Automotive Parts Traceability 2026 — IATF 16949 and Customer Audits

When a customer says “show us when this lot number was made, on which machine, and by whom,” how many minutes does it take you to answer? What is really being tested in automotive parts traceability is not whether you own a system. It is your power of proof: when something goes wrong, how narrow a range you can isolate, how quickly, and on the strength of what evidence. This article sets out a way of deciding the granularity of tracking and the span of records you retain, working backwards from the requirements of IATF 16949 and from the practical realities of customer audits — audits conducted on you by your OEM or Tier 1 customer — including the conditions specific to a plant in Thailand.

What this article covers, and what it does not

Material on traceability tends to split into two kinds: general arguments about why it is necessary, and feature lists for equipment and software. What a real automotive parts plant has to settle first, though, is neither an understanding of necessity nor a choice of features. It is a design boundary: how finely you will be able to trace (granularity), and how far your records will reach (scope). Collect quotations before that is settled and every supplier will scope on different assumptions, which makes both the price and the schedule impossible to compare from the outset.

Here is the scope of this article.

Note on terminology: a route tag here means what a Japanese plant calls a *genba slip* (現品票) — a kanban-style paper tag that travels with the physical parts and carries their identity. CSR means customer-specific requirements, the individual quality requirements each customer imposes on top of the standard. 4M means Man, Machine, Material and Method — the four categories of change a Japanese-style quality system tracks.

Cost breakdowns and how to structure a quotation are organised by layer in The Cost of Building a Traceability System, which is worth reading alongside this if you are at the budgeting stage. Food industry design built around temperature control and HACCP has a substantially different requirement structure, and is handled separately in Traceability in Food Factories. This article stays with automotive parts and digs into one thing only: designing the burden of proof.

Automotive parts traceability is the design of a burden of proof

The purpose is not to keep records

Once a traceability project starts, the conversation often opens with “let’s digitise the records first.” But records are a means, not an end. Pushed to its conclusion, the purpose of traceability in automotive parts comes down to a single sentence.

When a defect is found, to identify the range of affected product within a limited time and with evidence — and to demonstrate that everything outside that range is unaffected.

The second half is what matters. Being able to say “beyond this point, nothing is involved” is worth as much as being able to identify the suspect range. If you cannot narrow the range, everything shipped during the suspect period becomes the subject of action. That translates directly into the size of the loss, in the form of sorting, returns, arranging replacements, and the impact on your customer’s line.

Traceability, then, is not a mechanism for making quality better. It is a mechanism that determines the size of the loss when a quality problem occurs. For a defect that has already reached the market, the scope of the recall response is the cost. Whether or not that understanding is shared internally changes the quality of the investment discussion. The question stops being “how much do we spend to reduce defects” and becomes “how narrow a containment capability are we buying for the day a defect happens.”

What automotive parts customers actually ask

Lining up the questions that come up in customer meetings and internal quality reviews makes the shape of the required data visible.

The last row is the essential one. The real strength of traceability is set by its weakest link. You may be collecting equipment data at high resolution, but if goods receipt records for materials are still on paper, a material-borne defect cannot be narrowed down. When setting investment priorities, one line of thinking is that looking for the weakest link tends to produce more effect than strengthening what is already strongest.

The price of not being able to narrow the range

What happens when you cannot narrow the range, set out by stage.

The caution here is that lower down the table is not simply “worse.” A mechanism that tracks every individual part carries a corresponding load on the floor and a corresponding cost. Design means deciding how much containment capability is needed relative to the nature of the part and the customer’s requirements, which leads directly into the question of granularity in the next chapter.

Deciding tracking granularity by working backwards from customer requirements

Three levels: lot, sub-lot and serial

The choice of granularity is the single biggest fork in traceability design. Leave it vague and move on to equipment selection, and you will be rebuilding something later without fail. Here are the three levels side by side.

