Overseas Plant Quality Management System 2026 — Certification Does Not Guarantee One Standard Across Sites

When a quality problem surfaces at an overseas plant, the first thing the head-office quality assurance department usually checks is whether that site holds ISO 9001. But the presence or absence of a certificate is not a guarantee that multiple sites are being operated to the same standard. This article is written for executives and quality assurance leads who are considering an investment in an overseas plant quality management system, and it deals with the one thing you should confirm before you choose a system: whether you are currently able to measure how far each site has drifted from the head-office standard.

Overseas Plant Quality Management System 2026 — “Certified” Does Not Guarantee a Single Standard Across Sites

At Japanese-owned manufacturers with several overseas sites, quality decisions often take the following shape. Once a year, each site reports its defect rate and its count of corrective actions to head office. The numbers are flat year on year, or slightly improved. Every site maintains its ISO 9001 certification. Therefore, the conclusion goes, the quality management structure is working.

What is dangerous about that conclusion? A certificate is a record that a certification body has confirmed that a management system has been built and is being operated in a way that satisfies the requirements of a standard. It is not proof that the daily work at each site is being carried out the way head office assumes it is. These are two different things, and the gap between them widens as the number of sites grows.

The argument of this article comes down to a single point. Before you deploy a quality management system across your overseas plants, you should first put in place a way to measure how far apart the standards at your sites actually are; committing the same amount of money to every site from the outset is premature. There are two reasons. First, the certification scheme itself is not designed to look at every site every year. Second, cost of quality figures discussed as a company-wide average bury a badly deviating site inside that average.

Both of these are structural problems that arise from the nature of the scheme and the nature of the metric. Neither is a matter of insufficient effort on the shop floor. We will take them in turn.

Note that the broader question of how far to impose the head-office standard and where to leave things to the local site is not specific to quality; it is a common thread running through systems in general, including ERP and production management. The general design argument about data layers is covered in Overseas Plant System Rollout: Drawing the Line Between the Head-Office Standard and Local Optimization, so this article confines itself to the quality and audit domain.

One more point on timing. The revised edition of ISO 9001 is scheduled for publication in September 2026, with the development period extended to 36 months. After publication, there will be a three-year transition period running to September 2029. This comes from the official update published by the ISO/TC 176/SC 2 subcommittee, so it can be stated as a primary source. If the work of reviewing every site’s documentation for the revision is going to happen anyway, it makes sense to use that same moment to check whether the standards across your sites are aligned in the first place. The revision can serve as the occasion for taking stock of the drift.

Why Quality Standards Drift Apart as You Add More Sites

Things that were not a problem when you had one site become visible at three or four. The causes fall broadly into three groups.

The Blank Spaces in a Work Instruction Get Filled In Differently at Each Site

It has been reported that at overseas plants there is a working assumption on the ground that anything not written in the work instruction is left to the operator’s discretion. On a Japanese shop floor, the parts not written down are filled in by a shared, unspoken sense of how things are obviously done. Because that tacit understanding does not carry over, the blank spaces in the instruction turn directly into variation in how the work is performed. The same commentary lists loss of nuance in translation, high workforce turnover, and unauthorized process changes by suppliers as typical problems. This comes from media published by a vendor of shop-floor improvement tools rather than from statistically backed primary research, but anyone who has run multiple sites will likely recognize the description.

What matters is that the way each blank is filled in is decided independently at each site. Head office believes it distributed the same work instruction; in practice, the work that has actually taken hold has branched into as many versions as there are sites. The fact that a physical document is identical does not mean the work is identical.

“We Instructed Them” Quietly Becomes “It Was Understood”

There is a reported case from a Japanese-owned manufacturing site in Thailand in which managers worked out a countermeasure and rolled it out purely by issuing instructions to the operators. The only thing that registered on the operator side was that the work method had been changed by order, and the result was inspections performed using the wrong method, and check sheets filled in without any inspection having taken place at all. The same report also points to a vicious circle in which excessive dependence on Japanese expatriates and visiting staff drives up cost and creates confusion on the floor. This is secondary information from business media, but it is accompanied by a concrete description of the situation, and it shows that the way a countermeasure is rolled out is itself decisive for the outcome.

