How far up the chain you actually need to go

The short answer

Depth is a property of the fact, not of the supply chain. Before asking how far back your evidence has to reach, ask where the fact you are about to publish becomes true, because for most attributes that point sits a long way downstream of where a traceability programme begins.

Take the three attributes that get argued about most. Customs origin for a garment is conferred at making up, so records covering the spinning and the weaving describe steps the rule says do not decide it. Fibre composition is recoverable from the garment on your desk by a laboratory, so a supplier's attestation is a second opinion about something you can establish yourself. Recycled content is the exception, and it is a complete one: it becomes true several steps upstream, nothing done to the finished product establishes it, and an unbroken record is the only mechanism that exists.

None of that is an argument that knowing less about your supply chain is better. A deeper picture is worth having for reasons that have nothing to do with a passport, including buying decisions, risk and the questions a regulator may ask about something else entirely. The narrow claim here is that the depth a given field needs is decided by the field, and that three fields sitting next to each other in the same record can want three different answers.

So the useful question in a sourcing meeting is not how far back the programme reaches. It is which fields the reach is supposed to fill, and whether those fields become true anywhere the programme goes.

Where each fact becomes true

Customs origin is conferred when the garment is made up

Under the non-preferential rules, a garment takes the origin of the country where it was completely made up. Making up means everything following cutting or knitting or crocheting to shape. Spinning, weaving and dyeing are not origin-conferring operations for these goods, which means a mill relationship two steps upstream produces evidence about steps the rule has already excluded from the test.

That is the whole of it for this field. The evidence has to reach the place the garment was assembled and the records held there. It does not have to reach the loom, and reaching the loom does not make the origin statement any better supported than it already was.

One boundary matters more than anything else in this section. The rule above is a customs determination, made under customs law for duty and trade measures. It is not the test applied to a sentence written for a shopper, and it is not a defence to one. Those are two questions wearing the same word, and they are separated properly on the country of origin page rather than merged here.

Fibre composition is recoverable from the finished garment

The textile labelling regulation operates on the product as sold. It sets a closed list of fibre names and requires percentages by weight, and it provides for the stated composition to be checked against the result of analysis by a prescribed method. The fact is fixed at the mill, when fibres are blended into a yarn or a fabric, and it is still legible in the garment afterwards.

The consequence for a sourcing plan is unusually clean. A brand with no relationship at all beyond its immediate supplier can establish this attribute, because the evidence is the product. Upstream visibility adds nothing that a test does not already give you, and the fibre composition page covers what the field is allowed to say once you have it.

We read the fibre name list and the provision governing its use at the regulation itself. The analysis provision we confirmed and did not read in full, so this page states that a prescribed method exists and quotes no part of it, and it states no tolerance figure.

Recycled content becomes true before anybody you have a contract with

This is the attribute the rest of the page turns on and it gets its own section. The short form is that it originates at recovery well upstream of the garment, and that nothing about the finished product carries it.

The one where the depth cannot be avoided

For recycled content, depth is unavoidable. There is no shorter route, no laboratory shortcut and no version of the claim that a closer supplier can settle for you.

Whether a fibre was recovered or made from virgin feedstock is a fact about material that stopped being visible once it became yarn. No laboratory test at scale determines recycled from virgin origin on a finished product, which reaches us through a summary of the preparatory study rather than from the study's own file. Because nothing acting on the garment can establish it, the only mechanism anybody has is a record carried forward from the point of recovery, step by step, with every step intact.

That is what makes this attribute different in kind rather than in degree. For origin, a break in the record upstream of making up is irrelevant. For composition, a break anywhere is irrelevant, because the garment answers the question. For recycled content, a single break anywhere in the chain ends the claim, and the chain is long.

Nothing anybody has looked at closes that gap. This estate went through the published passport data models asking whether any of them carries a rule for turning a batch level record into a product level claim, and did not find one. That is an absence found by us in the models we looked at rather than a statement that no such rule exists anywhere, and the difference between those two sentences is set out on how we know. What a certificate does and does not support, once you hold one, is the subject of the recycled content page.

What the depth data shows, and where the break actually is

A brand whose central proposition is full traceability publishes its own account of how far its visibility reaches. The instructive part is not the headline figure. It is the distribution underneath it, because the weakest point is not where a sceptic would guess.

Visibility is strongest exactly where a traceability programme starts, at the mill and the fabric supplier, and weakest at trims: buttons, zips, labels, thread and interlinings. Those are the cheapest components in the garment and usually the most numerous. They are ordered late, substituted freely when a colour runs short, and bought from suppliers a brand has no data relationship with at all. Closing that gap is hardest precisely where the value of closing it is lowest.