“Serialise everything” is not the right answer. Serial control gives the strongest containment capability, but it carries ongoing cost across numbering, marking, reading and data retention. Apply it uniformly down to low-value mass-produced parts and the recording load eats into working time on the floor, so entry becomes a formality. Records that have become a formality are the state most likely to attract a finding in an audit.

Conversely, if you hold a safety-related part — or a part where the customer requires delivery records at serial level — at lot granularity because of cost, the affected range in a defect event widens by orders of magnitude. The realistic position is that granularity is decided part by part, not standardised across the plant.

The axes for deciding granularity

When deciding granularity part by part, organising the discussion around the following axes makes internal agreement easier to reach.

Fill this table in for each part family and most plants arrive at the same conclusion: serial control is needed for a subset, and sub-lot or lot control is sufficient for the rest. What matters is keeping the reasoning on record as a document. Audits ask less about the granularity itself than about the explanation of why you chose it.

Sub-lots: the practical middle ground

Sub-lots are the level most often overlooked. Between lot and serial sits an intermediate step that is extremely effective in practice.

A sub-lot means dividing a single lot further at the points where conditions that could affect quality change over. Typical dividing points are as follows.

That last item — the previous confirmed good point — matters most. When a defect is found, the lower bound of how far back you go is the last moment at which product was confirmed good. The longer the interval between those checkpoints, the wider the suspect range. Which means that before investing in finer granularity, simply increasing the frequency of confirmed good checks can improve containment capability. That is a comparatively low-cost place to start, and worth bearing in mind when setting priorities.

Traceback and trace forward need different data

Two directions built on two different mechanisms

Traceability is one word, but it contains two kinds of tracking running in opposite directions. They need different data, and it is not unusual to find a plant that has only one of them working.

In most plants, traceback is relatively well established and trace forward is weak. The reason is straightforward: the shop floor cares about in-process records, whereas linking shipping results to lots spans logistics and sales administration. Yet the very first thing a customer asks for in an initial report is trace forward information — how far the product has already shipped. When that is weak, you end up in the most uncomfortable position of all: you broadly know the cause, but you cannot state the range.

A quick self-diagnosis

A short check for working out which side you are weak on. Read each line and ask whether you could answer “yes” immediately.

Item 10 is the one people forget. When a part that has been reworked or re-inspected returns to good stock having lost its original lot information, that is where the chain breaks. In real defect response, the handling of these “returned” items is often the hole in range identification. How rework process records are kept is something to settle at the design stage without fail.

Which to tackle first

Fixing both at once is ideal, but if limited resources force a sequence, the following approach holds.

• If the customer requires speed in the initial report, get trace forward in place first, because being able to state the range takes priority
• If defects recur at a specific process and the cause has not been pinned down, get the relevant process on the traceback side in place first
• If neither applies, look for the weakest link described in the previous chapter and start there

One point on timing: the time allowed for an initial report is not fixed uniformly by the standard — it varies by customer-specific requirements (CSR). In practice the range runs from cases where a first report is required within a few hours to cases where an initial report is required within 24 hours. Check the CSR documents of your own major customers individually. Framing it as “the industry standard is X hours” makes for a weak basis both for internal targets and for audit responses.

Translating the identification and traceability requirements of IATF 16949 into your own language

Do not read the clause — translate it into implementation items

The identification and traceability requirements of IATF 16949 (clause 8.5.2, together with the automotive-specific supplemental requirement 8.5.2.1) feel abstract when read as a standards document. What works in practical audit preparation is not deploying the clause text internally, but building a list of implementation items that translates the intent of the requirement into the language of your own processes. No clause text is quoted here; instead, the points that are widely understood as the intent of the requirement are organised from an implementation perspective.

The right-hand column is the point of this table. What an audit looks at is not an explanation of the mechanism but what you can produce as evidence. “It is managed in the system” is not sufficient on its own; the question is whether you are in a state where you can retrieve it and show it there and then.

Three items that are commonly under-implemented

Of the items above, three tend to be implemented weakly.

