Batch-to-batch traceability in production: what the ERP must ensure

An Inditex buyer calls a garment maker near Famalicão on a Friday at five in the afternoon. They have detected a problem in a yarn from a specific batch and need to know, by Monday, which finished garments used that yarn, which customers they went to and which containers they are in. The correct answer takes eight minutes in an ERP with real traceability. In most Portuguese factories, it takes two days, three phone calls to the dyeing house and an Excel sheet the foreman put together from memory. The difference between the eight minutes and the two days is not luck — it is data architecture.

The thesis of this article is simple and uncomfortable: batch-to-batch traceability is not a module you switch on, it is a design decision that runs across purchasing, production, warehouse and invoicing — and most ERPs sold to Portuguese industrial SMEs record batches without linking them, which is traceability on paper and amnesia in practice. The distinction seems academic until the first incident. After the first incident, it is the only thing that matters.

It is worth clearing up a common confusion at the outset. Many people hear "traceability" and think of a barcode, a label, a reader. That is the surface. Traceability is what happens beneath the label: the link between what came in, what was transformed and what went out, kept unbroken at each change of hands. You can have all the labels in the world and no traceability at all, if the labels are not linked to one another in the database.

1. The real operational problem

Traceability almost always fails at the same place: at the joint between two processes. The yarn comes in with a supplier batch, is dyed and acquires a new dye batch, is knitted and acquires a roll batch, is cut and — here the chain dies — the cutting combines six rolls from three different dye batches in a single works order, and nobody records which panel came from which roll. From the cutting onwards, traceability is a fiction.

This is the general law that runs across every sector: information is lost at the transitions, not in the processes. Inside a machine, the batch keeps itself intact — the machine does not mix what it is not told to mix. It is when material passes from one person in charge to another, from one system to another, from one shift to the next, that identity dissolves. Anyone who wants to protect traceability must look at the borders, not at the centres.

Textiles and apparel: the chain that breaks at finishing

In the textile cluster of the Vale do Ave — Famalicão, Guimarães, Vizela, Barcelos — the problem is compounded because the value of traceability lies precisely at the point where it is hardest to maintain: finishing. A dye batch with a shade deviation (the notorious batch-to-batch colour problem) can contaminate thousands of metres. When the parent company demands compliance with the European strategy for sustainable and circular textiles, it wants to know the origin of the fibre, the water consumption per dye batch and the footprint of each finish. Without batch-to-batch traceability, this is answered with estimates — and estimates do not pass a brand audit.

Dyeing is a special case because colour is, in itself, a batch variable that is not fully controllable. Two loads of the same dye, the same baths, the same parameters, come out with slightly different shades. That is why the industry works with the concept of a shade batch: material guaranteed to match visually. A garment with panels from two different shade batches has a visible shadow at a seam — and that is what the brand's quality inspection detects and returns. Tracing which panel came from which shade batch is not bureaucracy; it is the difference between reworking one garment and reworking a container.

There is a structural aggravating factor here in the Portuguese sector. The garment industry of the North works under intensive subcontracting for the parent companies — Inditex, Decathlon, Mango, Tom Tailor, Lacoste. The raw material often arrives supplied or imposed by the brand, with the batch defined upstream, outside the factory's control. This means the factory inherits partial traceability from another and has to keep it intact throughout its own processes, on pain of being the one — the weakest link in documentary terms — to take the blame for a chain that broke before it arrived.

In a traceability chain, whoever has the weakest record always takes the blame for the problem — even when the problem did not originate there. The Portuguese garment industry learns this the expensive way.

Footwear: 800 SKUs and three axes the generic ERP does not model

In Felgueiras and Guimarães, a sample collection has between 800 and 1,200 SKUs, crossing colour, size and last (the fitting). Traceability here is not just of raw material — it is of the physical pair. When an international buyer visits in February for the women's collection and detects a bonding defect in a sample, the question is: which adhesive, which batch, which operator, which machine. An ERP that treats the sole, the lining and the leather as components without a batch answers "I don't know" — and the "I don't know" costs the order.