What makes this hollowing-out so awkward is that on paper everything looks normal. The check sheets are filled in. The defect rate does not go up. Nothing appears in the numbers that reach head office. It surfaces only when it is pushed in from outside, in the form of a customer complaint or an audit finding.

Lagging Digitalization Makes the Drift Invisible

According to the FY2025 survey of Japanese companies operating overseas, Asia and Oceania edition, published by JETRO, the share of companies using digital technology in ASEAN stands at only 52.1 percent, clearly below Australia, South Korea and India, which are all above 60 percent. Among the obstacles to digitalization, the difficulty of reconciling head-office-led direction with on-the-ground response ranks near the top.

As long as quality records stay locked in paper and local spreadsheets, all head office can see is the aggregated result each site chose to send. What was dropped during aggregation, and on what basis something was judged nonconforming, cannot be verified from the head-office side. Even if drift exists, there is simply no path by which it could be detected.

Why Some Sites Go Unseen by Head Office Even With ISO Certification — The Square-Root Rule of Multi-Site Certification

This is the part that many Japanese-owned companies have not fully grasped.

Under multi-site certification, where several sites are covered by a single certificate, the International Accreditation Forum’s mandatory document IAF MD1:2023 sets out different formulas for the number of sites to be sampled depending on the type of audit. For the initial audit, the guide figure is the square root of the total number of sites, rounded up, expressed as y equals the square root of x. For the surveillance audits conducted every year thereafter, it is that square root multiplied by 0.6, again rounded up, expressed as y equals 0.6 times the square root of x. In every type of audit, the central function, meaning the head-office function, is always included in the audit scope. This is primary information set out in IAF’s official document.

Let us put that provision into concrete terms. Suppose a company covers nine sites under a single multi-site certificate. The sampling target for the initial audit is the square root of 9, which is 3 sites. Once certification is in place, the annual surveillance audits target 0.6 multiplied by the square root of 9, which is 1.8, rounded up to 2 sites, fewer still than at the initial audit. In either type of audit, no auditor has set foot in the sites that were not selected for the sample that year. The provision does require the sample to be rotated from audit to audit, but it is not a mechanism that guarantees when any particular site will next be verified on the ground.

For the four-site configuration used as the model later in this article, the sampling target at each annual surveillance audit is 0.6 multiplied by the square root of 4, which is 1.2, rounded up to 2 sites. Half of the sites are treated as part of a “certified” group without having gone through a certification audit that year.

To avoid a misunderstanding, let us state this plainly: this is not a defect in the certification scheme. Multi-site certification is a rational design that keeps the audit burden within realistic bounds on the premise that the central function is controlling each of the sites. That design holds up only if the central function has visibility over every site through internal audits.

The problem is that a good number of companies accept the certificate as an outcome without ever verifying whether that premise matches reality. The scheme thins out external auditing on the assumption that head office is looking at every site. Where head office is not in fact looking, the portion that was thinned out becomes a blank spot that nobody is looking at. The certificate will not tell you that blank spot exists.

There is one further distinction worth drawing here, which is the handling of customer audits. ISO certification and head-office internal audits differ from second-party audits conducted by customers in purpose, in frequency and in the records demanded. The practical question of how to get records requested during a customer audit into a state where they can be submitted quickly is dealt with in Digitizing Records for Audit Response: The Time It Takes to Submit. This article stays on the axis of the certification scheme and head-office control.

How Company-Wide Average Cost of Quality (COPQ) Hides Your Worst Site

If there is a blank spot on the scheme side, it is natural to think you can compensate with numbers. But there is a trap here too.

