Ask three vendors for a warehouse management system and the quotes can differ by more than two to one. The gap is not caused by different views of your scale. It is caused by how many layers of WMS cost each vendor put on the page. This article opens the full five-year breakdown for a Japanese-owned factory warehouse in Chonburi, Thailand. The headline finding first — 76% of the five-year total sits outside the software.
Why WMS quotes diverge, and why size is not the reason
Picture the usual starting point. You decide to look at a warehouse management system, you approach three suppliers, and three very different documents come back. Vendor A sends an annual cloud subscription figure and nothing else. Vendor B sends a licence purchase price plus an annual maintenance percentage. Vendor C interviews your team first, then builds up a development estimate line by line. Laid side by side, the cheapest and the most expensive are more than twice apart.
At this point most project owners reach for one of two explanations. Either the vendors are sizing the warehouse differently, or the expensive one is over-engineering the solution. Neither is usually true. What actually differs is the number of cost layers printed on each page. Vendor A’s page contains software only. Vendor B’s contains software and a thin slice of implementation support. Vendor C’s contains interface development as well. You believe you are comparing three prices for the same thing. You are in fact comparing three different scopes that happen to share a product category.
Published price ranges reinforce the illusion that the software decision is the big one. In the Japanese market, cloud WMS products are commonly quoted at an initial cost of JPY 0 to 1 million with a monthly fee of JPY 30,000 to 300,000, on-premise packages at an initial cost of JPY 5 million to 30 million, and full custom development at JPY 30 million to over 100 million (Inter-Stock). Faced with that spread, it feels obvious that choosing the delivery model is the decision that matters most. But when you lay out five years of actual cash going out the door, the amount that moves with the delivery model is smaller than expected, and the layers that do not move at all are far larger. That inversion is the subject of this article.
Every figure below assumes the following model warehouse. These amounts are a planning model that TOMAS TECH has assembled from real project experience in Thailand. They are not published statistics or survey results. Read them as a structure to substitute your own conditions into, not as a price list.
| Item | Condition |
|---|---|
| Location | Warehouse attached to a Japanese-owned factory in Chonburi, Thailand |
| Storage area | 3,000 square metres |
| Stored SKUs | 2,400 |
| Warehouse headcount | 12 |
| Outbound volume | 320 lines per day, 25 operating days per month, 96,000 lines per year |
| Inventory value | THB 45,000,000 |
| Current method | Excel plus paper picking lists |
This profile is entirely ordinary for a Japanese manufacturer operating in Thailand. There is no dedicated IT person stationed in the warehouse. A Japanese plant manager and a Thai warehouse supervisor run it between them. Shipping cut-off concentrates in the late afternoon, and twice a year the line stops for eight hours so the full physical count can be done. That is the warehouse the numbers describe.
WMS cost is made of five layers
To compare quotes properly, split the cost into five layers first. Once the layers exist, you can see at a glance which layer each vendor left off the page — and the layers left off the page are the ones that come back to you later.
The five layers with initial and annual figures
Here is the full-scope deployment, layer by layer. Layer 1 shows the cloud delivery model. All amounts are in Thai baht.
| Layer | Contents | Initial | Annual |
|---|---|---|---|
| 1. Software | WMS product itself, cloud model | 0 | 360,000 |
| 2. Hardware | Handheld terminals, wireless access points, label printers | 1,150,000 | 115,000 |
| 3. Location setup | Rack labels, physical count, master data initialisation | 620,000 | 0 |
| 4. Interfaces | 4 links to ERP and production control | 1,400,000 | 140,000 |
| 5. Adoption and operation | Multilingual procedures, training, local support | 480,000 | 144,000 |
| Total | 3,650,000 | 759,000 |
Two things in that table deserve attention. The initial cost of layer 1 is zero, and the initial cost of layers 2 through 5 adds up to THB 3,650,000. When a cloud vendor tells you there is no upfront investment, that statement is true — of the software layer alone. To actually move goods through the warehouse, you still have to fund THB 3,650,000 across the other four layers.
