Purchase orders go out every day and the goods arrive on time. It is tempting to conclude that procurement is under control. Then an audit lands, or head office sends a query, and the document you cannot produce is not the purchase order. It is the record that shows why that supplier, at that price. This article is not about how to choose a whole order management system. It is about the buy side that a purchasing management system actually covers, and how to design one continuous line from competitive quotes through approval, ordering, goods receipt and payment matching.
What a purchasing management system is, and why the difference from an order management system is a matter of scope
Inside a single company, the phrase “purchasing management system” tends to carry about three different meanings. To one person it is the tool that issues purchase orders. To another it is electronic sign-off on internal approval requests. To a third it is the entire relationship with suppliers. Meetings talk past each other because all three meanings are in the room at once.
For this article, a purchasing management system means a mechanism that decides what to buy, at what price and from whom, then approves it, receives it and pays for it, and keeps all of that as one record. The scope is the buy side only. It does not cover the sell side, where your own company receives orders from customers.
Five core functions
What a typical purchasing management system does can be grouped into five functions.
| Function | What it does | Record left behind |
|---|---|---|
| Purchase request and competitive quotes | Takes a request from the floor and collects quotes from several suppliers | Requester, request date, the content of the quotes compared |
| Approval workflow | Routes to approvers according to value band and item category | Approver, approval timestamp, whether it was sent back |
| Purchase order issue and dispatch | Generates a PO automatically from the approved content and sends it | PO number, dispatch date, agreed unit price |
| Goods receipt management | Registers physical receipt and the check on quantity and quality | Receipt date, quantity received, inspector |
| Payment matching | Reconciles the purchase order, the receipt and the invoice | Nature of any mismatch and the history of its correction |
Of these five, the one that gets digitised first, in a plant in Japan or a plant in Thailand alike, is the third one, PO issue. The reason is simple. A purchase order is a document that leaves the company. Which also means that the first and second functions, the ones that never leave the building, tend to stay in spreadsheets and email. That is the part that comes back to bite you.
Where the boundary with an order management system sits
The wider frame, including the sell side and how to divide responsibilities across systems, is covered in the article on choosing an order management system and drawing the boundary with purchasing and MRP. That article exists to decide which function belongs in which system.
This one goes inside that boundary. Once you have decided which functions sit on the purchasing side, in what sequence and on what data structure should they actually run? Similarly, working out when and how many of each part you need is the job of requirements planning, and how much that accuracy depends on master data is set out in the article on MRP system master data accuracy. A purchasing management system takes that calculated result and decides who to buy from and at what price.
What has gone missing is not the purchase order but the record of quotes and approvals
Look concretely at what is actually absent in a plant running procurement on email and spreadsheets.
Three kinds of record disappear
The first is the price negotiation history. A quote arrives by email, someone calls and asks whether the price can come down, and a revised quote arrives. Of that exchange, what usually survives on the file server is the last message only. What the original number was, and on what grounds it moved, exists only in one buyer’s memory. When that buyer transfers, the memory goes with them.
The second is the record of who approved. When approval requests circulate on paper, the stamps remain. But the story behind them does not. “The amount went up so it went round again.” “It was sent back once and the terms were changed.” A paper approval form does not carry that. An electronic workflow records a send-back automatically. On paper and email, it exists only if somebody deliberately writes it down.
The third is the reasoning for choosing that supplier. “Company A hit the delivery date last time as well.” “Company B is cheaper but slow to ramp up.” These judgements happen inside the buyer’s head. Few plants can explain, after the fact, why they picked the second-cheapest of three quotes.
The practical damage from missing all three is close to zero in normal times. It shows up when prices rise, when deliveries slip, and when somebody outside asks you to explain yourself.
Governance expectations from head office are shifting
For a subsidiary in Thailand, the party most likely to ask for that explanation these days is the Japanese head office.
Part of the background is a review of transaction rules inside Japan. Japan’s Subcontract Act was amended by legislation promulgated on 23 May 2025, and from 1 January 2026 it was renamed and rewritten as the Act on Ensuring Appropriate Transactions with Small and Medium-Sized Contractors, often shortened in Japanese usage to the “Toritekiho”. Terminology was reorganised too. The parent business became the commissioning business, the subcontractor became the small and medium-sized contracted business, and subcontract payment became consideration for commissioned manufacturing and similar work.
