How to fashion supply chain traceability: a practical method

What can put a farm-origin claim on hold?

A farm-origin claim should stay on hold when an invoice quantity exceeds the transaction certificate and predates its coverage period. In a fictional brand pilot for one undyed cotton T-shirt, the certificate covers 10,000 kilograms, while the invoice lists 12,000 kilograms.

The supplier finds that the invoice combined eligible and conventional yarn. It issues corrected line items, ties the eligible quantity to the production order, and supplies shipment and receiving records. The brand reconciles output after documented loss, approves only the eligible batch, and describes the unverified quantity as generic cotton. A flax overshirt example reaches a similarly bounded claim for flax.

What makes a traceability pilot useful?

A useful pilot tests one defined claim or risk against evidence that can guide a named decision. Keep it to one product, one decision owner, and one bounded question, then expand only when the process works; a general promise of full transparency is too broad to test.

Set its scope as follows: For [product and period], verify [claim or risk] through [named tiers and events] so [decision owner] can [take action]. One example is confirming cut-and-sew, dyeing, weaving, spinning, and cotton transaction records for navy shirt lot 24-081 before approving its farm-origin claim. A sustainable fashion checklist can help frame the choice.

A broader sustainable fashion evaluation can inform the product decision alongside traceability. Keep records and final approval judgeable against stated evidence rather than a general ambition; a pilot that cannot connect its evidence to a decision has not tested the intended question.

Which product should I use to test the hard parts?

Choose a manageable, representative product that still exposes meaningful risks, transformations, and supplier relationships. The easiest item may hide gaps, while an oversized project can make the first effort unmanageable; select a bounded case that lets the team test evidence and working relationships.

Use one product and one claim, then extend the process after it works. Implementation may call for legal, labor, environmental, customs, privacy, and product-claim specialists. Definitions and a clear object hierarchy help those specialists coordinate, but they do not replace specialist advice when implementation requires it.

How narrowly should I define what is being checked?

Specify the product revision, color and size scope, production period, material components, countries, supplier tiers, chain-of-custody model, claim wording, and exclusions. Name whether the model is identity preserved, batch based, quantity allocated, or facility mapped, so the boundary is clear before records arrive.

The OECD due diligence framework asks companies to prioritize significant impacts, work with business partners and stakeholders, and improve over time. Traceability should inform sourcing changes, prevention, mitigation, monitoring, communication, and remedy. A map that does not influence those actions is not a complete due-diligence process.

Which identifiers should follow the product?

Use recognized identifiers wherever they fit, create controlled internal codes only to fill gaps, and maintain crosswalks to supplier codes. Each identifier must be unique, persistent, resolvable, attached at the correct level, and keep similarly named sites separate throughout the system.

GS1 identifiers include GTIN for trade items, GLN for parties and locations, and SSCC for logistic units. EPCIS is a GS1 standard for sharing event data, not an identifier. Keep those functions distinct in the data model and connect each code to the object it identifies.

A product model ID is not a batch ID. An RFID serial may identify one label while the database links it to a garment, lot, and product version. Specify which object each identifier resolves to and preserve those connections as information moves between systems.

What should my data dictionary define?

Define every field before collection, including its format, unit, permitted values, source, owner, update frequency, confidentiality, validation method, and retention rule. Shared definitions help suppliers provide consistent data and let reviewers interpret records across systems and handoffs without guessing what a field means.

Core objects may include:

Keep legal entity and facility in separate fields; a single free-text supplier entry must not stand for both the company and its production site. This distinction matters when similar business or location names appear in records, and gives each object a defined owner and position in the data model.

What should I confirm with suppliers before collecting records?

Ask each direct supplier to confirm its legal entity, production site, process, subcontractors, and upstream sources. Confirm system capability and responsible contacts, then check high-risk identity and scope using independent sources and site-level records rather than relying on a supplier list alone.

At EU level, the forced-labor products regulation, Regulation (EU) 2024/3015, prohibits placing or making available on the EU market, and exporting from it, products made with forced labor; it applies beginning December 14, 2027. The Corporate Sustainability Due Diligence Directive, Directive (EU) 2024/1760, requires in-scope companies to address adverse human-rights and environmental impacts.

In the United States, the Uyghur Forced Labor Prevention Act makes supply-chain evidence relevant to imports linked to Xinjiang. A supplier roster alone cannot show where or by whom goods were made; evidence must connect relevant production sites with records.

Separately, the Ecodesign for Sustainable Products Regulation (ESPR, Regulation (EU) 2024/1781) establishes a framework for Digital Product Passports. Product-specific requirements depend on delegated acts, and the regulation provides for a Digital Product Passport registry; check applicable textile rules before treating a passport as mandatory.