ASQ, the American Society for Quality, indicates that cost of quality can reach 15 to 20 percent of sales, and in some industries as much as 40 percent. Narrowing to manufacturing, the average is broadly in the region of 15 percent, with a range of 5 to 35 percent depending on the industry and product complexity. This is primary information published by the industry body itself.

Many companies compare this kind of benchmark against a company-wide figure. If company-wide COPQ lands somewhere around 15 percent, they judge themselves to be in line with the industry. What gets overlooked here is the nature of an average as a metric.

An Average Cancels Out the Drift

The following is a hypothetical numerical example, presented in order to show the mechanism by which drift between sites gets buried in an average. These are not the figures of any real company; please read them as a scenario built out from ASQ’s 15 percent benchmark.

Suppose group-wide sales are 3,000,000,000 THB, there are four overseas sites, and each site accounts for an equal 750,000,000 THB of sales. Suppose three sites incur COPQ at 15 percent, equivalent to the ASQ benchmark, and one site alone is at 22 percent. The gap is 7 percentage points.

In this scenario, 7% multiplied by 750,000,000 THB gives 52,500,000 THB, the amount arising in excess at that one site every year.

Yet viewed as a company-wide average, that gap barely stands out. A state in which just one of four sites is 7 percentage points higher, once smoothed into a company-wide average, amounts to only a slight uplift, and it sits comfortably inside the 5 to 35 percent range ASQ indicates for manufacturing. As long as the comparison is against the benchmark, this company’s cost of quality will be judged to be in line with the industry. You want to identify the deviating site by name, and the metric you are looking at has the property of canceling that deviation out. That is the essence of the problem.

It is worth emphasizing that the 7 percentage point gap is itself an assumption. We are not claiming that your company has a 7 point gap. The gap might be zero points, or it might run in the opposite direction. That is exactly where the core of this article’s argument sits. How much drift you actually have cannot be known unless you measure it site by site. And as long as you are looking at a company-wide average, you have not measured it.

Three Reasons It Cannot Be Broken Down by Site

There are practical reasons why COPQ is not broken down by site at many companies.

First, the components of COPQ span multiple departments. Scrap and rework land in manufacturing cost, complaint handling lands in selling expenses, and inspection labor lands in the quality department’s personnel cost. No mechanism bundles them by site.

Second, the definitions of the accounts differ from site to site. Even for the same item labeled rework cost, if the line on what to include is not aligned, comparing across sites will not produce a difference that means anything.

Third, some failure costs never make it into the records at all. If something noticed on the line and reworked on the spot is never registered as a formal nonconformity, then as far as the numbers are concerned it does not exist. How to design the granularity of that record is a within-a-single-site question, and it is dealt with in Quality Data Management System: A Three-Layer Design. This article treats the fact that the record layer itself differs from site to site as the problem.

What an Overseas Plant Quality Management System Actually Needs — Four Questions That Set Your Standardization Level

So what should you have in place? Before you start comparing product catalogs, here are four questions for confirming where you stand today. All of them are framed so that they can be answered with a yes or a no.

Question 1 — Can You Break COPQ Down by Site?

As described in the previous section, a company-wide average will not show you the drift. Can you aggregate scrap, rework, complaint handling and inspection labor by site, using the same definitions? It does not have to be rigorous cost accounting. Being able to line the sites up and compare them on the same definition matters more than precision. If the definitions are aligned, you can read the differences between sites even if the absolute values are somewhat rough.

Question 2 — Can You See Corrective Action (CAPA) Closure Rates by Site?

Can you track, by site, the number of cases and the number of days from a nonconformity occurring to the corrective action being completed? The number of CAPAs sitting open and stagnating is a good reflection of the real state of that site’s quality management. If there is a site with a count of zero, that is not excellence; you should suspect that nonconformities are simply not being registered there.

Question 3 — Can You Confirm That the Local Versions of Head-Office Standard Documents Are Current?

Are the latest editions of the quality manuals and inspection standards issued by head office actually the ones in use at each site? When you issue a revision, can head office confirm that the old edition has definitively been replaced at every site? Without a means of confirming that, head office has merely distributed documents while feeling that it has exercised control. Where translated versions exist, whether the timing of translation has fallen out of step with the timing of revisions to the original is the same problem.