Substantively, the amendment added a prohibition on setting a price without proper consultation, which addresses the practice of leaving prices frozen, a prohibition on payment by promissory note and similar instruments, the addition of transport commissioning to the covered transaction types, and an expansion of scope through an employee-count criterion. The stated aim is to make appropriate price pass-through stick across the whole supply chain, given the rise in labour, raw material and energy costs.
One point needs to be stated plainly to avoid a misreading. The Toritekiho is Japanese law covering commissioned transactions within Japan, and it does not apply directly to transactions between a Thai subsidiary and its suppliers in Thailand. If you come across commentary that suggests procurement in Thailand is bound by this law, treat it with suspicion.
The practical ripple effect, on the other hand, is real and already happening. When a Japanese head office responds to the amendment by codifying the issue of written purchase orders, the recording of price consultations and the clear statement of payment terms as internal standards, those standards frequently get rolled out as group-wide internal control requirements. The result is that a subsidiary in Thailand starts receiving questions such as “are purchase orders always issued in writing or as data” and “is there a record of consultation behind how unit prices were set”. Not as a legal obligation, but as conformity with a group standard.
At that moment, without the three records described above, you have nothing to answer with. Put the other way round, the honest motivation for adopting a purchasing management system is more often this explainability than cost reduction. As the costing section will show, trying to justify it on money alone gets uncomfortable.
Six variables to isolate where the effort in your purchasing process really sits
“Should we adopt a purchasing management system” is not a question that can be answered as asked. Which step of procurement is heavy differs by plant, and buying functionality that does not touch the heavy step produces no result.
So score your own purchasing process on six variables. Rate each from 0 to 3 points, for a total between 0 and 18.
| Variable | What it measures |
|---|---|
| Variable 1 Monthly PO count | Purchase orders issued. Counted by number of orders, not by value |
| Variable 2 Active supplier count | Suppliers with at least one transaction in the last 12 months |
| Variable 3 Share of orders requiring competitive quotes | Proportion of all orders for which internal rules mandate competitive quotes |
| Variable 4 Number of approval levels | Steps of approval between raising a request and confirming the order |
| Variable 5 Share of imported procurement | Share of purchase value sourced from overseas suppliers |
| Variable 6 Complexity of payment currencies and withholding tax categories | Number of currencies, and how far tax categories split by supplier |
The scoring criteria for each variable follow. Read them with your own figures to hand.
Variable 1 Monthly PO count
Transaction count is the right measure because approval and matching effort scales with the number of documents, not with value. A plant that places two equipment orders a year at 10,000,000 THB each gets less out of a purchasing management system than a plant placing 200 consumable orders a month at 5,000 THB each.
| Points | Condition |
|---|---|
| 0 points | Fewer than 50 |
| 1 point | 50 or more, fewer than 150 |
| 2 points | 150 or more, fewer than 400 |
| 3 points | 400 or more |
Variable 2 Active supplier count
Measure the number you actually trade with, not the number of records in the supplier master. A hundred companies you bought from once a decade ago add almost nothing to operational difficulty. Active suppliers, on the other hand, each add unit prices, payment terms and tax categories to manage.
| Points | Condition |
|---|---|
| 0 points | Fewer than 20 |
| 1 point | 20 or more, fewer than 60 |
| 2 points | 60 or more, fewer than 150 |
| 3 points | 150 or more |
Variable 3 Share of orders requiring competitive quotes
Measure this as how many orders your rules say require competitive quotes, not how many actually get them. This is not a compliance rate. A low compliance rate is a symptom that argues for a system, not the value of the variable.
| Points | Condition |
|---|---|
| 0 points | Under 5%. Almost everything runs on unit price agreements or nominated sourcing |
| 1 point | 5% or more, under 15% |
| 2 points | 15% or more, under 30% |
| 3 points | 30% or more |
Variable 4 Number of approval levels
Look not only at the number of steps but at whether the approver changes with the value band. Value-band branching on paper and email produces mistakes almost without fail, because a human decides who to send it to.