The difficulty of footwear does not lie only in the number of SKUs, it lies in the three-dimensional structure that generates them. A model has colours; each colour has sizes; each combination may have different lasts or widths. A generalist ERP models this by stacking flat SKUs — it creates one thousand two hundred independent items that do not know they are the same model. Then, when the buyer asks "how many pairs of this model, across all colours and sizes, used this batch of soles", the system cannot aggregate because it never knew those one thousand two hundred items shared a common axis. The vertical ERP models the axes as dimensions, not as separate items — and it is that modelling that enables both planning and cross traceability.

Add to this the pace of the sector. International buyers visit twice a year — men's footwear around August, women's around February. In those windows, the entire collection is decided, and a sample with an untraceable defect is not just a lost sample: it is a doubt about the factory's ability to guarantee consistency in mass production. Sample traceability is, in practice, a demonstration of industrial competence to those about to place the big order.

Food distribution: recall in hours, not days

In a food distribution centre in the Lousada/Paços de Ferreira corridor, batch-to-batch traceability is a legal obligation, not an option. Regulation (EC) 178/2002 requires one-step-back and one-step-forward traceability for foodstuffs. When there is a withdrawal alert, the warehouse manager needs to know, without stepping away from the radio for more than two hours, which pallets contain the affected batch and which customers it has already gone to. If picking did not record the batch at the moment of separation, the recall becomes a manual sweep of the entire warehouse.

The friction point in food is almost always dispatch, not receipt. Almost everyone records the batch when goods come in — the supplier's invoice requires it, the expiry is on the label, receipt is a calm and controlled moment. The problem is the outbound: the picking operator, under time pressure, grabs the box nearest to hand on the shelf and not the one the system told them to, because the two boxes look identical and they have thirty lines to pick before the distribution run closes. If the system does not force confirmation of the batch at the moment of separation, the outbound record becomes disconnected from the physical. And it is the outbound record — downstream traceability — that the recall demands.

There is a human detail here that no requirements document captures: the warehouse manager in the Lousada/Paços corridor will resist any process that takes them off the radio for more than two hours. They coordinate unloading, loading, shortages and complaints in real time, and traceability that forces them to stop and type into a slow terminal will be sabotaged in silence — not out of bad faith, but because the operation does not wait. Any implementation that ignores this fact is dead before it starts. Batch capture has to be faster than the manual alternative, or the operator will find a way around it.

Traceability is not proven when everything goes well. It is proven on Friday afternoon, when someone asks "where is this batch now" and the answer has to be exact within minutes.

Metal and plastic: the mixing of granulates and the one-off mould

In the Aveiro–Marinha Grande corridor, the mould-making and plastic injection industry lives two opposite traceability realities within the same company. In series injection, the problem is the mixing of granulates: a batch of virgin raw material mixed with recycled material or with colour masterbatch in a hopper that feeds continuously — and the parts that come out over the following hours have overlapping genealogies that only a precise time record can distinguish. In mould-making, by contrast, each mould is a unique piece of very high value, and traceability has to descend to the individual serial number, with a record of each steel, each heat treatment, each machining operation.

It is the perfect example that the right granularity is not a property of the company, it is a property of the product. The same factory needs batch traceability in injection and unit traceability in the mould — and an ERP that can only do one of the modes fails half the operation.

2. What batch traceability in Portuguese industrial production is, exactly

Batch Traceability is the ability to reconstruct, for any unit produced, its entire genealogy (which raw materials, which batches, which processes, which equipment, which operators) and its entire destination (which works orders consumed that batch, which finished products resulted from it, which customers they went to). These are two distinct directions and both are necessary.

Upstream and downstream traceability

• Upstream (backward / genealogy): from a finished product, knowing all the raw material and all the processes that formed it. This is the one that answers "where did this defect come from".
• Downstream (forward / dispersion): from a suspect batch of raw material, knowing all the finished products that contain it and where they are. This is the one that answers "what I have to withdraw from the market".