Question 4 — Do Primary Records of Nonconformities Exist at the Same Granularity at Every Site?

This is about the primary record, not the final aggregated figure: when, on which line and for which production lot, what happened, and who judged it how. If that granularity differs from site to site, then even comparing the numbers that come up to head office will not tell you what the differences mean.

Of these four, whichever ones come back as a no are precisely the areas to take on first.

Draw the Line on Standardization at Three Levels

Having answered the four questions, decide what to unify and how far. In practice, aligning everything to the head-office standard is neither realistic nor rational. Drawing the line at the following three levels makes it easier to organize.

Note that remotely monitoring equipment operating data from head office is a separate investment decision from controlling quality records. For the cost and payback of monitoring infrastructure, see Overseas Plant IoT Deployment: The Cost and Payback of Remote Monitoring. What this article addresses is not whether the equipment is running, but whether quality records and quality judgments are aligned across sites.

Designing the Investment Around a Diagnosis First — When It Costs Less Than Investing Uniformly Across All Sites

From here we turn to cost. Everything that follows uses a hypothetical model company and is an assumed scenario, not the figures of any real company. The currency is THB, Thai baht.

As the scenario, assume the following company. A Japanese-owned manufacturer with four overseas sites: two in Thailand, one in Vietnam and one in Indonesia. Each site is an assembly or machining plant with 50 to 150 people. The quality assurance department at head office in Japan oversees them, but there is no dedicated overseas quality specialist. This is close to the configuration of many mid-sized Japanese-owned manufacturers.

Comparing Two Approaches in Numbers

There are broadly two ways to proceed.

One is to skip the diagnosis and make the full investment uniformly at all four sites: document standardization, deployment of a CAPA tracking system, and reinforced site visits, carried out at every site. In this scenario, the cost is taken to be 3,400,000 THB up front and 600,000 THB per year.

The other is to start with a diagnosis. First, across all four sites, run a simplified site-by-site COPQ measurement and a survey of CAPA closure rates over 90 days. In this scenario, the cost is taken to be 180,000 THB per site, or 720,000 THB for four sites. On that basis, the full investment is then made only at the sites where drift from the head-office standard has been confirmed. In this scenario, the full investment is taken to be 850,000 THB up front and 150,000 THB per year, per site.

What the diagnosis-first approach comes to depends on the number of sites where drift is found. Let us compare three cases.

In the case where drift is found at 2 of the 4 sites, the total up-front cost comes to 2,420,000 THB, which is 28.8% less than the 3,400,000 THB of the uniform investment. The annual cost is also 300,000 THB, half the 600,000 THB of the uniform investment. Even bearing the entire cost of the diagnosis, it still comes out cheaper.

In the case where drift is found at not a single site, the up-front cost ends at the 720,000 THB of the diagnosis alone, which is 78.8% less than the uniform investment. In this case, what the diagnosis bought was the conclusion that you do not need to invest. That is not wasted spending; it is spending to obtain the grounds for not having to spend 3,400,000 THB.

On the other hand, in the case where drift is found at all four sites, the up-front cost comes to 4,120,000 THB, which is 21.2% more than the uniform investment, because the diagnosis cost is simply added on top.

The Break-Even Sits at Around Three Sites

Expressing this structure as an equation, the value of n that satisfies 720,000 plus 850,000 multiplied by n equals 3,400,000 is approximately 3.15. In other words, if you expect that three or fewer sites will need the full investment, going in with the diagnosis first is advantageous; if you already know from the outset that all four sites are in scope, the uniform investment is cheaper.

What deserves attention here is that second condition itself. A state in which you already know from the outset that all four sites are in scope for the full investment is not normally possible without a diagnosis. Being confident that every site is in scope while not having diagnosed anything is, in reality, nothing more than a guess. The choice is between committing 3,400,000 THB on the basis of a guess, or paying 720,000 THB to convert the guess into fact before committing.