| Points | Condition |
|---|---|
| 0 points | One level. The buyer completes it at their own discretion |
| 1 point | Two levels on a fixed approval route |
| 2 points | Three levels, or an approver that branches by value band |
| 3 points | Three or more levels, and sign-off from an overseas head office occurs routinely |
Variable 5 Share of imported procurement
More importing means a longer lead time from order to arrival, and with it the question of who carries the exchange rate and price movement in between. The unit price at order time stops matching the amount at payment time, which lifts the difficulty of payment matching by a notch.
| Points | Condition |
|---|---|
| 0 points | Under 5%. Almost entirely domestic sourcing |
| 1 point | 5% or more, under 20% |
| 2 points | 20% or more, under 50% |
| 3 points | 50% or more |
Variable 6 Complexity of payment currencies and withholding tax categories
This is where running procurement in Thailand differs most from Japan. The detail comes in the supplier master section, but the scoring belongs here.
| Points | Condition |
|---|---|
| 0 points | Baht only. Effectively a single withholding tax category |
| 1 point | Baht only, but withholding tax splits into several categories |
| 2 points | Two currencies, with withholding tax split into several categories |
| 3 points | Three or more currencies, with withholding tax split into several categories |
Reading the total
Map the six-variable total onto three bands.
| Total | Verdict |
|---|---|
| 0 to 5 points | Do not proceed. A standard spreadsheet ledger format and a fixed storage location will do |
| 6 to 11 points | Partial adoption. Start with either competitive quotes or the approval workflow |
| 12 to 18 points | Full adoption. Connect competitive quotes through to payment matching in one line |
If you land in the middle band, which end to start from is decided by the scores on variables 3 and 4. If variable 3 is greater than or equal to variable 4, start with the record of competitive quotes. If variable 4 is greater than variable 3, start with the approval workflow. The rule is nothing more than starting where the centre of gravity of the effort lies.
There is one override. If variable 6 scores 3 points, put redesigning the supplier master at the top of the list regardless of the total. With three or more currencies and split tax categories, digitising only the workflow moves the problem rather than solving it. Approval gets faster while payment stalls every single time.
Running the numbers on a model plant, one year of indirect materials in a Japanese-owned factory in Thailand
Rather than keep this abstract, here is a plausible plant with figures attached.
Assumptions
A Japanese-owned maker of moulded resin components in Rayong province, 210 employees. Direct materials, meaning resin pellets and colourants, run on unit price agreements and requirements planning, so they are excluded from this costing. The scope is indirect materials and consumables, meaning mould components, tooling, maintenance parts, secondary materials, office supplies and outsourced equipment work.
| Assumption | Value |
|---|---|
| Annual indirect material spend | 28,800,000 THB |
| Annual PO count | 2,880 orders (240 a month) |
| Average order value | 10,000 THB |
| Competitive quote band (20,000 THB or more per order) | 480 orders / 14,400,000 THB (average 30,000 THB) |
| Low value band (under 20,000 THB per order) | 2,400 orders / 14,400,000 THB (average 6,000 THB) |
| Registered suppliers | 210 |
| Active suppliers (last 12 months) | 86 |
| Buyers | 3 |
| System users including requesters | 8 |
| Share of imported procurement (indirect materials, by value) | 18% |
| Payment currencies | Two, baht and Japanese yen |
| Approval levels | Two. Japanese head office sign-off is added only above 500,000 THB, which applies to roughly 3 orders a year and is not routine |
| Hourly cost of purchasing and finance staff including employer burden | 260 THB |
Applying the scoring model
Scoring this plant on the six variables produces the following.
| Variable | Condition applied | Points |
|---|---|---|
| Variable 1 Monthly PO count | 240 orders (150 or more, fewer than 400) | 2 points |
| Variable 2 Active supplier count | 86 (60 or more, fewer than 150) | 2 points |
| Variable 3 Competitive quote share | 480 of 2,880 orders = 16.7% (15% or more, under 30%) | 2 points |
| Variable 4 Approval levels | Two levels, but branching to head office by value band. Head office sign-off is non-routine, so the 3 point condition does not apply | 2 points |
| Variable 5 Share of imported procurement | 18% (5% or more, under 20%) | 1 point |
| Variable 6 Currencies and tax categories | Two currencies, several withholding tax categories | 2 points |
The total is 11 points. That falls in the 6 to 11 band, so the verdict is partial adoption, and it sits right at the top of the band. Variables 3 and 4 are tied on 2 points each, so the rule gives the conclusion that this plant should start with the record of competitive quotes.
