking, stocktaking, item verification	Handheld terminal (one-handed, fast scanning, trigger key)

Tablets can read barcodes too. Reading with the camera adds no extra cost. But in work that involves hundreds of scans a day, one-handed holding and the presence of a trigger key change the work time considerably. Conversely, it is unreasonable to make someone fill in a 20-field inspection form on a handheld’s small screen.

Rather than pushing everything onto one or the other, splitting by task is the right answer. The details of that judgement are set out in Handheld terminals and barcode picking. The 40 devices in this article are tablets covering form entry and work instruction checking; handhelds for warehouse work are a separate budget line.

Frequently asked questions (FAQ)

Are consumer tablets good enough for a factory?

Deciding to “hand out cheap consumer devices and see how it goes” really is cheaper in initial investment. In this model, Design A’s initial investment is THB 1,158,000 against Design B’s THB 1,658,000, a gap of THB 500,000. On the 5-year total, however, Design A is THB 3,044,000 and Design B is THB 2,302,800, which reverses the order and makes Design A THB 741,200 more expensive. There are two reasons: the THB 637,200 difference in 5-year failure cost, and the fleet-wide replacement of THB 604,000 caused by OS updates ending after 3 years. If the conditions are clear, such as a trial with only a few units, or a plan that uses the devices up within the update window, consumer devices are a legitimate option. If the plan is to keep handing them out for 5 years, confirm the update deadline date by model name before you decide. Note again that the failure rates of 30%, 12% and 4% are model values for this article, not measured values.

How many shop floor devices do we need?

It is decided by peak concurrent use plus spares. In this model, against 60 workers we set peak concurrent use at 35 devices and add 5 spares for 40 devices. What you should count is not headcount on the payroll but the number of people using devices at the same time. If those 60 people are split across 3 shifts, the peak is much smaller. The number of spares is decided not by “how many break per year” but by “how long they stay broken.” With a contract where repairs take three weeks you need a thicker spare pool; with next-day exchange you can go thin. The spare ratio in this model is 5 / 35 = 14.3%, fixed in common across all three designs.

What does a factory tablet system cost?

In this model (40 devices, 5 years, a Japanese-owned factory in Thailand), initial investment is THB 1,158,000 to 2,698,000, annual running cost is THB 105,920 to 256,400, and the 5-year total is THB 2,302,800 to 3,227,600. The annual running cost breaks down into licences of THB 80,000 (MDM 900 + business app 1,100 = 2,000 baht per device per year x 40 devices) and failure cost of THB 25,920 to 176,400. The three items most often missed in a quotation are the Wi-Fi upgrade at THB 450,000, the charging and storage racks at THB 120,000, and kitting at THB 1,200 per device, and none of the three changes when you change the device model. In Design A, the device price of THB 480,000 is only 15.8% of the 5-year total.

Does factory Wi-Fi need to cover the whole site?

Full coverage is ideal, but in reality you set priorities. The test is “will this area fall back to paper?” If an area has no coverage, paper forms come back in that process alone, and parallel paper and digital management begins. Parallel management takes more effort than the baseline, so the savings can even swing negative. Where you cannot cover everything, fill the gap with offline entry and later sync on the business app side. That is a specification to check at package selection, however; retrofitting it is difficult. In this model we book the Wi-Fi upgrade of THB 450,000 as initial investment common to all three designs. Please start the radio site survey before device selection.

Is it better to develop the business app for smartphones or for tablets?

It depends on the work. Forms with many input fields, viewing drawings and procedures, and entering results while scanning several fields at once suit a tablet. The larger screen reduces scrolling and screen transitions, and reduces input errors. On the other hand, checking things while moving, one-handed work, and receiving notifications suit a smartphone. Business app development for smartphones requires a design that limits the amount of information per screen; simply shrinking a tablet screen does not work. If you plan to deploy both, check at package selection whether the design is responsive, so that development effort does not simply double. The estimates in this article assume 40 tablets.

Should the business app be an off-the-shelf package or developed in-house?

The estimates in this article assume an off-the-shelf package at a business app licence of 1,100 baht per device per year. We assume a package partly because there is no upfront development cost, and partly because the responsibility for keeping up with OS updates sits with the vendor. An in-house developed app needs rework every time the OS major version rises, and that effort occurs in step with the device update deadline. The structure that is the subject of this article, that the update deadline decides the total cost, gets stronger when the app is developed in-house. Even where your forms are unusual and do not fit an off-the-shelf product, first consider whether the work can be brought onto the package’s standard features, and decide the development scope after that.

When we replace devices, do we have to swap the whole fleet at once?