An ERP that does only one of the directions is half traceability. And half traceability, in a recall, is as useful as none — because the question that costs money is always the downstream one. The upstream one serves for the post-mortem: the defect has been identified, you want to know the cause, there is time to investigate. The downstream one is the urgency: there is a contaminated batch somewhere in the market and every hour that passes is more product sold, more reputational risk, more legal exposure. That is why systems built only for upstream genealogy — the ones that answer the "where did it come from" question well — tend to fall silent on the burning question.

Internal vs. chain traceability

Internal traceability lives within the four walls of the factory: it links inbound batches to outbound batches across the processes. Chain traceability extends to the supplier upstream and the customer downstream — and depends on each link communicating the batch in the documents. In Portugal, internal traceability is what separates a serious industrial ERP from an accounting ERP with a production module bolted on. Chain traceability depends on documentary discipline that most suppliers still do not have.

Chain traceability is a coordination problem, not a technology one. You can have the best internal system in the world and still be blind beyond your gate, if your supplier delivers material with no batch on the document and your customer does not ask for a batch on the delivery note. What changes this equation is almost always the powerful brand at the end of the chain: when Inditex or Decathlon impose traceability requirements on their direct suppliers, those requirements propagate backwards, link by link, because each supplier has to demand them of theirs in order to be able to comply. It is the downstream pressure that civilises the whole chain — and Portugal, being a subcontracting territory for major brands, feels that pressure earlier than many competitors.

Batch, sub-batch and serial number: do not confuse them

Three granularities that the manuals treat as synonyms and the factory treats as different things:

The classic Portuguese mistake: wanting a serial number where a batch will do (it drives up cost and nobody records it), or using a batch where a serial number is needed (the individual part is lost). The right granularity is the coarsest that still answers the questions you are asked.

There is a direct economic consequence in this choice that requirements documents tend to forget. Each level of granularity adds recording work to each operation — and that work is paid for every day, forever, while the benefit only appears on incident days, which are rare. Serialising a part that was worth tracing by batch is imposing a permanent daily cost for a benefit that never materialises. The discipline of choosing the right granularity is, in essence, the discipline of not over-recording — of not drowning the factory in confirmations that nobody will use.

Granularity is not a matter of ambition. It is a matter of proportion: the finest record costs every day and serves on the rare days. Choose the coarsest that still saves you on the rare day.

3. The landscape in Portugal today

Commerce in Portugal moved 201.8 billion euros in 2024, spread across around 218,800 companies and 860,600 workers, according to the INE. Within this universe, the overall commercial margin in wholesale trade was just 4.6% in 2024 — which means that stock and logistics efficiency, not the selling price, is what decides profitability. A poorly executed product withdrawal, with the destruction of good stock because the affected cannot be distinguished from the unaffected, eats that entire margin in a single incident.

It is worth dwelling on this margin figure, because it explains the whole economics of traceability. With a margin of 4.6% in wholesale trade, a company needs to sell more than twenty euros of goods to generate one euro of margin. If a poorly executed recall forces it to destroy a thousand euros of good stock — because it could not isolate the affected batch from the rest — it needs to sell more than twenty thousand euros just to recover that loss. Fine traceability is not an administrative cost; it is what separates destroying a pallet from destroying a warehouse.

The business structure that shapes maturity

The Portuguese industrial fabric is dominated by micro and small companies — the overwhelming majority of companies in Portugal have fewer than ten workers, and even within manufacturing, medium-sized units are relatively few. This has a direct consequence for traceability: system maturity correlates strongly with size, because a micro-company rarely has anyone looking after data architecture full time. The knowledge lives in the head of the owner or the foreman, and traceability "works" as long as that person is present and remembers. It is human traceability, not systemic — and human traceability fails on exactly the day the person is on holiday.

The gap between sectors

Traceability in Portugal is not uniform. In sectors with reinforced legal obligation — food, pharmaceutical, automotive safety components (IATF 16949 standard) — maturity is reasonable because the audit forces it. In sectors where traceability is a customer requirement and not a legal one — textiles, apparel, footwear, moulds — it depends entirely on who the parent company is. A garment maker that works for a demanding German brand traces; the same maker, on an order for a customer that does not ask, lets the discipline slip.