Now open each layer. Layer 2, hardware is 14 handheld terminals at THB 42,000 each for THB 588,000, plus 8 wireless access points at THB 28,000 each for THB 224,000, plus 4 label printers at THB 65,000 each for THB 260,000, plus THB 78,000 of associated installation work, giving THB 1,150,000. Annual maintenance runs at 10% of the initial figure, so THB 115,000. The 14 terminals cover 12 operators with 2 spares. Cut the spares and the first day a unit fails, one person goes back to paper. The 8 access points exist because a 3,000 square metre high-bay warehouse creates radio shadows in the aisles between racks. Reusing office-grade access points and leaving those shadows in place produces read failures, and read failures become a reputation — the system does not work — which destroys adoption before the project has a chance.
Layer 3, location setup is 3,200 rack labels at THB 35 each for THB 112,000, plus 480 hours of full physical counting and master data initialisation at THB 850 per hour for THB 408,000, plus THB 100,000 of location design, giving THB 620,000. There is no annual figure. This is a one-time cost. Layer 4, interfaces is 4 links at THB 350,000 each — sales orders, shipment results, inbound plans and stock balances — for THB 1,400,000, with annual maintenance at 10%, or THB 140,000. Layer 5, adoption and operation is THB 280,000 for procedure documents and training in Japanese, Thai and English, plus THB 200,000 for two weeks of on-site attendance at go-live, giving THB 480,000. The annual figure is local support at THB 12,000 per month, or THB 144,000.
Layers 2 to 5 cost the same whichever model you choose
This is the point the whole article turns on. Changing the software delivery model moves layer 1 substantially. It does not move layers 2 through 5 at all. You need 14 handheld terminals whether the software is cloud, package or custom-built. You need 3,200 rack labels physically applied, because an unlabelled rack is not a location. And you need the same 4 interfaces, because the systems you are connecting to have not changed.
| Model | Initial | Annual | Layer 1 over 5 years | 5-year total |
|---|---|---|---|---|
| Cloud | 0 | 360,000 | 1,800,000 | 7,445,000 |
| Package | 800,000 | 144,000 | 1,520,000 | 7,165,000 |
| Custom development | 3,600,000 | 540,000 | 6,300,000 | 11,945,000 |
Layers 2 to 5 come to THB 5,645,000 over five years regardless of model, which is the initial THB 3,650,000 plus THB 399,000 per year for five years. Every five-year total in the table above is that fixed block plus the five-year cost of layer 1.
Look hard at the gap between cloud and package. It is THB 280,000 over five years. Against a cloud total of THB 7,445,000, that is roughly 3%. In most selection committees, the hours of debate go into exactly that 3%. Custom development is a genuinely different case — layer 1 alone reaches THB 6,300,000 and the five-year total reaches THB 11,945,000 — but that decision should be made on whether your operation contains processes that genuinely exist nowhere else, not on cost.
The practical conclusion is uncomfortable. Time spent arguing cloud versus package is, in pure cost terms, poorly invested. The same hours spent on the number of interfaces and the granularity of locations move far more money. If you are still deciding what class of system you need at all, our comparison of factory inventory management systems works through the same choice from the functional side and pairs well with this cost view.
76% of the five-year total sits outside the software
Stated plainly — of the THB 7,445,000 five-year total for a full-scope cloud deployment, the THB 5,645,000 in layers 2 through 5 accounts for 76%. The software layer that everyone calls the price of the WMS is the remaining 24%.
What that ratio means in practice is that choosing the proposal with the cheapest software barely moves your total. Turn it around and the risk becomes obvious. Accept a quote that omits layers 2 through 5 because it looks cheap, and THB 3,650,000 of supplementary quotations will arrive between contract signature and go-live. When the amount approved internally and the amount eventually paid differ by that much, it stops being a budget problem and becomes a credibility problem for whoever sponsored the project.