And there is one more difference that does not show up in the money. If you do the diagnosis first, you are left holding data on which site deviates from the head-office standard, on which metric, and by how much. That information serves not only the investment decision but also as the baseline against which you measure whether the investment produced an effect. With a uniform investment, because the pre-investment state was never measured, verifying the effect becomes a matter of impressions.

The Minimum the Diagnosis Phase Should Cover

What should be carried out during the 90-day diagnosis phase corresponds almost exactly to the four questions in the previous section: a simplified site-by-site COPQ measurement, aggregation of CAPA closure rates and stagnating case counts, a reconciliation of head-office standard documents against the documents actually in local use, and sampling of primary nonconformity records. There is no need to bring in a new system. Simply taking the existing records, aligning them to the same definitions across sites and lining them up will reveal a good deal about whether drift is present.

Where Thai and Vietnamese Regulatory Trends Bear on Quality Management System Design

An external factor that makes it harder to keep postponing the alignment of standards across sites is the direction of regulation on the ASEAN side.

Vietnam’s Amended Law on Product and Goods Quality

The amendment to Vietnam’s Law on Product and Goods Quality is reported to have been passed by the National Assembly on 18 June 2025, with 408 votes in favor out of 420, and to have come into force on 1 January 2026. It is described as the first major overhaul in 17 years, classifying products into three risk tiers of low, medium and high, and making conformity certification mandatory for high-risk goods along with the assignment of a digital product passport, meaning traceability by means such as a barcode. The transition period is reported to run for up to three years. This is secondary information from a firm that supports companies entering the local market, but it comes with a specific passage date and vote result. For the detailed scope of application, you will need to check the responsible authority’s published material against your own products individually.

The implication for design is clear. Anything equivalent to a digital product passport requires that an individual product or a production lot be linked to its manufacturing and inspection records. If only the Vietnamese site holds records at that granularity while the other sites carry on as before, the granularity of records will diverge structurally across sites. Viewing them side by side as quality data for the same group becomes difficult.

Seen this way, the regulatory response is better designed not as remedial work for the Vietnamese site, but as an opportunity to align the granularity of records across the entire group.

The ISO 9001 Revision as a Shared Moment

As noted earlier, the revised edition of ISO 9001 is scheduled for publication in September 2026, with a three-year transition period running to September 2029. Because every site needs to respond to the revision of the same standard, there could not be a better moment to align document structures that have until now been developed separately at each site.

Conversely, if you let each site handle it individually here, the drift will become even more entrenched through the revision. Responding to a revision is work each site can manage on its own, but the result of letting them do it on their own is document structures that branch into as many versions as there are sites.

For Thai Sites, Think in Structural Terms Rather Than Regulatory Ones

We have not been able to confirm primary sources on how Thailand’s investment promotion regime or the quality audit requirements of the industrial authorities relate to the argument of this article. Accordingly, this article makes no claims that rest on assumed regulatory requirements on the Thai side. For Thai sites as well, our view is that site-by-site measurement is needed for the structural reasons set out in the preceding sections, namely the square-root rule and the nature of averages, rather than for regulatory reasons.

Note that in some industries customer requirements arrive ahead of regulation. For automotive components, the traceability granularity required by IATF 16949 will take effect first. Industry-specific traceability design is covered in Automotive Parts Traceability: IATF 16949, so this article stays with multi-site governance without narrowing to a single industry.

Common Failures in Rolling Out an Overseas Site Quality Management System

Finally, here are the failure patterns most often seen in practice. Each is the flip side of a point made in this article.

Failure 1 — Making Maintenance of the Certificate the Goal

Certification is something to maintain, but it is not in itself the objective. If the absence of findings in a certification audit is treated as proof that there is nothing wrong with the quality management structure, the year will pass without anyone checking the state of the sites that were thinned out under the square-root rule. Certification confirms a floor; it does not prove a ceiling.