This is the most dangerous pattern we see in the field: intermittent traceability. The factory knows how to trace — it has the system, has the processes, has the competence — but only does it when the customer presses. On the orders where nobody asks, the operator saves the fifteen seconds of recording, and a hole is created in the database. The problem is that incidents do not choose the traced orders: when it blows up, it blows up on the order where the discipline was let slip. Traceability only protects if it is systematic — if the recording is mandatory on the screen, always, regardless of who the customer is. The kind that switches on and off according to the pressure is theatre traceability.

Traceability is not maintained by conviction. It is maintained by system. If the batch record is not mandatory on the screen where the operator is already working, it does not happen.

The weight of the new European textile regulation

The pressure comes from the European strategy for sustainable and circular textiles and from the European ecodesign regulation for sustainable products (ESPR), which introduces the Digital Product Passport. When it is fully in place for textiles, it will require verifiable information at product level on composition, origin and durability. A factory without batch-to-batch traceability will have nothing to feed that passport with — and the brand will choose another supplier that does.

The point that most business owners have still not internalised is that the Digital Product Passport turns traceability from a cost into a condition for market access. Today, those who do not trace lose a few orders from demanding customers. Tomorrow, when the passport becomes mandatory to place textiles on the European market, those who do not trace simply do not sell to the European Union. The fibre composition, the origin, the water consumption, the recyclability — all of this will have to be verifiable information linked to the physical product, not a generic declaration. And verifiable information at product level only exists if there is batch-to-batch traceability feeding it. The factories that start building that foundation now will have years of advantage over those that wait for the obligation.

There is also a funding window that intersects with this timetable. The instruments of PT2030, PRR and Norte 2030 fund investment in digitalisation and industrial sustainability — and traceability fits both axes. Those who align investment in traceability systems with a well-built application, with a defensible technical report, find support. Those who wait for the legal obligation to invest do so with no support and under pressure — the worst moment to negotiate anything.

4. The implementation models

There are four practical approaches to batch-to-batch traceability, and the wrong choice costs years. I describe them in increasing order of cost and fidelity.

Model 1: Documentary traceability (the pretend legal minimum)

The batch is recorded on the inbound and outbound, but consumption is not linked to production. It is known that batch X came in and that product Y was produced in the same month, but it is not known whether Y used X. This is what many generalist ERPs call "traceability" and it is, in practice, a warehouse audit — not a production one. It serves to pass a superficial inspection; it does not survive a real recall.

The danger of this model is that it gives a false sense of security. The business owner sees the "batch" field filled in on the documents, sees reports that list batches, and believes they are covered. They are covered until the day of the first downstream recall, when they discover that the system knows five yarn batches came in that month and that twenty references were produced, but does not know how to cross-reference the two things. Documentary traceability is the one that deceives best in sales demonstrations — it looks complete because it shows data — and serves worst in reality.

Model 2: Works-order traceability

The consumption of raw material batches is linked to each works order, and the resulting production to that same order. Genealogy is now answered at the order level: "this order consumed these batches and produced these quantities". This is the model sufficient for most textile and footwear SMEs. The limitation: if the order mixed several batches of the same component, it does not distinguish which panel came from which batch within the order.

This is the realistic starting point for most medium-sized Portuguese factories, and it is far more powerful than it seems. If the factory has works-order discipline — if each production is associated with an order, and each consumption is posted against that order — order-level genealogy answers nine out of ten of the real questions customers and regulators ask. The limitation of mixing within the order only bites in extreme cases, and even those can be mitigated with management rules: for example, not mixing different shade batches in the same order, which solves the problem at source rather than at the record.

Model 3: Shop-floor capture traceability

The operator confirms, on the industrial terminal, which batch they are consuming at the moment they consume it. This is where real-time production capture comes in, of the KORA Productivity kind, with barcode or QR reading on the raw material. Traceability becomes a by-product of the production recording the factory already does for efficiency KPIs and OEE. It is the cost-benefit sweet spot for Portuguese industry with terminals on the floor.