So when a quote lands on your desk, do not start with the number. Start by asking how many of the five layers the document covers. Any layer that is missing from the page is not missing from the project. It is simply going to be paid for later, usually at a worse moment and without competitive tension.
Interface count is what actually moves the total
Among the four layers outside the software, one swings far harder than the others. Layer 4, interfaces, is where the five-year total is genuinely decided.
Interface count against five-year total
Hold everything else constant — same warehouse, same cloud model, identical layers 2, 3 and 5 — and vary only the number of interfaces.
| Interfaces | Layer 4 over 5 years | 5-year total |
|---|---|---|
| 0, WMS standalone with manual entry | 0 | 5,345,000 |
| 2 | 1,050,000 | 6,395,000 |
| 4, standard | 2,100,000 | 7,445,000 |
| 8 | 4,200,000 | 9,545,000 |
Each interface costs THB 525,000 across five years, which is THB 350,000 initial plus THB 35,000 of annual maintenance for five years. Between zero interfaces and eight, the total moves from THB 5,345,000 to THB 9,545,000 — roughly 1.8 times. The warehouse is identical. The software model is identical. Interface design alone produces that spread, and it is a large part of why the three quotes on your desk disagree.
What THB 525,000 per interface actually buys
The most common pushback we hear is that THB 350,000 for one interface seems steep for what amounts to passing a file between two systems. Here is where the effort actually goes.
| Stage | Main work |
|---|---|
| Field definition | Reconciling item code length, leading zeros and character width, unit conversions, and code value mapping tables across both systems |
| Timing design | Polling interval versus nightly batch, and which date records land on when data crosses the cut-off time |
| Exception design | What happens when the counterpart is down, a field arrives empty, or a code appears that does not exist in the master — block or pass through |
| Resend design | The key that prevents double ingestion, who is authorised to trigger a manual resend, and how stock is rolled back on resend |
| Testing | Not only the happy path, but month-end, count day and post-holiday boundary data |
The stages that consume budget are the first one and the exception and resend work. Item codes are almost never clean across systems. The ERP holds a 15-character alphanumeric code, production control holds 12 characters, and the label on the shelf carries only the last 4 digits because that is what the operators actually read. Connect two systems without designing through that mismatch and, in the first week of live operation, someone reports that WMS stock does not agree with ERP stock. Finding out why takes weeks. You are not paying for the act of connecting two systems. You are paying for the guarantee that stock still reconciles six months after they were connected.
When comparing quotes on this layer, the number itself tells you less than one specific sentence does. Look for a statement of which system is authoritative when the two disagree. If that sentence is absent, the price you are looking at covers connectivity testing and nothing beyond it.
What comes back to the floor when you cut an interface
If eight interfaces cost THB 9,545,000 and zero cost THB 5,345,000, the obvious move is to cut them all. It is not that simple. Every interface you remove hands the data it was carrying back to a human being.
Drop the sales order interface and someone keys shipping instructions into the WMS by hand. At 96,000 outbound lines a year, even at an optimistic 10 seconds per line, that is 267 hours annually — and keying errors turn directly into wrong shipments. Drop the stock balance interface and you now have two versions of the truth, one in the ERP and one in the WMS, which someone has to reconcile monthly. In practice that reconciliation lands immediately before the accounting close, which is the single busiest window in the month.
A workable rule of thumb — count how many times a day each data set changes. Anything that changes at least once a day should be interfaced. Anything that changes a few times a month can be handled manually. Sales orders and shipment results fall clearly on the interface side. Inbound plans and stock balances depend on how you operate. Master data such as items, business partners and units of measure changes rarely, so starting manual and adding the interface once operations have stabilised is a realistic sequence. Interfaces can always be added later. What does not change is the THB 525,000 per interface. Deciding the count up front is usually cheaper than discovering it.