The first is maintaining identification between processes. A part may carry a number inside the process and lose its identity the moment it is set down flat in a staging area. In particular, situations where several lots end up mixed in one container, or where a route tag is attached to the container and stops corresponding to the contents, are common on the floor. Whether identity is held at container level or on the product itself has to be decided to suit the process layout.

The second is identifying held and re-inspection states. A binary of accepted and rejected does not keep the floor running. In reality, intermediate states arise: awaiting judgement, under re-inspection, concession request pending. If these states are identified neither in the system nor on the physical part, room remains for product to escape as good before judgement is made. Listing the state types and giving each one a rule for labelling and for where it is placed is what implementation actually consists of.

The third is the accessibility of records. Being stored and being retrievable when needed are different problems. Records for a lot from five years ago sitting in a cardboard box in a warehouse, or in an Excel file on the PC of someone who has left, are hard to describe as “retained.” Alongside defining the retention period, you need to settle who can retrieve it, by what means, and within roughly how long.

How to decide the retention period

Retention periods sit where several requirements overlap. Rather than setting one figure for everything, one approach is to line up the following elements and take the longest.

The last item is one to consider without fail when going digital. The question is whether you have any guarantee of being able to read that data format in ten years. Data retrievable only in a proprietary format may effectively be lost the moment you replace the system. Confirming at contract stage that bulk export in a general-purpose format is possible reduces that concern.

Connecting 4M change management to traceability

If changes are not linked to records, isolation depends on memory

4M change management (Man, Machine, Material, Method) is operated as part of quality management in most plants. Seen from the traceability perspective, though, what is decisive is not the existence of the 4M mechanism itself but whether 4M change records are linked to lots.

Root cause analysis almost always produces the question “did you change anything around then?” If the 4M change notice is managed in a separate file, someone has to search their memory and recall that “come to think of it, we switched material suppliers last month.” This is one of the points a customer audit probes hardest, because a state in which awareness of changes depends on an individual’s memory is judged to lack repeatability.

Conversely, if the design raises a flag on the lot number side when a 4M change occurs, then the moment you search for the defective lot, information such as “this lot is the first production run after the material lot changeover” becomes visible automatically. This is not a particularly difficult thing to build; simply adding a field to the change notice for “the lot from which this applies” produces a large improvement on its own.

What to record for each of the four Ms

Here is what should be recorded as a change for each element of the 4M, and how it links to the lot.

The “revision of the instruction” under Method is especially easy to miss. When a work instruction is revised, if there is no record of which lot the new revision started from, you cannot separate whether a defect originates from the old revision or the new one. A mechanism linking the revision history of instructions to production lots falls into the higher end of cost-effectiveness as a target for systematisation.

What a change notice should carry

Check whether your 4M change record — paper or electronic — carries the following items.

The item in bold — the first applicable lot number — is missing from most change notices. Adding it alone changes the amount of information available during traceback substantially. As an improvement that can be started without any system investment, it is worth raising in priority.

Note also that the range of changes requiring prior notification to the customer may be defined by CSR. What counts as “a change requiring customer approval” differs by customer, so it is safest to confirm that your own decision criteria are consistent with the CSR.

What a customer audit tests is repeatability, not the mechanism

Can you answer “produce it now”?

In a customer audit or an internal IATF audit, what actually happens on traceability is not limited to checking documents. In most cases, the auditor names one lot number on the spot and asks you to produce the required information in front of them.

What is being tested at that moment is not whether you have a system, but the following.

• Whether someone other than the usual owner can reach the same result using the defined procedure
• Whether the required information comes together within the expected time
• Whether the information produced is consistent with the physical part and with other records
• Whether you are aware of where information is missing

The fourth is more important than it looks. Nothing has to be perfect; what is valued is being aware of your weak points and holding a plan to address them. Explaining that “there are no issues” without knowing your weaknesses, and then failing to produce during the demonstration, makes a distinctly poor impression.

Build mock trace exercises into your operations

The surest way to secure repeatability is to practise in peacetime. The activity is called a mock trace exercise, or traceability drill: with no actual defect in play, you take a randomly chosen lot and produce the information for it.