As long as the model and the update deadline are the same, in practice you do end up swapping them all at once. That is why we book THB 604,000 as a lump sum at the end of year 3 for Design A in this model. If you want to split it, one possible design is to deliberately split the model or the deployment timing at the point of purchase. That does mean, however, that kitting procedures, spare pools and app validation targets all multiply by the number of models. If the only purpose of splitting is to smooth out spending, it often does not balance against the increase in management cost – that is what this model suggests. It is more effective in money terms to choose a model with a longer update commitment and push the replacement itself outside the evaluation period. That is the thinking behind Designs B and C having zero replacements within 5 years.

Summary

What decides the 5-year total cost of the shop floor devices you hand out in a factory is not the device price; it is the cut-off date for OS and security updates.

1. The baseline is THB 1,035,200 per year. That is floor writing and transcription 221,760, office entry and compilation 253,440, paper, printing and storage 180,000, rework from transcription errors 260,000, and searching for records 120,000. Annual savings from digitisation are THB 782,480, or THB 3,912,400 over 5 years. Those savings are identical across all three designs.
2. The 5-year total ranking flips relative to the device price ranking. Devices are A 540,000 < B 1,040,000 < C 2,080,000, but the 5-year totals are B 2,302,800 < A 3,044,000 < C 3,227,600.
3. The THB 741,200 of A minus B is the replacement of 604,000 plus the failure difference of 637,200, less the device difference of 500,000. You think you saved THB 500,000 on hardware, and you pay THB 741,200 more over 5 years. That is 1.48 times what you saved.
4. Simple payback does not reflect the replacement. A 2.20 years, B 2.54 years, C 3.99 years makes A look best, but Design A turns positive at around month 2.4 of year 3 and then falls back to -183,760 with the replacement at the end of year 3. On cumulative cash, B settles into positive first (year 3), while A and C do so in year 4.
5. On the 5-year net, B is 1.85 times A (1,609,600 / 868,400). The difference of THB 741,200 matches the difference in 5-year totals. Simply changing the metric from simple payback to cumulative cash changes the conclusion.
6. The device price is only 15.8% of Design A’s 5-year total. Of the remaining 84.2% (THB 2,564,000), the largest piece is 5-year failure cost at 882,000 (29.0%), followed by the replacement at 604,000 (19.8%). When you compare device prices alone across quotations, 15.8% is all you are seeing.
7. Before you decide the device, decide charging and storage, spares, accounts and connectivity. In particular, the Wi-Fi upgrade of THB 450,000 is common to all three designs and should be started before device selection. On standards, note that IP65/IP67 (dust and water) and MIL-STD-810H (drop, vibration, temperature) measure different things. MIL-STD-810H is a test framework, not a pass/fail certification, and “MIL-rated, therefore waterproof” is not a valid reading.

And here is the most important caution. There is one baseline and three designs. A, B and C are three alternatives measured against the same baseline of THB 1,035,200, and their effects cannot be added together. The moment you choose Design B, neither Design A’s THB 868,400 nor Design C’s THB 684,800 exists any more. Design B’s THB 1,609,600 is the only number there. If an approval document builds up a stack in the form of “the effect of A plus the effect of B,” you are counting the same saving twice, and the figures will not match actual results one year after go-live.

Finally, a word about the nature of the numbers in this article. The failure rates of 30%, 12% and 4% are model values for this article, not measured values. The update commitments of 3, 5 and 5 to 7 years are settings informed by policies published by vendors in each class, not guarantees from us. The effective hourly rate of 70 baht for shop floor workers is a value we assume for this article, while the fact that Thailand’s statutory minimum wage is 337 to 400 baht per day (400 baht in Bangkok) is published fact. Designs A, B and C are product classes, not specific products, and this is not a statement about the superiority of any particular vendor. When you run the estimate for your own company, please replace all of these with your own actual figures.

TOMAS TECH is based in Bangkok, Thailand, and provides integration of production management, factory IT/OT and FA systems for Japanese-owned manufacturers. On shop floor device selection, if you can show us how many forms you use today, how many people use them at the same time, and the state of your wireless coverage, we can produce a 5-year total and cumulative cash estimate in the same format as this article, using your own numbers. It is also fine if you are still at the stage before narrowing down models and simply want to check “how many devices we actually need, and what that comes to over 5 years.” Please feel free to get in touch through our contact page.

Sources

• Best Rugged Tablets 2026: Complete Buyer’s Guide (RuggON)
• Are Consumer Devices Failing? Best Rugged Tablet for Manufacturing (OneRugged)
• LifeGuard for Android Updates (Zebra)
• Android Enterprise Recommended (Zebra)
• Android Enterprise in 2025: Security Clashes with Operations (VDC Strategy)
• What is Android Enterprise? Features, pricing and deployment (CLOMO)
• About MIL-STD-810H testing (Japan Nova System)
• IP65 vs MIL-STD-810H: Key Differences for Rugged Tablets (Valano)
• Shock and vibration resistance (Panasonic Connect TOUGHBOOK)
• Minimum Wage in Thailand (Thai Law Online)