Cut location setup from the quote and it returns after go-live
Layer 3, the THB 620,000 for location setup, is the line most often deleted to make a quote fit a budget. The reasoning sounds sensible every time. We can stick the labels on ourselves. We already do a physical count every year, so we do not need to schedule extra hours. And then it comes back, without exception, after the system is live.
Take the 3,200 rack labels at THB 35 each. The unit price is not high because a label is expensive. It covers a material that survives several years inside a warehouse that regularly exceeds 35 degrees, that stays readable after forklift exhaust has dirtied it, and it covers designing the numbering scheme before anyone starts applying anything. Substitute office stationery labels and within a couple of years you have racks nobody can read. A rack nobody can read gets operated as roughly around here, and location accuracy quietly collapses.
The heavier item is the 480 hours of physical counting and master data initialisation, THB 408,000 at THB 850 per hour. This work makes the opening stock balance in the WMS correct. It is a precondition of the system, not an optional extra. If the opening values are wrong, the WMS will manage wrong values with great precision. The floor then says the new system does not match reality, and they are right about the symptom and wrong about the cause. Lose credibility at that point and operators quietly resume keeping their own spreadsheets, and dual bookkeeping becomes permanent.
How to run the count itself, and how to size the hours honestly, is covered in our guide to physical inventory and stocktaking efficiency. Working out what 480 hours becomes under your own SKU count and rack layout is worth doing before you request quotes, not after.
The THB 100,000 of location design pays for one decision — whether you manage to the rack bay or all the way down to the individual shelf level. Finer granularity shortens search time but increases the number of scans during put-away and multiplies label count. Coarser granularity means fewer labels, but multiple SKUs share a location and physical searching survives. There is no universally correct answer. With 2,400 SKUs and 12 operators, you are paying someone to find where the balance sits for your specific case.
A WMS without location accuracy becomes a box where only the quantities are right. The totals reconcile, but where anything physically sits still lives in someone’s head. That is not meaningfully different from running on paper. A quote that omits layer 3 looks cheaper because it is only buying you the box.
Where the benefit comes from, and the THB 2,264,000 annual breakdown
That covers cost. Now the return. Full-scope deployment produces an estimated THB 2,264,000 of annual benefit in this model. Like the cost figures, this is a TOMAS TECH planning model rather than a published statistic.
The benefit breakdown
| Source of benefit | Before | After | Annual reduction |
|---|---|---|---|
| Count variance write-offs, 2% of inventory value falling to 1% | 900,000 | 450,000 | 450,000 |
| Wrong shipments, 500 PPM falling to 125 PPM at THB 20,000 each | 960,000 | 240,000 | 720,000 |
| Time spent physically searching for stock | 90,000 | 30,000 | 60,000 |
| Line stoppage for counting, twice a year falling to once | 288,000 | 144,000 | 144,000 |
| Picking labour, 30 seconds saved per line | — | — | 80,000 |
| Storage cost released by inventory reduction, 10% less stock at an 18% carrying rate | — | — | 810,000 |
| Total | 2,264,000 |
Now the arithmetic behind each line. Wrong shipments — 96,000 lines a year at 500 PPM is 48 incidents, and at 125 PPM it is 12. At THB 20,000 per incident to cover collection, re-shipment, customer handling and the report, THB 960,000 becomes THB 240,000, a reduction of THB 720,000. This is the single largest line item, and it is entirely a function of how well the inspection step is built. How far you can realistically push PPM down depends on the checking design, which our article on shipping and receiving inspection systems works through in detail.
Physical searching is 12 people spending 15 minutes a day over 300 days, or 900 hours a year, at THB 100 per hour for THB 90,000. That falls to 300 hours and THB 30,000. Picking is 30 seconds saved across 96,000 lines, or 800 hours, again at THB 100 per hour for THB 80,000. Line stoppage for counting is 8 hours per count at THB 18,000 per hour, or THB 144,000 per event. Going from two counts a year to one saves THB 144,000.