The elegance of this model lies in not asking for new work. A factory that already captures production on the terminal to measure efficiency — how many parts, in how much time, with what stoppages — already has the operator in front of the screen at the right moment. Adding the reading of the batch they are consuming is an additional two-second gesture, not a new task. Traceability comes for free on top of an investment made for another reason. That is why we often recommend tackling production capture and OEE first, and letting fine traceability hitch a ride — because the factory takes to a system that shows it its own efficiency better than to a system that only serves for the day of the recall.

Model 4: Unit traceability with serialisation

Each unit has a unique identifier and each operation is recorded against that unit. Necessary in pharmaceuticals, aerospace, automotive safety components and high-value moulds. Expensive in devices, discipline and data. It is only justified when the law or the customer requires it, or when the unit value of the product pays for the effort.

Serialisation is the extreme of fidelity and the extreme of cost. Each operation, on each unit, generates a record — the volume of data grows with the number of parts multiplied by the number of operations, and the discipline required of the shop floor is total, because a missed reading leaves an orphan unit that breaks the chain. It makes perfect sense in a mould worth tens of thousands of euros that takes weeks to produce; it makes no sense at all in a pair of socks. The mistake we see is the temptation to "do it right the first time" by serialising everything — and the result is a system the shop floor sabotages because it asks an effort disproportionate to the value of what is being recorded.

Choosing between the models is not choosing the most sophisticated. It is choosing the coarsest that still answers the question your customer or regulator is going to ask you.

Hybrid models: the reality of the factory

In practice, almost no factory uses a single pure model. The common — and correct — thing is to have different models in different parts of the operation, according to risk and value. A footwear factory may use shop-floor capture (Model 3) at bonding, where defects concentrate, and works-order traceability (Model 2) at stitching, where the risk is lower. The mould-making and injection factory we saw uses serialisation (Model 4) on the moulds and batch (Model 3) on series injection. What matters is that the ERP supports this heterogeneity without forcing everything to be levelled to the most expensive model or the weakest. A system that can only do one model forces compromises that cost at both ends.

5. How to assess whether your company needs it

Every factory needs some traceability. The question is the level. The golden question: when your biggest customer or the regulator asks for the genealogy of a batch, how long does it take to answer and with what confidence? If the answer is "days" or "with a sheet the foreman has to reconstruct", you have an architecture problem, not an effort one.

Step by step of the diagnosis

1. Map the process joints. List every point where a batch changes identity — raw material receipt, dyeing, cutting, assembly, packaging. Each joint is a point where the chain can break. The joints are the map of your risk.
2. Identify the joints where several batches mix. The cutting that combines rolls from different dyeings, the mixing of granulates in injection, the silo that accumulates. These are the critical joints: it is where fine traceability is won or lost.
3. List the questions you are asked. Write down the real questions customers, brands and regulators have already asked you in the last year. "Which yarn batch did this defect come from?" "Which garments used this finish?" The granularity needed is the one that answers these questions — no more, no less.
4. Measure the current response time. Time, with a real case, how long it takes to answer the hardest downstream question. If it takes more than half a day, your current system is not traceability — it is archaeology.
5. Assess where recording already happens. If operators already record production on terminals, fine traceability costs little to add. If they record on paper that someone types up at the end of the day, the problem is first one of capture, and only then of traceability.

This diagnosis gives you the level you need before any conversation with a software vendor. Anyone who goes into a demo without knowing which questions they have to answer buys what they are sold.

The absent foreman test

There is a quick and cruel test that reveals the real maturity of a factory's traceability: imagine that the foreman who "knows everything" is on holiday, without a phone, for two weeks. A customer calls with a problem in a batch. Can whoever is standing in answer with only what is in the system? If the answer depends on calling the foreman, your factory's traceability is not in the ERP — it is in one person's head. And knowledge that lives in a head is the most fragile asset a company can have, because it walks out the door with that person, whether on holiday, on sick leave or in retirement.