Option A, first establish what the current state is already losing
The baseline for comparison is “carry on without a system”. But the current state does not cost zero. It simply leaks a fixed amount every year without appearing on any invoice. To avoid exaggeration, only three items are counted as benefits. Administrative time savings are not counted, for reasons given below.
The first leak is duplicate ordering. Because orders already placed but not yet received are invisible, the same request comes up from the floor twice and gets ordered twice. Assume this happens on 1.25% of the 2,880 annual orders, that is 36 orders. At 10,000 THB each that is 360,000 THB, but not all of it is lost. Most of it either gets consumed as stock or returned. Put the real loss, as dead stock and return handling charges, at 40%, giving 144,000 THB.
The second is weakened price negotiation. Of the 14,400,000 THB in the competitive quote band, quotes are actually collected on 55%, and the remaining 45%, or 6,480,000 THB, flows to “the same supplier as last time, on the same terms as last time”. Put the price reduction achieved by collecting quotes at 2.0%, and the missed saving is 2.0% of 6,480,000 THB, which is 129,600 THB. On top of that, past unit prices in the 14,400,000 THB low value band cannot be searched, so last time’s price simply gets rubber-stamped. Assume that laying out historical prices and reviewing once a year takes 0.5% off, which is 72,000 THB. Together, 201,600 THB leaks on the price side.
The third is payment matching errors. The plant receives 2,400 invoices a year, fewer than the 2,880 orders because some suppliers combine several orders on one invoice. Reconciling purchase order, receipt record and invoice throws up a mismatch on 3.5% of them, that is 84 invoices. Correcting one takes 45 minutes of combined purchasing and finance time, so 84 invoices at 45 minutes is 3,780 minutes, or 63 hours, and at 260 THB an hour that is 16,380 THB. On top of that, 12% of the mismatches, roughly 10 invoices, go unnoticed and are overpaid. Put the overpayment at 4,500 THB each and that is 45,000 THB. The total is 61,380 THB.
Adding the three items, the current state leaks 406,980 THB every year.
A word on why administrative time savings were left out. A purchasing management system genuinely shortens the handling time per transaction. But this plant has 3 buyers. Even if a full person’s worth of hours is freed up, that person is not made redundant, they move to other work. If you count a saving that does not reduce spending as a benefit, your year three report on the investment will fall apart. Not counting it is the honest choice.
Option B, partial adoption, digitising from request to PO issue
In line with the verdict, this configuration starts with quote attachment and approval records. Purchase requests are raised on screen, attaching the quote file is mandatory, the approval route runs automatically according to the value band, and the purchase order is generated from the approved content and sent. Goods receipt and payment matching continue on spreadsheets and paper as before.
| Cost item | Amount (THB) |
|---|---|
| Initial cost (configuration, approval route definition, supplier master migration, training) | 380,000 |
| Initial cost amortised over five years | 76,000 |
| Annual subscription (8 users at 1,000 THB for 12 months) | 96,000 |
| Total annual cost | 172,000 |
The benefits are estimated as follows. Duplicate ordering falls by 60%, because order history becomes searchable and the duplicate is caught at approval, giving 60% of 144,000 THB, or 86,400 THB. On price, mandatory quote attachment lifts the competitive quote rate from 55% to 80%. Of the 129,600 THB missed saving, 25 of the 45 percentage points are recovered, so 129,600 x 25 / 45 = 72,000 THB. The quoting function does not reach the low value band, so that contributes 0. Payment matching sees fewer mismatches because the order unit price becomes confirmed data, but the reconciliation work itself stays manual, so a 25% reduction gives 25% of 61,380 THB, or 15,345 THB.