The largest single item is inventory reduction. Ten percent of THB 45,000,000 is THB 4,500,000 of stock that stops being necessary, and at an 18% carrying rate covering storage, financing and obsolescence, that releases THB 810,000 a year. The mechanism is straightforward — once location accuracy exists, the pattern of ordering more because nobody could find what was already on the rack disappears. Setting the right target levels in the first place is a separate discipline, covered in our article on optimal inventory level management. A WMS does not calculate optimal inventory levels. It creates the conditions under which they can actually be held.
Only 6% of the benefit comes from labour
This is the second core argument of the article. Split those six lines into labour-derived and inventory-and-quality-derived.
| Category | Components | Amount | Share |
|---|---|---|---|
| Labour-derived | Physical searching 60,000 plus picking 80,000 | 140,000 | 6% |
| Inventory and quality derived | Count variance 450,000 plus wrong shipments 720,000 plus count stoppage 144,000 plus inventory reduction 810,000 | 2,124,000 | 94% |
Of THB 2,264,000 in annual benefit, the part that comes from reducing the hours people spend working is THB 140,000, or 6%. The other 94% comes from compressing inventory value and from quality, meaning wrong shipments and count variance.
This structure follows directly from the Thai wage level. The minimum wage in Bangkok rose to THB 400 per day on 1 July 2025, up from THB 372, affecting roughly 700,000 workers (JETRO). At a daily wage of THB 400, removing an hour of work simply does not recover much money. The THB 100 per hour used throughout this model is that wage level with allowances and indirect costs layered on top, which is the realistic loaded figure to plan against.
Make it concrete. Covering the THB 759,000 annual running cost through labour savings alone would require eliminating 7,590 hours a year at THB 100 per hour. If one person works 2,400 productive hours a year, that is more than three full headcount — and that is just to break even on the running cost, before touching the initial investment. In a warehouse of 12 people, removing more than three is removing over a quarter of the operation. A WMS does not deliver that. What it delivers is a thin layer of search time and picking seconds.
The practical implication is a warning. Do not write headcount reduction of three people into the investment approval. The moment that sentence is written, the project is scheduled to be judged a failure twelve months later. Write count variance write-offs, wrong shipment handling costs, and inventory value reduction instead. Those three are visible in the finance numbers and can be tracked honestly. If the head office approval template in Japan demands a labour saving figure, explain the Thai wage structure first and get agreement that the benefit case rests on a different axis.
Full scope or phased, and how to choose
If THB 3,650,000 of initial investment is too heavy to approve, narrowing the scope is a legitimate option. Here is full scope against a phased deployment with 2 interfaces and location setup limited to the shipping area.
| Item | Full scope, 4 interfaces, all locations | Phased, 2 interfaces, shipping area only |
|---|---|---|
| Initial cost | 3,650,000 | 1,915,000 |
| Annual cost | 759,000 | 519,000 |
| Annual benefit | 2,264,000 | 1,577,000 |
| Annual net | 1,505,000 | 1,058,000 |
| Payback | 29 months | 22 months |
| 5-year total | 7,445,000 | 4,510,000 |
| 5-year net | 3,875,000 | 3,375,000 |
The phased build-up works out as follows. Layer 2 is 8 handheld terminals, 5 access points and 2 label printers for THB 650,000, with THB 65,000 maintenance. Layer 3 covers the shipping area only at THB 315,000. Layer 4 is 2 interfaces at THB 700,000, with THB 70,000 maintenance. Layer 5 is THB 250,000 with an annual THB 120,000. Layer 1 is a small cloud plan at THB 22,000 per month, or THB 264,000 a year. That gives an initial THB 1,915,000 and an annual THB 519,000. The benefit falls to THB 1,577,000 because inventory reduction and count variance only half materialise — count variance 270,000, wrong shipments 720,000, searching 30,000, count stoppage 72,000, picking 80,000, inventory reduction 405,000. Wrong shipments keep their full THB 720,000 because inspection works at full strength even when the scope is limited to the shipping area.