One example of how to design it.

The value of this exercise lies in making weaknesses visible as numbers. Once measurements such as “identifying the material lot took 40 minutes” and “the delivery destination summary came out in 5 minutes” are on record, the discussion about where to invest becomes concrete. A vague wish to “strengthen traceability” turns into a specific requirement to “automate the linkage at material receipt.”

The exercise records themselves also become audit material. A history of running exercises regularly and acting on the issues found is evidence that the mechanism is genuinely in operation.

Questions that come up in audits, and what to have ready

We recommend preparing that last line. Plants with no weak points barely exist in reality. Showing that you have identified them yourself and are addressing them in priority order is far better received than answering “no issues” and then coming apart during the demonstration.

Designing record carriers and data capture

Choose on three axes

Route tags, barcodes, QR codes, RFID, direct part marking (stamped or laser-marked) — there are several options for record carriers, but the decision is not about which is superior; it is about which fits the process environment and the way you will read it. Organising around the following three axes makes the judgement easier.

1. Process environment: is it exposed to oil, cutting fluid, heat, washing, grinding dust or paint
2. Reading frequency and method: at how many process steps is it read, handheld or fixed, is non-contact required
3. Acceptable cost per piece: relative to the part’s unit price, how much can you spend on identification

Laying each carrier out against those three axes gives the following.

RFID is not always at the top. RFID has the clear advantage of non-contact bulk reading, but its reading characteristics change near metal parts and liquids, and tag unit cost is higher than the alternatives. It also raises no issue when fitted to returnable containers or fixtures, but if it is attached to product that ships out, you need an operating design for recovering and reusing the tags.

Direct part marking, at the other end, is the most durable and has a strength nothing else offers: the part remains identifiable after it has reached the market. It does, however, require an assessment of whether the marking affects the strength or function of the part, and the investment in marking equipment is not small. The usual judgement is to adopt it only for safety-related parts and parts where serial control is mandatory.

Assume you will mix carriers

In practice, rather than covering every process with a single carrier, the realistic design is to change carrier by process and connect them on the data side.

What matters in this design is that the linkage does not break where the carrier changes. The point at which information passes from RFID on a returnable container to a QR code on the product is the critical junction in implementation. Leave that to human transcription and it becomes your weakest link.

Records become a formality unless you reduce the entry burden

As important as carrier selection is the data entry effort on the floor. Traceability records are continuous work that recurs with every production run. Even at a few tens of seconds per instance, a few hundred instances a day add up to time you cannot ignore.

When the recording burden is high, the following happens. People batch up the entries later on a busy day; they copy the values from the previous lot; values that differ from reality get entered. All of these damage the reliability of the record. Reducing the entry burden is not a question of convenience but a question of record accuracy.

Ways to reduce the burden include the following.

• Reduce the items a person types and capture directly from equipment and measuring instruments
• Turn entries into selections from a list rather than free text
• Require no record under normal conditions and entry only for exceptions (only where the normal condition can be judged automatically)
• Place the reading point on the natural work path so that no re-gripping of the part or extra walking is required
• Match the screen language to the operator’s language

Capturing data from existing equipment

Taking data directly from equipment reduces the entry burden considerably, but in real plants the majority of machines are older models with no communication capability. There is no need to conclude from this that “traceability is impossible unless we replace the equipment.” Signal tapping, retrofitted sensors, counters, and image-based judgement are among several ways of capturing data from existing machines. The options and the reasoning behind them are set out in IoT Retrofits for Legacy Equipment, which is worth reading if data capture from equipment is your bottleneck.

One caution specific to traceability. Are the timestamps on data captured from equipment aligned with your other records? If the clock on the machine has drifted, production lots and machine condition data cannot be matched when you put them side by side. Where you are gathering time-stamped data from multiple machines and systems, the method of clock synchronisation has to be settled at the outset. It is a difficult thing to fix afterwards.