Adding up, the annual benefit is 86,400 + 72,000 + 15,345 = 173,745 THB. Against an annual cost of 172,000 THB, the net is plus 1,745 THB.
Option C, full adoption, connecting competitive quotes through to payment matching
Sending and collecting quote requests, approval, ordering, entering goods receipt and three-way matching against the invoice all happen in one system, integrated with the accounting system.
| Cost item | Amount (THB) |
|---|---|
| Initial cost (requirements definition, supplier master redesign, accounting integration, receipt terminals, training) | 1,450,000 |
| Initial cost amortised over five years | 290,000 |
| Annual subscription and maintenance (8 users at 2,750 THB for 12 months) | 264,000 |
| Additional annual operating effort (master maintenance and supplier registration, 120 hours at 260 THB) | 31,200 |
| Total annual cost | 585,200 |
Now the benefit side. Duplicate ordering falls by 70%, because receipt and stock status are visible too, giving 70% of 144,000 THB, or 100,800 THB. On price, the competitive quote rate rises to 95%, so 129,600 x 40 / 45 = 115,200 THB, plus 60% of the 72,000 THB low value band review benefit, or 43,200 THB, for a combined 158,400 THB. Payment matching falls by 70% because three-way matching completes inside the system, giving 70% of 61,380 THB, or 42,966 THB.
Adding up, the annual benefit is 100,800 + 158,400 + 42,966 = 302,166 THB. Against an annual cost of 585,200 THB, the net is minus 283,034 THB.
The three options side by side
| Option | Annual cost (THB) | Annual benefit (THB) | Net (THB) | Recovery of the 406,980 current leak |
|---|---|---|---|---|
| A Current state (email and spreadsheets) | 0 | 0 | 0 | 0% |
| B Partial adoption (request to PO) | 172,000 | 173,745 | +1,745 | 42.7% |
| C Full adoption (quotes to payment matching) | 585,200 | 302,166 | -283,034 | 74.2% |
Here is the honest reading. At this scale, full adoption does not pay for itself on money alone. The benefits capture 74.2% of the current leak, but since the ceiling on what can be captured is 406,980 THB, there was never any room to justify an annual cost of 585,200 THB financially. Partial adoption roughly breaks even, which in substance means “spending money to acquire records”.
This does not mean a purchasing management system is useless. It means that at this plant’s scale, an approval request built on money will not get signed. If you want full adoption approved, the case has to rest on explainability and control.
How large do you have to get before it pays back
So what size does pay back? Take the case where indirect material procurement for three plants in Thailand is consolidated into one purchasing centre. Spend and order volume triple, and system users rise from 8 to 14.
| Item | Amount (THB) |
|---|---|
| Annual benefit (302,166 for three plants) | 906,498 |
| Initial cost (1,450,000 plus 450,000 to roll out to the other two sites) | 1,900,000 |
| Initial cost amortised over five years | 380,000 |
| Annual subscription and maintenance (14 users at 2,750 THB for 12 months) | 462,000 |
| Additional annual operating effort (240 hours at 260 THB, doubled rather than tripled because consolidation removes duplicated work) | 62,400 |
| Total annual cost | 904,400 |
| Net | +2,098 |
Consolidating three plants finally produces a surplus of 2,098 THB a year. That is within the margin of error, so in substance it is the break-even point. In other words, full adoption for indirect material procurement is an investment that only stacks up financially once you bundle three plants together.
Sensitivity analysis, what changes if the quote-driven price reduction is halved
The most fragile assumption in the costing above is the 2.0% price reduction from collecting competitive quotes. If many of your items are already competitively priced, quotes will not move the number much. Change this one input only.