Failure 2 — Propping Up Quality With Expatriates and Visiting Staff

As noted in the report cited earlier, excessive dependence on Japanese expatriates and visiting staff is said to invite a vicious circle of rising cost and confusion on the floor. Support that depends on individuals works in the short term, but the quality level reverts once that person is reassigned. As long as quality rests on people rather than on a mechanism, the drift between sites will be determined by where people happen to be posted.

Failure 3 — The Head-Office Standard Gets Replaced by a Different Local Format

It is not unusual for the quality manuals and inspection standards issued by head office to be rebuilt locally to make them easier to use, until they have quietly become a different format. There is no bad intent; in most cases it is a reasonable adaptation by the people doing the work. The problem is that head office is unaware of the substitution. Head office reads the numbers on the assumption that operations follow the format it distributed, while the local site is making its judgments on a different one.

This kind of substitution can be detected to a considerable extent simply by keeping a document control register at head office and periodically collecting samples of the forms actually in use locally. There is still plenty you can do before any system investment.

Failure 4 — Putting the Same Thing Into Every Site Uniformly

As the numbers in the earlier section showed, making the full investment at sites where there is no drift tends to work against you financially. On top of that, putting a new mechanism into a site that does not feel the need for it raises the probability that the operation becomes an empty formality. The reported case in which a rollout delivered by instruction alone ended in check sheets being filled in without inspection is precisely this pattern.

Failure 5 — Collecting Metrics Without Connecting Them to Corrective Action

A dashboard is built, and site-by-site metrics are now lined up on it. Some organizations stop there, satisfied. A metric only functions once it is settled who does what when a threshold is crossed. When the CAPA closure rate falls, who at head office intervenes, and when? A metric without that decided ends up doing nothing but adding aggregation cost.

Failure 6 — Deploying Something That Cannot Be Operated in the Local Language

A recording system is deployed with an interface only in Japanese or English, and local operators end up recording in a separate notebook first and entering the data afterwards, producing a duplicated workflow. In that state, the granularity of the primary record is not assured. Being able to record in the local language, within the flow of the work, is what determines the quality of the record.

Summary — Start by Listing What You Are Not Measuring

To bring the key points together.

• Holding a certification such as ISO 9001 is not a guarantee that multiple overseas sites are being operated to the same quality standard.
• Under multi-site certification, the sampling target for the initial audit is the square root of the total number of sites, rounded up, and for the surveillance audits conducted every year thereafter it is that figure multiplied by 0.6, rounded up. In either type of audit, the sites outside the sample are not verified on the ground that year. The scheme thins out external auditing on the assumption that head office is controlling every site.
• A company-wide average cost of quality buries drift between sites inside the average. As long as you are comparing a company-wide average against the ASQ benchmark, you cannot determine whether drift is present.
• Therefore, ahead of selecting a quality management system, you should get to a state where you can measure four things by site: COPQ, CAPA closure rate, currency of documents, and the granularity of primary records.
• In the hypothetical scenario, an approach that diagnoses first and invests only in the sites where drift was found requires less up-front cost than a uniform investment across all sites, provided that three or fewer sites need the full investment. It is 28.8% less in the case of 2 of 4 sites, and 78.8% less if there is no drift at all. Conversely, if all four sites are in scope, it is 21.2% more.
• The amendment to Vietnam’s Law on Product and Goods Quality is reported to have come into force on 1 January 2026, and the revised edition of ISO 9001 is scheduled for publication in September 2026. Both are the kind of change that entrenches drift if you let each site respond to it separately.

Put differently, the first thing to do is not to deploy something, but to list what your company is currently not measuring. In most cases, the list of things you are not measuring is itself the order of priority for your investment.

It is perfectly fine to be at the stage of not even knowing whether you have any means of measuring the drift between your sites. TOMAS TECH is based in Thailand and has worked on making quality records and production data visible at the overseas sites of Japanese-owned manufacturers. You are welcome to start with a conversation about whether your current records are in a state where they can be compared across sites, via our contact page.