Issues specific to automotive parts plants in Thailand

Imported material procurement is where linkage breaks

Automotive parts plants in Thailand commonly procure a substantial share of their materials through imports. A report published by JETRO in November 2025 notes that, in the Thai automotive industry, parts imports from China have increased since 2013, with the trade deficit widening particularly since 2021.

From a traceability perspective, a high share of imported procurement shows up as difficulty establishing linkage at the point of receipt.

Goods receipt is the most fragile point in the whole of traceability. If it is broken there, no amount of precise in-process recording will let you trace a material-borne defect. When setting investment priorities, receiving is an area that often belongs near the top.

It is possible to ask a supplier to mark lot numbers, but depending on their production setup it may be difficult for them to comply. A plan is more stable if the design absorbs the problem on your own side rather than assuming the other party will change.

Maintaining records amid multiple languages and staff turnover

In Thai plants it is normal for Japanese, Thai and English to be running at the same time. On top of that, many workplaces see more frequent operator turnover than is typical in Japan. Keeping record quality high in that environment requires a design that does not depend on understanding the procedure.

“Can a newcomer learn it in a few hours?” is a practical and effective criterion. If the recording operations for traceability are complex, record quality drops with every turnover. Reducing the number of steps in the operation translates directly into the reliability of the record.

BOI, local content requirements and production lot records

Plants holding a BOI (Thailand Board of Investment) import duty exemption on raw materials are required to report their duty-exempt raw material allowances against actual usage. When the ledger for that report and the production lot records are managed separately, the reconciliation between them ends up as manual work in Excel.

Local content is also a policy issue in the electrified vehicle field. A report published by JETRO in February 2025 describes Thailand’s EV policy: within a free zone, meeting a local content ratio of 40% or more of the ex-works (EXW) price allows finished products to be shipped into Thailand duty-free; and a special measure, limited to the period from 8 October 2022 to 31 December 2025, allowed the CIF price of imported battery cells to be counted towards the local content ratio, up to a cap of 15% of the BEV’s EXW price. The deadline for that special measure has already passed as of the time of writing, and whether it has been extended or succeeded by another scheme should be confirmed with the relevant authority. The same report also describes the HEV support measures approved in July 2024 — excise duty of 6% for CO2 emissions under 100 g/km and 9% for 101–120 g/km, with an implementation period stated as 2026 to 2032 — and the MHEV support measures approved in December 2024, under which the excise rate is reduced to 10–12%, again with an implementation period stated as 2026 to 2032.

These are primarily schemes on the vehicle manufacturer side, but for a parts supplier they mean that requests to submit information on procurement ratios and origin may become more frequent. If production lot records and the records of where materials were procured and where they originate sit in separate ledgers, every such enquiry generates a manual aggregation exercise.

That said, whether these schemes apply to you, and exactly which records would be required, depends on conditions including industry, investment category and the timing of the application. Nothing in this article can be used to determine your own eligibility, so confirm directly with the BOI or the relevant authority. As a matter of traceability design, holding receiving lot and procurement source information inside the same mechanism as production lots, regardless of whether you hold incentives, keeps the additional effort small if requirements grow later.

The EV shift brings new parts and new suppliers

The composition of the electrified vehicle market in Thailand is moving. According to the JETRO report published in November 2025, HEV sales in 2024 were around 130,000 units (up 49.9% year on year), BEVs around 70,000 units and PHEVs around 9,000 units, while new BEV registrations in the first half of 2025 were up 52.0% year on year. At the same time, total vehicle sales in 2024 are given as around 570,000 units (down 26.2% year on year), so the picture is one of the electrified share shifting within a shrinking overall market.

For a parts supplier’s day-to-day work, that change shows up in the following forms.

The last item becomes a design issue. Where parts requiring serial control and parts for which lot control is sufficient run mixed down the same line, the recording mechanism has to be designed so it can switch per part. Forcing everything to the strictest granularity makes the effort for mass-produced parts excessive.