Cutting the reduction from 2.0% to 1.0% takes the missed saving in the competitive quote band from 129,600 THB down to 64,800 THB. What Option C recovers becomes 64,800 x 40 / 45 = 57,600 THB, and what Option B recovers becomes 64,800 x 25 / 45 = 36,000 THB. The 43,200 THB low value band review benefit depends on a different mechanism, visibility of past unit prices rather than competitive quoting, so this coefficient is not applied to it. The duplicate ordering and payment matching benefits are likewise unrelated to the price reduction rate and stay as they were.
| Item | Option B base (THB) | Option B at 1.0% (THB) | Option C base (THB) | Option C at 1.0% (THB) |
|---|---|---|---|---|
| Duplicate order prevention | 86,400 | 86,400 | 100,800 | 100,800 |
| Price in the competitive quote band | 72,000 | 36,000 | 115,200 | 57,600 |
| Unit price review in the low value band | 0 | 0 | 43,200 | 43,200 |
| Reduction in payment matching errors | 15,345 | 15,345 | 42,966 | 42,966 |
| Total annual benefit | 173,745 | 137,745 | 302,166 | 244,566 |
For Option C at a single plant the net becomes 244,566 minus 585,200, or -340,634 THB. Even consolidated across three plants, 244,566 for three plants is 733,698 THB against a cost of 904,400 THB, so the net is -170,702 THB. The conclusion that full adoption does not pay back is not reversed. If anything it gets stronger.
Option B, on the other hand, does flip. The benefit falls from 173,745 THB to 137,745 THB, and against an annual cost of 172,000 THB the net turns into a loss of -34,255 THB. Even the partial adoption that the scoring model recommended loses its surplus after moving a single assumption. That is the clearest thing this costing has to say. Almost no plant can predict with confidence, before adoption, whether quote-driven price reduction will come in at 2.0% or 1.0%. Which means an approval request built on money has no firm footing under either option.
Note also that at a 1.0% reduction the current leak itself falls from 406,980 THB to 342,180 THB. The recovery percentages in the three-option table are for the base case at 2.0%.
Settle the supplier master design first
Everything above has been about system functionality. But what actually stalls an implementation is master data more often than functionality. In Thailand in particular, there are specific fields whose design differs from Japan.
The withholding tax category is already fixed at the moment the order is raised
In Thailand, withholding tax applies when a supplier that is a juristic person is paid for services and similar items. The rate splits by the type of payment, running roughly from 1% to 5%. However, the line between which transaction falls into which category involves practical judgement, so always confirm the category with your tax advisor. This article deliberately does not publish a rate table.
Filing is done on PND3 for individuals and PND53 for juristic persons. For payments made in a given month, the deadline is within the first 7 days of the following month for a paper filing, or within the first 15 days where the Revenue Department e-filing service is used, under an extension granted by a Ministry of Finance notification.
For the design of a purchasing management system, the rate itself is not the important part. What matters is that this category is not information determined at payment time. It is information already determined at the moment the purchase request is raised. The instant you decide what you are buying, it is settled whether this is a purchase of goods, a supply of services, a lease or transport. Therefore the category is settled too.
Despite that, in many plants this information is not carried on the order data. The purchase order has only item name, quantity and unit price, and at the payment stage finance decides the category by looking at the invoice, then calls purchasing whenever it is unsure. Those calls are a leading cause of delay in payment matching.
The remedy is straightforward. Hold a tax category field on both the supplier master and the item category, and have it populate automatically when the order is raised. One caution here. The tax category is an attribute of the transaction, not an attribute of the supplier. If you buy both goods and services from the same company, holding a single category on the supplier master guarantees errors. Hold “the categories that can arise with this supplier” on the supplier master, and let the item category decide the actual one. Without that two-stage design, fixing it later is extremely painful.
Payment terms and lead time belong in the master
Beyond the withholding tax category, other information is fixed at order time and has to travel all the way to payment.
| Field | What is settled at order time | What happens without it |
|---|---|---|
| Payment terms | Cut-off and payment dates, and whether the clock starts at receipt or at invoice receipt | Finance memorises them supplier by supplier and no payment forecast is possible |
| Standard lead time | The reference number of days from order to arrival | You cannot judge whether a delivery was late, and supplier evaluation becomes subjective |
| Receipt method | Physical receipt note, EDI receipt data, or registration on a receiving terminal | The receipt date differs depending on who entered it, and payment matching does not reconcile |
| Currency and exchange handling | Order currency, and which date the conversion uses | Order value and invoice value do not agree, and every difference needs explaining |
Of these, receipt method is the one plants surprisingly often have not settled. When “the day the goods arrived”, “the day the warehouse accepted them”, “the day quality assurance passed them” and “the day finance cut the voucher” all coexist, the matching key for payment reconciliation has no fixed definition. Before implementation, pick exactly one and decide which of them counts as the receipt date.