Compare the two columns and one difference matters more than the rest. Phased deployment pays back 7 months sooner, 29 months against 22, but leaves THB 500,000 less over five years, THB 3,875,000 against THB 3,375,000. Put another way, buying a faster payback means giving up a little of the five-year return.
Which is correct is not settled by the return calculation. It is settled by which constraint is binding for your company. If the head office investment standard requires payback within 24 months, full scope at 29 months will not be approved and phased is the only route available. If instead the goal is to maximise what is left after five years, or if SKU count is forecast to grow over the next three years, doing it once at full scope ends up cheaper.
One caution if you choose phased. Fix the unit rates for the later expansion in the original contract. Get the incremental price per rack label, per interface and per handheld terminal in writing. Without that, the expansion phase gets priced as a fresh project with no competitive pressure behind it. Phased deployment is not a way to spend less. It is a way to spread the same spending across more budget cycles, and it should be presented internally in exactly those terms.
What is different about deploying a WMS in Thailand
The same system, deployed in Thailand rather than Japan, changes several judgements. Four points matter most.
Warehouse rent and occupancy. In the second quarter of 2026, Thailand’s ready-built warehouse market held a stock of 6.05 million square metres at 85.28% occupancy, with average rent at THB 160 per square metre per month, up from THB 158 the previous quarter, and zero new supply delivered in the quarter (Cushman & Wakefield). A separate study puts 2024 warehouse demand at 6.4 million square metres with 88.0% occupancy and national average rents between THB 110 and THB 230 per square metre per month, with Samut Prakan highest at THB 230 (Krungsri Research). What this environment means in practice is that solving a stock increase by adding floor area is an expensive answer. Adding 1,000 square metres to a 3,000 square metre warehouse at THB 160 per square metre per month costs THB 1,920,000 a year and THB 9,600,000 over five years. That exceeds the entire THB 7,445,000 five-year cost of the WMS. Raising location efficiency to absorb the growth inside your existing footprint is, in pure cash terms, the far more rational move.
Minimum wage. As noted above, the Bangkok minimum wage has been THB 400 per day since 1 July 2025, up from THB 372, covering roughly 700,000 workers. Approaching the project with a Japanese instinct that headcount reduction will fund the payback will always miss. The benefit case has to be built on inventory and quality from the first draft, not retrofitted after the first annual review.
Workforce turnover. Across Asia Pacific, 77% of employers report difficulty securing the talent they need, the highest of any global region (Value Chain Asia). Warehouse roles turn over faster than production line roles. The requirement this produces is specific — if the procedure lives in a person, it collapses when that person leaves. The real test of a warehouse system is whether shipping still runs the week after your most experienced operator resigns. If the handheld terminal tells the operator what to do next on screen, a new hire can stand on the shipping floor the day after induction. You cannot evaluate this from a feature list. During the vendor demonstration, hand the terminal to someone who has never seen the screen before and watch what happens.
Languages and local support. Screens and procedure documents need Thai, Japanese and English. The THB 280,000 in layer 5 covers all three. Deploy a system with an English-only interface and the supervisor layer will manage while the people actually holding the terminals cannot read it. Alongside that, put the local support response time into the contract itself. Two clauses do most of the work — a first response within a stated number of hours for any fault that stops shipping, and business hours defined in Thai time rather than Japanese time. When contracting directly with a Japanese vendor, the time zone question genuinely does get signed without ever being addressed. The THB 144,000 annual local support line, at THB 12,000 per month, is what buys that response time.