How can I request data without creating extra work?

Explain why the data is collected, who can access it, how corrections are handled, and which fields are required. Small suppliers may need templates and training, while repeated manual requests without stable definitions create fatigue and inconsistent replies; set expectations before routine collection begins.

If a supplier has no API, use a controlled template or portal with validation, then plan integration according to volume and risk. Manual input still needs a named owner and an audit history, so the lack of an API does not remove the need for defined fields and review controls.

Which production events should I record?

Record each movement or transformation as a linked input-output event, including the site, time, quantity, transaction references, and supporting evidence. Capture normal losses and units too; convert kilograms to meters only with fabric specifications and tolerances, not with an unexplained formula.

How can I tell whether quantities and certificates agree?

At each actor, compare opening eligible stock plus verified receipts, less documented losses and outputs, against closing stock. Flag negative stock, impossible yields, output recorded before input, duplicate transaction IDs, expired certificates, and claims above entitlement for review before accepting a claim.

Relevant schemes include the Global Organic Textile Standard (GOTS), Organic Content Standard (OCS), Recycled Claim Standard (RCS), Global Recycled Standard (GRS), and Responsible Wool Standard (RWS). GOTS and Textile Exchange standards have scheme-specific chain-of-custody documents, including transaction certificates where required.

Check the current scope held by the issuing certification body, along with certificate validity, product or material, quantity, and transaction dates; a certificate does not prove every claim. A certification file review can help structure those checks before you rely on a document for product-level wording.

Discuss exceptions with the supplier and preserve the original record, correction, reason, approver, and time. Never overwrite a failed result without a trace; retain the change history so a reviewer can understand what was corrected and why before the revised evidence is accepted.

What evidence supports a claim?

Build a claim-to-evidence matrix that connects every statement with the records needed to substantiate it. Set out the expected proof before review, so the team can assess scope, completeness, and whether the available records support the wording being considered.

For farm origin, list every custody transfer and transformation. For recycled content, document input classification, model, allocation, percentage method, tests, and production match, tracing the relevant records from source through the finished item so each part of the claim can be checked.

Use independent verification in proportion to potential harm, legal exposure, claim breadth, supplier risk, and data weakness. Record the sampling method and do not present sampling as complete coverage. A claim diagnosis guide can help examine scope, evidence, and certification issues.

Examples of textile certification or verification bodies include Control Union, Bureau Veritas, and IDFL; traceability platform examples include TrusTrace, TextileGenesis, and retraced. Confirm provider independence, relevant scheme and material scope, and coverage of actual sites and transactions. A platform can organize records, but it does not verify provenance by itself.

What should a customer be able to understand?

Ask a reviewer unfamiliar with the project to read the label or digital page and identify the origin, material, labor, and environmental conclusions it communicates. Narrow wording when its likely meaning extends beyond the evidence; label verification steps can help test the overall impression.

Show the date, product identifier, mapped tiers, verification status, chain-of-custody model, and known gaps. Avoid a green map graphic that suggests every listed site passed every social or environmental criterion; communicate what the evidence establishes and where it stops.

Can I test the system before a real incident?

Test the system with one material batch and a simulated problem: locate affected finished goods, warehouses, customers, and remaining stock, then insert a false facility record. The controls should detect it, quarantine it, correct it, and document the response without relying on a presentation.

Measure lookup time, unresolved units, false matches, and correction time. Those checks indicate whether the process works during a real problem, not just in a presentation, and whether the team can locate relevant records and document its response within its working systems.

What can RFID tell me about origin?

An RFID label can carry or point to an identifier, but it does not verify the database entry or establish product provenance by itself.

Accurate attachment, event records, evidence, and controls remain necessary. Serialization improves identification granularity but does not guarantee truth; those controls determine whether the identifier refers to accurate garment information.

How often should I check records again?

Validate records at entry, reconcile them at defined production or transaction intervals, and reverify high-risk fields when claims, suppliers, sites, or standards change. Set timing to match the activity being checked and the risk of relying on outdated information, rather than treating every record as equally current.

When should I hold back publication?

Stop when material identity, quantities, facility, chain-of-custody status, or claim scope cannot be reconciled to reliable evidence. Do not publish a broader result while an unresolved mismatch could change what the statement means to a customer.

What should I do first?

Start with one bounded pilot and make publication conditional on reconciled, verified evidence. Connect traceability records to risk assessment and corrective action, then use exceptions, corrections, and lookup drills to decide whether the process is ready to expand.