Approval routes cannot be settled until the master is
When designing an approval workflow, the condition that selects an approver is usually value. In practice, though, other conditions always join it. Whether it is imported. Whether it is capitalised. Whether the supplier is new. Whether it sits within a unit price agreement. Every one of those is an attribute held in the master.
Which means approval route design comes after master design. Reverse the order and you end up with a design where a human picks the branch condition by hand, and errors do not go down even though you digitised.
Matching after the invoice arrives is a separate story
Everything covered so far happens before the invoice arrives. Collect quotes, obtain approval, issue the purchase order, receive the goods. If that chain of records is in order, matching when the invoice arrives becomes a simple act of comparison.
The other side of the line, meaning how to read an invoice that arrives on paper or as a PDF and get it into three-way matching, is a different technical area. Reading and matching automation using AI-OCR is written up separately in the article on invoice processing automation.
The two are adjacent in time but independent as investment decisions. If you deploy invoice OCR while the buy-side records are still incomplete, there is nothing on the other side to match the scanned invoice against, so nothing improves. Purchasing comes first in sequence. Conversely, once purchasing is in order, the invoice side can be added afterwards.
Points specific to a Japanese-owned plant in Thailand
Do not confuse conformity with group standards for legal compliance
As stated earlier, Japan’s Toritekiho does not itself apply directly to a Thai subsidiary. But the internal control standards that head office builds do come down to group companies as operating rules for procurement.
What tends to go wrong locally is importing the head office standard verbatim and watching operations seize up. Approval hierarchies and document formats premised on Japanese business custom sometimes do not work with Thai suppliers. It is far more practical to redesign, locally, a different means of satisfying the intent of the standard, which is that order content is fixed in writing or as data, that there is a record of how the price was set, and that payment terms are stated explicitly. Systemisation is a useful tool for building that “different means of satisfying the intent”.
Currencies and languages are mixed
Suppliers to a Japanese-owned plant in Thailand usually fall into three layers. Japanese trading houses and component makers, local Thai companies, and import sources in China or elsewhere in ASEAN.
| Supplier layer | Working language | Currency | Purchase order format |
|---|---|---|---|
| Japanese trading houses and makers | Japanese | Baht or Japanese yen | A Japanese-language order form is sometimes requested |
| Local Thai companies | Thai or English | Baht | An English PO usually works |
| Import sources | English | US dollars and others | An English PO with explicit invoice terms is essential |
This mixture is an easily overlooked requirement during selection. Whether the product can hold purchase order templates by language, and whether the supplier master can carry a language attribute, are things to check at the demo stage. If it cannot, you end up sending a separate Japanese order form built in a word processor, and the purchase order in the system stops matching the document you actually sent.
Who performs goods receipt
One more thing that often causes friction in plants in Thailand is who performs goods receipt. Does the warehouse receiving clerk complete the goods receipt, or does the requesting department check the goods and sign them off? With indirect materials, it frequently happens that the requester takes delivery directly and no record reaches the warehouse or purchasing at all.
If you are putting in a system, settle this as an operating rule first. Simply establishing the principle that an order with no receipt registration cannot proceed to payment removes a large share of the missing records. That principle does add work on the floor, however, so pair it without exception with a design that lowers the effort of registering, such as a mobile device that finishes the job in a few taps.
Contractual questions that arise when the system is built by an external vendor are covered in the article on system development contract issues.
Frequently asked questions
What is a purchasing management system
It is a mechanism that decides what to buy, at what price and from whom, then approves it, receives it and pays for it, and keeps all of that as one record. The core functions group into five, purchase request and competitive quotes, approval workflow, PO issue and dispatch, goods receipt management, and payment matching. The scope is the buy side only. It does not cover the sell side, where your company receives orders from customers.
How does it differ from an order management system
The coverage differs. An order management system is the wider frame including both the sell side and the buy side, and it comes with the work of deciding which function belongs in which system. A purchasing management system takes only the buy side and goes deep. For how to draw the boundary itself, read the article on order management systems first.