The period immediately after launch is also one where process conditions are unstable and the probability of defects is relatively high. This is precisely when traceability earns its keep — yet in practice there is a structural tendency for the recording mechanism to be pushed back while everyone is busy with the launch itself. Explicitly including “are traceability records being captured as designed” among the criteria for transition to mass production is the practical countermeasure.

How to think about cost, and where to start

Think of cost as buying containment capability

Traceability investment tends to be discussed in terms of the price of hardware and software, which makes for an awkward internal approval case. The argument that passes more easily is “how narrow a containment capability are we buying for the day a defect happens.”

Suppose your suspect range today is one month’s production, and you become able to narrow it to one day. The assumed cost of sorting, returns and arranging replacements changes accordingly. That difference is the part you can present as the effect of the investment. If it escalates as far as a field recall, the difference in the number of vehicles in scope is directly the difference in cost. The actual figures vary greatly with unit price, production volume and the customer’s disposition policy, so no single number can be given — but collating what your own past defect responses cost makes this discussion concrete.

How to break the cost down (hardware, software, installation, system integration, operation and maintenance) and structure a quotation is organised by layer in The Cost of Building a Traceability System, which is the place to go when you need the money framework. This article stays with the question of where to spend a limited budget first.

How to set priorities

When deciding the order of work, laying items out on two axes makes the judgement easier: contribution to containment capability, and ease of starting.

The top three or four items can be started with almost no system investment. Do these first, confirm the remaining weaknesses as numbers through a mock trace exercise, and only then decide the targets for capital-intensive measures — that sequence is one way of reducing wasted investment.

An example of a staged approach

We recommend not skipping Stage 0. Move into equipment selection with granularity unsettled, and when it later emerges that “this part needed serial control,” you will be revisiting the equipment configuration. Deciding granularity is document work and costs almost nothing, but it is the premise for every judgement that follows.

Budget for internal effort

One final point. Building traceability does not reduce internal effort to zero even when it is outsourced. Judging granularity, classifying part families, reading the CSR, putting 4M change operations in order, negotiating receiving rules with suppliers — all of these are areas only someone who knows your own business can decide.

Build a schedule without allowing for this effort and the plan will stall at the granularity decision and the receiving rule negotiation. Put the other way round, getting those two moving internally in advance makes the start-up much faster once you do bring in outside help.

Frequently asked questions

How does 4M change management differ from traceability?

The purposes differ. 4M change management is a mechanism for verifying validity in advance and controlling the impact when a change occurs in people, equipment, materials or methods. Traceability, by contrast, is a mechanism for recording the linkage between product and production conditions so that it can be tracked afterwards. The two are nonetheless closely connected. If 4M change records are not linked to production lots, root cause analysis falls back on an individual’s memory of “what was changed around then,” and the isolation process loses repeatability. The practical point is to add a “first applicable lot number” field to the 4M change notice so that change history can be looked up in reverse from the lot. This is an improvement that can be started with almost no system investment.

Should we choose RFID or QR codes?

Neither is superior as a general proposition; the decision comes down to three things — the process environment, the reading method, and the acceptable cost per piece. RFID allows non-contact bulk reading and, with sealed tags, stands up well to oil and soiling, but its reading characteristics change near metal and liquid, and tag unit cost is higher. It suits returnable containers and fixtures, but if it is attached to product that ships out, you need an operating design for recovering and reusing the tags. QR codes have low label cost, can carry data in a small area, and remain readable with partial damage, but they require an application step and measures against peeling and soiling. In practice, rather than covering every process with a single carrier, the realistic design is to change carrier by process and connect them on the data side. In that case, how you preserve the linkage where the carrier changes becomes the critical design point.

Should every part be under serial control for automotive parts traceability?