Cost of the problem vs. cost of the solution

Before investing, do the honest sums on both sides. On the problem side: how much did the last traceability incident cost — in destroyed stock, in investigation hours, in customer trust, in lost orders? How many times a year does it happen? On the solution side: what is the implementation cost and the daily operating cost of the model that solves it? If the expected annual cost of incidents is a small fraction of the cost of the most sophisticated solution, choose a coarser model. If a single incident can cost more than the entire solution — which is the case in food and in safety components — the investment pays for itself on the first bad day.

Any traceability that depends on a person remembering is not traceability — it is luck with a calendar. And luck fails on exactly the day that person is on holiday.

6. What to choose and why (decision by company size)

The size of the company and the complexity of the product decide the model. There is no single answer, but there are wrong answers for each profile.

The small subcontracted garment maker

A garment maker with fewer than fifty people, working on a make-up basis for parent companies, lives a specific tension: it does not control the raw material (it comes imposed by the brand), it has very tight margins and an often ageing workforce, uncomfortable with terminals. For this profile, serialisation is operational suicide — nobody will maintain it. Works-order traceability, well configured, with mandatory batch recording at receipt and dispatch, is the right point. The goal is not perfection; it is being able to answer the brand with confidence when it asks, and not being the weakest link in the chain.

The medium textile or footwear company with terminals

A medium-sized company, with complete internal processes and some terminals on the floor, is at the ideal point for shop-floor capture. Here it is worth investing seriously in vertical modelling — the colour-size-last axes in footwear, the shade batches in textiles — because the product's complexity justifies it and the company's size supports the recording effort. It is also the profile that benefits most from linking traceability to BI: with enough volume of genealogy data, you begin to see defect patterns by supplier and by machine that pay off the investment far beyond the day of the recall.

The distributor with a large warehouse

A distribution centre of five to fifty thousand square metres has the inverted challenge: production is simple (there is no transformation), but the volume of movements and the time pressure at dispatch are enormous. Here, everything is decided at picking: if the batch is not confirmed at the moment of separation, with a fast device that does not take the operator out of the flow, downstream traceability does not exist. Warehouse management with mandatory batch at picking is the heart of the system. The fatal mistake — which we see repeatedly — is investing in receipt, which is easy, and neglecting dispatch, which is hard and is the one the recall demands.

Why the vertical ERP wins here

A generalist ERP treats the batch as an optional text field on a movement line. A vertical MULTI ERP for industry treats the batch as an entity with its own genealogy, which runs across production, warehouse and invoicing without being lost at the joints. The difference does not appear in the demo — it appears two years later, in the first serious recall. For very high configuration complexity, QAD Adaptive ERP in low-code allows modelling genealogies that a closed ERP does not support.

On the decision itself: in Portuguese family SMEs, it is decided by the trio of CEO, CFO and the head of IT — and the latter is often a self-taught person with fifteen years of business knowledge and no formal degree, who knows exactly where the batch chain breaks because they have patched it up by hand dozens of times. It is them the demo has to convince, not the sales rep's slide deck. They will ask the uncomfortable questions — "and when the order mixes three shade batches, how do you tell them apart?" — because they have lived the problem in the flesh. A vendor who answers those questions well earns the trust of the person who really decides; one who answers with generalities loses it in the first technical meeting. It is worth reading what we say about what the ERP reveals that management did not see and about change management in an ERP implementation.

7. Applicable regulatory framework and compliance

Traceability intersects with several layers of legal obligation in Portugal, and each asks something different of the ERP.

Invoicing and communication to the AT

DL 28/2019 and Ordinance 195/2020 govern certified electronic invoicing, ATCUD and the monthly communication of the SAF-T. Traceability is not directly required by these instruments, but the batch must be able to travel to the dispatch document — because the transport document is the last point where chain traceability materialises for the customer. An ERP that loses the batch before invoicing breaks the chain at the most visible link.