Seven decisions to make before you compare quotes
Settle the following seven points internally before the request for quotation goes out. Send it without them and each vendor will assume a different scope, which makes the comparison meaningless no matter how carefully you build the spreadsheet.
| Decision | What to weigh |
|---|---|
| 1. Location granularity | Rack bay or individual shelf level. Finer means faster searching but more scans during put-away and more labels |
| 2. Number and direction of interfaces | Four or two. One-way or two-way. Which system is authoritative when they disagree |
| 3. Handheld terminal count and spares | Headcount plus spares. Cut the spares and the first failure sends someone back to paper |
| 4. Handling of exceptions | Emergency shipments, returns and partial shipments — carried in the system or absorbed by procedure |
| 5. Ownership of master data | Which system is authoritative for items, business partners and units. No duplicate registration anywhere |
| 6. Counting method | Full physical count or cycle counting. Cycle counting reduces line stoppage but adds procedural complexity |
| 7. Post-go-live change rates and response times | Price for one additional report or one additional field, and the guaranteed first response time on faults |
Of these seven, the ones with the largest financial impact are the first three. Point 2 in particular, at THB 525,000 per interface, means the choice between four and eight interfaces alone moves the five-year total by THB 2,100,000. The two most commonly overlooked are points 4 and 7.
Exception handling gets waved through in requirements workshops with the phrase that those cases are rare so procedure will cover it. Rare cases are precisely the ones that generate private spreadsheets on the warehouse floor, and six months later that spreadsheet is the only place where the real answer lives. Count how many times a year each exception actually occurs, and set a threshold — anything over ten occurrences a year goes into the system.
Point 7 bites from the third year onward. Whether an additional report costs tens of thousands of baht or a few thousand determines the shape of your operating expenditure for the rest of the system’s life. Ask for the rate card before signing and attach it to the investment approval. It saves a great deal of argument later.
Frequently asked questions
How much does a WMS cost
For the model warehouse in this article — 3,000 square metres, 2,400 SKUs, 12 operators — a full-scope cloud deployment is THB 3,650,000 initial, THB 759,000 annual, and THB 7,445,000 over five years. A narrower phased deployment is THB 1,915,000 initial, THB 519,000 annual, and THB 4,510,000 over five years. These are TOMAS TECH planning figures and will shift substantially with your interface count and location granularity. Interface count alone moves the five-year total between THB 5,345,000 and THB 9,545,000.
What is a WMS, and how does it differ from an inventory management system
A warehouse management system manages where things physically are and in what order work happens inside the warehouse. An inventory management system holds what exists and how many. A WMS holds which rack and which shelf level, how many are there, and who goes to collect them in what sequence. Handheld-based inbound and outbound control plus location management are the core. Anything lacking those two is a stock ledger, not a WMS. The overlap with ERP inventory functions is apparent rather than real — the ERP is looking at value and period close, while the WMS is looking at physical goods and work sequence.
Which is cheaper, cloud or package
Under the conditions in this article, five-year totals are THB 7,445,000 for cloud and THB 7,165,000 for package. Package is THB 280,000 cheaper, which is around 3% of the total. Rather than deciding on that margin, decide on how far you need to suppress initial expenditure, since the cloud model puts layer 1 initial at zero, and on whether you have anyone in-house to run a server. Custom development sits in a different category entirely at THB 11,945,000 over five years, and should be judged on whether your operation contains genuinely unique processes.
How long does a WMS take to pay back
The model gives 29 months at full scope and 22 months phased. Both assume the benefit is taken from inventory reduction and quality improvement. Attempting to pay back on labour savings alone does not work. Covering THB 759,000 a year at THB 100 per hour would require eliminating 7,590 hours annually, more than three full headcount, and a WMS does not release that much.
Can we start small and expand later
Yes. Starting with the shipping area and 2 interfaces is a sound choice. Two conditions apply. Lock the incremental unit rates for expansion — per label, per interface, per terminal — into the original contract, or the expansion will be priced as a separate project. And decide location granularity at the start regardless of scope. Beginning at rack bay level and later moving to shelf level means relabelling everything and running the full physical count a second time.