There is no need for that. Serial control gives the strongest containment capability, but it carries ongoing cost across numbering, marking, reading and data retention. Apply it uniformly down to low-value mass-produced parts and the recording burden on the floor becomes excessive, with a risk that entry degenerates into a formality. Records that have become a formality are the state most likely to attract a finding in an audit. As decision axes, line up the following per part family: customer requirements (CSR), safety and regulatory involvement, unit price, production volume, sources of quality variation, whether combinations with mating parts must be recorded, and whether identification in the field is necessary. In most plants the conclusion is that serial control is needed for a subset of parts, and sub-lot or lot level is sufficient for the rest. What matters is keeping the reasoning behind that decision on record as a document.

What specifically do customer audits look at?

Alongside checking documents, a demonstration is frequently requested. The auditor names one lot number on the spot and asks you to produce the material lots consumed, the equipment, the operator, the inspection results, whether there were 4M changes, and the delivery destinations and quantities. What is being tested is not the existence of a system but repeatability: whether someone other than the usual owner can reach the same result using the defined procedure, whether the information comes together within the expected time, whether it is consistent with the physical part and other records, and whether you are aware of where information is missing. That last item matters particularly, because plants with no weak points barely exist in reality. Showing that you have identified your weaknesses and are addressing them in priority order is better received than explaining that “there are no issues” and then coming apart during the demonstration. We recommend running mock trace exercises in peacetime and keeping the records.

Within how many hours do we have to produce information when a defect occurs?

There is no uniform standard. This time may be defined by customer-specific requirements (CSR) and differs by customer — some customers set no such requirement. In practice the range runs from cases where a first report is required within a few hours to cases where an initial report is required within 24 hours. Since the standard does not set a uniform figure, it is safer to avoid building internal targets on the premise that “the industry standard is X hours.” The practical approach is to check the CSR of your major customers and set your target time against the strictest requirement. On top of that, measuring actual elapsed time in a mock trace exercise and deciding improvement items in the order that closes the gap to the target makes investment priorities concrete.

Should we tackle traceback or trace forward first?

It depends on which side you are weak on. In most plants, attention goes to in-process records, so traceback is relatively well established and trace forward — which requires linking shipping results to lots — is weak. Yet the very first thing a customer asks for in an initial report is how far product has already shipped. When that is weak, you end up in the uncomfortable position of broadly knowing the cause but being unable to state the range. As a rule of thumb: if customer requirements press hard on the speed of the initial report, take trace forward first; if defects recur at a specific process and the cause has not been pinned down, take the relevant process on the traceback side first. If neither applies, the reliable method is to walk the linkage from product back to material one step at a time and find the weakest point.

What should we watch out for particularly at a plant in Thailand?

Several things. The largest is linkage at the point of receipt, given the volume of imported material procurement. Situations arise where the supplier’s route tag carries no production lot marking, where mixed languages make it unreadable, or where the lot unit does not match your own management unit. You can ask the supplier to mark lot numbers, but compliance may be difficult for them, so a design in which you assign your own receiving lot number and relabel at receipt is more stable. Next is designing to hold record quality in a multilingual environment with operator turnover: screen and instruction language, reducing free-text entry, and keeping the number of operations low enough for a newcomer to learn quickly all matter. In addition, the ledger for reporting actual usage against BOI duty-exempt raw material allowances easily ends up separate from production lot records. Whether these schemes apply, and what records they require, depends on industry, investment category and other conditions, so confirm with the BOI or the relevant authority.

Can we capture data from old existing equipment?

There are several ways of capturing data even from machines with no communication capability: tapping the machine’s signals, adding external sensors or counters, or judging from images. Which suits depends on the type of machine, the process and the granularity of data you need. It is not the case that traceability is impossible unless you replace the equipment. The options and the reasoning behind them are set out in IoT Retrofits for Legacy Equipment. One caution specific to traceability: check at the outset whether the timestamps on data captured from equipment are aligned with your other records. If the machine’s clock has drifted, production lots and machine condition data cannot be matched when you put them side by side. It is a difficult thing to fix afterwards.

What should we do first?