There is a practical synergy here worth exploring. The infrastructure the factory already has to maintain to comply with tax obligations — AT-certified software, document communication, ATCUD — is the same that carries the dispatch document. If the batch is linked to the document line, it travels for free on the structure that already exists by tax requirement. Whoever designs the system with this link in mind gains chain traceability with no additional infrastructure; whoever forgets it discovers too late that the batch was lost precisely at the jump to the document the customer receives.

Food safety

For food distribution and retail, Regulation (EC) 178/2002 imposes one-step-back and one-step-forward traceability. It is not optional and has no grace period: whoever does not have it cannot legally place foodstuffs on the market. ASAE enforces it, and a food safety incident without adequate traceability is not just a withdrawal problem — it is exposure to a fine and to liability that can escalate far beyond the value of the product.

Automotive safety components

For those who supply the automotive industry — and there are Portuguese suppliers of plastic and metal components in this chain — the IATF 16949 standard imposes very demanding traceability requirements, frequently at batch level with complete genealogy and, for safety components, at individual part level. Here traceability is neither negotiable nor gradual: it is a condition for qualification as a supplier. A carmaker does not work with anyone who does not guarantee auditable traceability, because a vehicle recall for a defective component costs millions and the liability propagates across the entire supply chain.

Data protection, cybersecurity and data that becomes sensitive

When traceability records the operator who performed each operation, it creates personal data subject to the GDPR and to Law 58/2019. The purpose must be legitimate and proportionate — tracing quality, not monitoring individual productivity without basis. And because this genealogy data is a critical asset, the NIS2 Directive (EU 2022/2555), transposed in Portugal by DL 65/2025, and ISO 27001 certification become relevant for those who store it. A system with EDR and adequate monitoring protects the integrity of the traceability record itself — which is worth nothing if it can be tampered with.

This point about the operator deserves care because it touches a sensitive line in labour relations. Recording that operator X consumed batch Y in operation Z is legitimate for quality traceability purposes — if there is a method defect, you want to know who performed it in order to train them better, not to punish them. But the same data, used to measure and rank individual productivity without basis and without prior information, enters territory of surveillance that the GDPR and Portuguese labour case law do not tolerate. The practical rule: define the purpose in writing, inform the workers, and do not use data collected for traceability for disciplinary ends that were not declared. An implementation that ignores this wins the system and loses industrial peace — and a factory in labour conflict traces nothing well.

Traceability and cybersecurity are the same problem seen from two sides: a record that can be altered without leaving a trace is not traceability, it is literature.

8. How INFOS approaches this

We work traceability not as a separate module, but as a property that runs across the vertical ERP. In the MULTI ERP, the batch is an entity with its own genealogy: it links from the supplier to the dyeing, to the cutting, to the assembly and to the dispatch document, without being lost at the process joints where generalist ERPs let it drop. In the footwear and apparel sectors, this means modelling the colour-size-last axes and the by-component genealogy that a sample collection requires.

Shop-floor capture with KORA Productivity turns fine traceability into a by-product of the production recording the factory already does to measure efficiency — the operator confirms the batch on the same terminal where they already log the operation, with no duplicated work. In the warehouse, the KORA Inventory Suite makes the batch mandatory at picking, so that the downstream recall is a matter of minutes and not days. And Qlik Sense turns the accumulated genealogy into analysis: which suppliers, which batches and which machines concentrate the defects.

We learned some of these things the hard way, in projects that did not go as the plan said. We have seen implementations where batch capture was imposed across the whole factory at once, and the shop floor reacted by slowing down — because every second of additional recording, multiplied by thousands of operations, is felt at the end of the shift. The lesson was clear: fine traceability enters by critical joint, one at a time, never in a big bang. We also learned that a system that gives nothing back to the operator — that only asks them for records and does not show them their own efficiency — is a system the operator will circumvent. That is why we link traceability to the production capture the operator wants to see, so that the batch record hitches a ride on something that already matters to them.