Can we digitise competitive quoting alone
Yes. In fact the scoring model in this article recommends starting from competitive quotes whenever the share of orders requiring them (variable 3) is greater than or equal to the number of approval levels (variable 4). Single-purpose products exist that cover sending and collecting quote requests, generating comparison tables automatically, and searching past quotes. Be aware, though, that without a connection to PO issue you are left with a state where the approved quote and the order actually placed do not necessarily agree.
How much does a purchasing management system cost
The order of magnitude changes with the configuration. For the model plant in this article, partial adoption covering request through PO issue was set at 380,000 THB initial and 96,000 THB annual subscription, and full adoption covering competitive quotes through payment matching at 1,450,000 THB initial and 264,000 THB annual subscription and maintenance. Those are one example scoped to indirect material procurement only. Include direct materials or outsourced processing and the requirements change.
Is it worth adopting at a smaller plant
On money alone, in most cases it is not. For the model plant in this article, with annual indirect material spend of 28,800,000 THB and 2,880 orders, full adoption produced an annual benefit of 302,166 THB against an annual cost of 585,200 THB, so it did not pay back. Only consolidating three plants brings it to break-even. If it is worth adopting, the reason is not the money. It is that the record of quotes and approvals survives, which is to say explainability towards head office and auditors. Whether you can write that squarely into the approval request is what decides it.
Summary
Evaluating a purchasing management system should not start from digitising the purchase order. That is usually already digitised, and what has disappeared is the record of competitive quotes and approvals. The price negotiation history, the record of approvers and send-backs, and the reasoning for choosing that supplier. The damage from losing those three surfaces when prices rise and when somebody outside asks for an explanation.
How much depth you need can be judged by scoring six variables from 0 to 3 each, monthly PO count, active supplier count, share of orders requiring competitive quotes, number of approval levels, share of imported procurement, and complexity of payment currencies and withholding tax categories, then reading the total. In the middle band, whichever of variables 3 and 4 is larger decides where you start.
Then there is the money. For the model plant here, the current state leaks 406,980 THB every year. Partial adoption roughly breaks even, with an annual benefit of 173,745 THB against a cost of 172,000 THB, and full adoption does not pay back, with a benefit of 302,166 THB against a cost of 585,200 THB. Consolidating three plants gets you to a net of just +2,098 THB. Run a sensitivity analysis that drops the quote-driven price reduction from 2.0% to 1.0%, and full adoption turns to a loss of -170,702 THB even when consolidated, while partial adoption flips from +1,745 THB to -34,255 THB. Since moving a single assumption wipes out the surplus, justifying this on money alone is a stretch at this scale. Put the justification on the explainability side.
Finally, one thing you should not rearrange is the order of design. Give the supplier master and the item category their withholding tax category, payment terms, standard lead time, receipt method and currency handling. Then design the approval routes. Reverse that order and you get a system that is digital but no less error-prone.
We are happy to talk at the stage of working out what your own six-variable score is, or whether partial or full adoption is the right fit. Give us three numbers, monthly PO count, active supplier count, and the share of orders requiring competitive quotes, and we will return a first-pass verdict and an approximate benefit figure. As a way of getting organised before you request product quotations, please get in touch through our contact page. We will describe the scope that is neither excessive nor underpowered, based on what we have seen implementing these systems in Thailand.
References
- Keiyaku Watch, Commentary on the Subcontract Act amendment effective 1 January 2026 and the Act on Ensuring Appropriate Transactions with Small and Medium-Sized Contractors, keiyaku-watch.jp
- Japan Fair Trade Commission, Leaflet on the Act on Ensuring Appropriate Transactions with Small and Medium-Sized Contractors, jftc.go.jp
- Gentle Law IBL, Thailand Withholding Tax Filing 2026 PND3 vs PND53 Deadlines and e-Filing Rules, gentlelawibl.com
- ExpatDen, Withholding Tax in Thailand, expatden.com
- Thai Revenue Department, English Forms, rd.go.th
- IT Trend, Core functions of a purchasing management system and how to choose one, it-trend.jp