Decide granularity, and measure where you stand. First, decide for each part family which granularity — lot, sub-lot or serial — you will track at, and keep the rationale on record. It costs almost nothing but becomes the premise for every judgement that follows. Next, run one mock trace exercise and measure how long it takes to produce the required information for a randomly chosen lot, and what cannot be assembled. With that measurement in hand, investment priorities become concrete. From there, starting with what can be fixed through forms and operating rules — adding a first applicable lot field to the 4M change notice, recording lot numbers in shipping records, assigning your own lot numbers at receipt — is one of the least wasteful routes in cost-effectiveness terms. Measures involving capital investment should then be considered only for the weaknesses that remain at that point.

Summary

What automotive parts traceability is really tested on is not whether records exist, but your power of proof: how narrow a range you can contain when a defect occurs. If you cannot narrow the range, everything shipped during the suspect period becomes the subject of action. In that sense, traceability is less a mechanism for improving quality than a mechanism that determines the size of the loss when a quality problem occurs.

The biggest design fork is tracking granularity. Choosing lot, sub-lot or serial changes both the range you can isolate and the load on the floor substantially. Putting everything under serial control is not the right answer; granularity is decided per part family against axes such as customer requirements, safety involvement, unit price, volume and sources of variation. What an audit asks about is less the granularity itself than the rationale for choosing it.

Traceback and trace forward need different data. The former is about linkage inside the process, the latter about linkage to shipping results. In most plants the latter is weak, yet the first thing a customer asks for is the shipped range. It is worth actually walking your own records to confirm which side you are weak on.

If 4M change records are not linked to production lots, isolation depends on someone’s memory. Simply adding a first applicable lot number field to the change notice changes the amount of traceable information substantially. As an improvement that can be started without system investment, it deserves a higher priority than it usually gets.

Record carriers are chosen on three axes — process environment, reading method and acceptable cost per piece — and RFID is not always at the top. Changing carrier by process and connecting them on the data side is the realistic design, in which case the linkage at each carrier boundary is the critical point.

For a plant in Thailand, the specific issues are that linkage breaks easily at receipt of imported materials, that records degenerate into a formality unless entry burden is reduced in a multilingual environment with staff turnover, that BOI usage reporting and production lot records easily end up in separate ledgers, and that quality assurance tends to thin out during the launch periods that the EV shift makes more frequent.

And what a customer audit looks at is repeatability, not an explanation of the mechanism. Given this lot number, how many minutes would it take you to produce the required information — and could someone other than the usual owner do it? Running mock trace exercises in peacetime and recording elapsed times and gaps gives you audit material directly.

You are welcome to talk to us before you have decided which parts to track at which granularity. In fact, the conversation is usually more productive before granularity is settled, because we can work through the reading of the customer requirements together. Describing the situation as it is on the floor is enough — “we only have route tags and an Excel ledger,” “material lots stop being traceable at receipt,” “we ship, but there is no record of which lot went where.” Feel free to get in touch through our contact form, and we will work through where it is realistic to start, making use of the equipment and records you already have.

References

• ISO/IATF SCHOOL, “IATF 16949 Clause 8.5.2: Identification and Traceability” (in Japanese)
• ISO/IATF SCHOOL, “IATF 16949 Clause 8.5.2.1: Traceability” (in Japanese)
• IATF, “IATF 16949:2016 Frequently Asked Questions,” Japanese edition (PDF) — general information on the standard and certification scheme
• Japan Quality Assurance Organisation (JQA), “IATF 16949 (Automotive)” — general information on the standard and certification scheme, in Japanese
• OMRON Traceability Navi, “Standards and Regulations in the Automotive and Automotive Parts Industry” (in Japanese)
• OMRON Europe, “Traceability in Automotive”
• JETRO, “Thailand (2): The Current State of the Automotive Industry, Global Supply Chains and the Japanese Response,” November 2025 (in Japanese)
• JETRO, “The Current State of Thailand’s EV Policy: Raising Local Content and New HEV Support Measures,” February 2025 (in Japanese)
• Thailand Board of Investment (BOI), press releases (Japanese page)