Frequently Asked Questions (FAQ)

Where should we start with an overseas plant quality management system?

The sensible starting point is measuring where you stand today, not selecting a product. Specifically, four points: whether you can break COPQ down by site using the same definitions, whether you can track CAPA closure rates and stagnating case counts by site, whether you can confirm that the latest editions of head-office standard documents are in use at each site, and whether primary records of nonconformities are at the same granularity across sites. Whichever of these four comes back as a no is the area to take on first. Simply realigning existing records to the same definitions across sites will tell you a great deal before you bring in any new system.

If we hold ISO 9001 certification, can we assume our overseas sites are on the same quality standard?

Better not to. Under multi-site certification, the provisions of IAF MD1:2023 mean the sampling target for the initial audit is calculated as the square root of the total number of sites, rounded up, and for the surveillance audits conducted every year thereafter it is that figure multiplied by 0.6, rounded up. With nine sites, the initial audit covers 3 sites, and each subsequent annual surveillance audit covers 0.6 multiplied by the square root of 9, which is 1.8, rounded up to 2 sites; no auditor visits the remaining sites that year. This is not a defect in the scheme; it is a design premised on the central function managing all sites through internal audits. Whether that premise matches reality is something nobody will check unless head office checks it itself.

How does overseas plant traceability differ from a quality management system?

Traceability refers to being in a state where you can retrospectively trace which materials, processes and operators a given product came from, while a quality management system refers to the whole mechanism that uses those records to judge conformity against a standard and to deal with deviations. Traceability is one component of a quality management system. In a multi-site context, if the granularity of traceability differs from site to site, you end up with a situation where records can be traced but cannot be compared across sites. Designing the record layer within a single site is covered in the quality data management system article, and industry-specific granularity design in the automotive parts traceability article.

How much does it cost to deploy a quality management system at overseas sites?

It varies widely with the size of the sites, the state of existing records and the number of sites in scope, so no generalized figure can be given. This article set out an assumed scenario using a hypothetical model company, with a four-site configuration in which the diagnosis phase is 180,000 THB per site, and the full investment at a site where drift is confirmed is 850,000 THB up front and 150,000 THB per year, per site. In that scenario, a uniform investment across four sites is 3,400,000 THB up front, whereas if drift is found at 2 of the 4 sites the total including the diagnosis cost is 2,420,000 THB, which is 28.8% less. These are assumed values only, not actual results from any real company.

How do we find the drift in quality standards between sites?

Even without deploying a new system, simply taking existing records, aligning them to the same definitions across sites and lining them up will reveal a large part of it. Specifically, four things: a simplified site-by-site COPQ measurement, aggregation of CAPA closure rates and stagnating case counts, a reconciliation of the standard documents issued by head office against the forms actually in use locally, and sampling of primary nonconformity records. The hypothetical scenario in this article assumes this diagnosis is carried out over 90 days across four sites. In particular, where a site has an extremely low count of registered nonconformities, it is worth suspecting that the registration practice is different rather than that quality is good.

References

• IAF MD1:2023 – Mandatory Document for the Audit and Certification of a Management System Operated by a Multi-Site Organization – primary source
• Multi-site ISO certification – how it works and common pitfalls – supplementary commentary on IAF MD1
• ISO/TC 176/SC 2 – Update – Revision of ISO 9001 – primary source
• ASQ – Cost of Quality – primary source
• JBpress – report on quality countermeasures becoming an empty formality at Japanese-owned manufacturing sites in Thailand, in Japanese – secondary source
• tebiki Genba Note – quality control at overseas plants, in Japanese – secondary source
• CAST Vietnam – amendment to Vietnam’s Law on Product and Goods Quality, in force 1 January 2026, in Japanese – secondary source
• JETRO, Japan External Trade Organization – FY2025 Survey on Business Conditions of Japanese Companies Operating Overseas, Asia and Oceania edition – primary source