We do not sell serialisation to those who need works-order traceability. Our job is to get the level right — the coarsest that still answers your customer's questions. For those funding this, it is worth knowing what PT2030 funds and what it does not fund in industrial software, and the concrete case of total production control in footwear.

9. 30/60/90-day roadmap

Batch-to-batch traceability is not implemented in a big bang. It is implemented by process joint, starting with the most critical. Here is a 90-day path with verifiable milestones.

Days 1-30: map and win the first quick win

• Complete the five-step diagnosis from section 5 and identify the single critical joint where the chain breaks the most.
• Quick win: make batch recording mandatory at raw material receipt. It is a configuration change, not a project, and it closes the entry door.
• Quick win: print the batch on the raw material labels in the warehouse, so that reading on the shop floor is possible right away.
• Time the current response to the hardest downstream question, to have a baseline against which to measure progress.
• Involve from day one the self-taught head of IT and the foreman of the critical joint — they are the ones who know where the chain breaks and they are the ones who will make the system live or die.

Days 31-60: link consumption to production at the critical joint

• Configure works-order traceability (Model 2) at the identified critical joint, linking batch consumption to the resulting production.
• If terminals already exist on the floor, activate batch confirmation at consumption (Model 3) only at that joint — not across the whole factory at once.
• Quick win: make batch recording mandatory at dispatch picking for the products with the highest recall risk. It closes the exit door.
• Run a simulated recall test: choose a batch and time the downstream response. Compare with the baseline.
• Define in writing the purpose of the operator record and inform the team, to settle the GDPR question before it turns into conflict.

Days 61-90: extend, measure and institutionalise

• Extend batch capture to the remaining joints in order of risk, not all at o

Frequently asked questions

What is batch-to-batch traceability in an ERP?

It is the continuous link between what came in, what was transformed and what went out, kept unbroken at each change of hands. It is not just labels or barcodes — it is the data structure that connects batches to one another across the different processes, from purchasing through to invoicing.

Where does traceability fail in most Portuguese factories?

It fails at the joints between two processes. For example, when material passes from one person in charge to another or from one system to another, the batch's identity dilutes. In textiles, the chain frequently breaks at cutting, when six rolls from three different batches are combined in a works order without recording which panel came from which roll.

Why is traceability critical in dyeing?

Because colour is a batch variable that is not fully controllable. Two identical loads come out with slightly different shades, creating shade batches. A garment with panels from two different shade batches shows a visible shadow at the seam, detected in the brand's quality inspection.

How does subcontracting affect traceability in garment makers?

The raw material often arrives supplied by the brand with the batch defined upstream, outside the factory's control. The garment maker inherits partial traceability and has to keep it intact. In a broken chain, whoever has the weakest record always takes the blame, even if the problem originated earlier.

What is the specific challenge of footwear in traceability?

A collection has between 800 and 1,200 SKUs, crossing colour, size and last. Traceability is of the physical pair — which batch of soles, which adhesive, which operator. A generic ERP treats components without a batch and cannot aggregate data when the buyer asks about a whole model.

Why does a generic ERP not solve traceability in footwear?

Because it models the three-dimensional structure by stacking flat SKUs — one thousand two hundred independent items that do not know they share a common model. A vertical ERP models the axes as dimensions, enabling both planning and cross traceability across colours, sizes and lasts.

What is the difference between having labels and having traceability?

You can have all the labels in the world and no traceability at all, if the labels are not linked to one another in the database. Traceability is what happens beneath the label — the connection maintained in the data structure, not on the surface.

Sources

• Regulation (EU) 2023/1115 on deforestation and forest degradation (EUDR) — traceability requirements for textile and footwear supply chains
• EU Strategy for Sustainable and Circular Textiles (2022) — European Commission, Directorate-General for the Environment
• Standard ISO/IEC 27001:2022 — Information Security Management Systems, applicable to data architecture and traceability in ERP
• CNPD (National Data Protection Commission) — Guidelines on the processing of personal data in industrial traceability systems
• INE (National Statistics Institute) — Classification of Economic Activities (CAE) for textiles, apparel and footwear in Portugal