What is different about deploying a WMS in Thailand compared with Japan
Four things. First, at a minimum wage of THB 400 per day, labour savings cannot fund the payback, so the benefit case has to be built differently. Second, warehouse rents are rising, at an average of THB 160 per square metre per month in the second quarter of 2026, with occupancy at 85.28%, which makes adding floor area expensive and raises the priority of location efficiency. Third, in a high-turnover labour environment, with 77% of Asia Pacific employers reporting hiring difficulty, procedures must live in the terminal screen rather than in experienced individuals. Fourth, you need Thai, Japanese and English support and a contractual response time expressed in Thai business hours. Those four points separate a Thai deployment from a Japanese one.
Summary
WMS cost is not determined by the price of the software. Of a THB 7,445,000 five-year total, 76%, or THB 5,645,000, sits in the four layers outside the software — hardware, location setup, interfaces, and adoption and operation. Those four layers cost the same whether you buy cloud or package, and the five-year difference between cloud and package is THB 280,000, about 3% of the total.
What genuinely moves the total is the number of interfaces. At THB 525,000 each, going from zero to eight moves the five-year total from THB 5,345,000 to THB 9,545,000, roughly 1.8 times. When quotes diverge, look at interface count and at whether location setup is included, not at assumptions about scale.
The benefit case has to be built differently from a Japanese deployment as well. Of THB 2,264,000 in annual benefit, only THB 140,000, or 6%, comes from labour. The remaining 94% comes from inventory reduction, wrong shipment elimination and count variance. Building the approval case on headcount reduction guarantees the project fails its review. And the scope decision comes down to a straight trade — phased deployment pays back 7 months sooner, full scope leaves THB 500,000 more after five years, and which one is right depends on which constraint actually binds in your organisation.
Deciding just two things in advance, the number of interfaces and the granularity of locations, is enough to turn an incomparable pile of quotes into a genuine comparison. TOMAS TECH builds factory and warehouse systems for Japanese manufacturers across Thailand, and we are happy to work through just those two questions with you. There is no need to be at the quotation stage — get in touch while you are still deciding whether to proceed, and we can start by laying out the five layers against your own warehouse conditions.
References
- Cushman & Wakefield, “Inside Thailand’s Logistics & Industrial Market”, Q2 2026 https://www.cushmanwakefield.com/en/thailand/insights/inside-thailands-logistics-and-industrial-market (Q2 2026 ready-built warehouse stock 6.05 million square metres, occupancy 85.28%, average rent THB 160 per square metre per month, up from THB 158 the previous quarter, with zero new supply in the quarter)
- Krungsri Research, “Industry Outlook 2025-2027 Warehouse”, 29 August 2025 https://www.krungsri.com/en/research/industry/industry-outlook/logistics/warehouse-space/io/warehouse-2025-2027 (2024 warehouse demand 6.4 million square metres, occupancy 88.0%, national average rent THB 110 to THB 230 per square metre per month, Samut Prakan highest at THB 230)
- JETRO, “Bangkok raises daily minimum wage to 400 baht”, 4 July 2025 https://www.jetro.go.jp/biznews/2025/07/b21007a1ac8f7fca.html (Effective 1 July 2025, raised from THB 372 to THB 400, covering roughly 700,000 workers)
- Value Chain Asia, “Asia warehouse labour shortage 2026” https://valuechainasia.com/articles/human-resources/asia-warehouse-labour-shortage-2026 (77% of Asia Pacific employers report difficulty securing talent, the highest of any global region)
- Inter-Stock, “The cost of implementing a WMS warehouse management system” https://www.inter-stock.net/column/no525/ (Japanese domestic market rates. Cloud initial JPY 0 to 1 million with a monthly fee of JPY 30,000 to 300,000, on-premise initial JPY 5 million to 30 million, custom development JPY 30 million to over 100 million)