We hold that brand's figures second hand and carry the shape of them rather than restating the numbers, with the brand's own framing attached: it publishes them as its own measure of its own chain, not as an industry benchmark. The finding worth taking is structural. A programme that has reached the mill has done the part that is technically hard and commercially easy, and has not touched the part that is technically easy and commercially thankless.

Depth achieved is not quality achieved

Suppose a programme does reach all the way back. That settles how far the record goes and it settles nothing about whether the record is right, and the two get conflated constantly because the first one is easy to put on a slide.

A peer-reviewed pilot reached the farm and still concluded that transparency beyond the second tier of suppliers remains a significant challenge. A practitioner quoted in it named the mechanism plainly: data gets altered as it moves between systems, sometimes intentionally and sometimes not. Reaching a node is one thing. Receiving from that node a value nobody has re-keyed, re-rounded or quietly adjusted on the way is another.

There is also a documented case in which a deep certified chain was defeated outright. Forged transaction certificates were passed downstream, and a buyer who tried to verify them was directed to a cloned version of the issuing body's website, which confirmed them. Every structural feature people rely on was present, including a certified chain, documents that referenced each other and an online check. It is carried here as a case and never as a rate. The fraud rate most often quoted in this territory is an indicative figure more than a decade old, with no denominator that survives inspection, and this page does not publish it.

The standards body that governs chain of custody says the limit itself, in its own published abstract, more usefully than any critic has.

A chain of custody document is not, on its own, able to support claims about an organisation's materials or products.

Only the abstract has been read. The standard sits behind a purchase, this build did not buy it, and every freely available account of the chain of custody models it governs turned out to be a vendor page, so none of those models is named anywhere on this page. That is a narrower position than the category usually takes, and it is the position the reading supports.

Read the quoted sentence carefully, because it is not a criticism of chain of custody records. It says what the record is: an input to a claim rather than the claim itself. Something else has to connect the document to the product, and that something else is where most passport programmes quietly fail.

Attribute by attribute

The table below is the page compressed. It asks the same four questions of each attribute, and the answers do not travel between rows, which is the whole reason a single depth target for a whole programme misallocates effort. What each attribute is, and which of them a passport is expected to carry, sits on the attributes we track.

Two columns do the work here. The laboratory column tells you whether a dispute can be settled without anybody upstream cooperating. The reach column tells you where to stop asking. When those two disagree with what a programme is scoped to do, the programme is being scoped by its vendor's product rather than by your record.

What to stop doing

The practical output of this page is subtractive. Most of what follows is work to remove rather than work to add.

1. Stop asking suppliers to attest to composition. The garment answers it and a test settles a disagreement about it. Asking is not harmful, but it should not sit on the critical path of a passport, and it should never be the evidence of record.
2. Stop chasing origin evidence upstream of making up. A mill certificate does not improve an origin statement that is conferred somewhere else. Spend the same effort confirming that the making-up site named in your system is the one the goods were actually made in.
3. Stop treating one depth target as a programme. A target expressed as a number of steps upstream is a target about a supply chain, and every field it is meant to fill is a question about a fact. Sort the fields first and the depth falls out of the sorting.
4. Ask once, properly, about the things that genuinely need asking. The attributes that do require somebody upstream to write something down deserve a single well-formed request rather than a recurring one, which is set out on asking suppliers.
5. Put the recovered effort into trims. It is the part of the chain the published depth data says nobody reaches, and it is cheap to start on because the suppliers are small and the questions are simple.

Two things this page cannot tell you, and they are worth stating rather than implying. What any of this costs is not established: no figure we found carries a denominator or a scope we would publish, so this page publishes none. And whether an act covering textile products will mandate a depth of its own is unknown. If one does, the requirement replaces this analysis rather than adjusting it, and the estate's method for handling questions in that state is on how we know.

The rest of the practical sequence, from what a catalogue already holds to what has to be asked for, is on the Shopify guide, which is where this guide's family starts.

Sources

• Commission guidance on non-preferential rules of origin for textiles and garments
  In forceRelevant provisions reviewed

  Read for one rule, which is the strongest row on this page: for garments, origin is conferred by complete making up, meaning everything following cutting or knitting to shape, and spinning, weaving and dyeing do not confer it. It is a customs origin statement. It is never used here to answer an origin claim made to a shopper, and that boundary is held on the country of origin page rather than blurred on this one. This build did not establish the guidance's own address, so the source is named without a link.

• Regulation (EU) No 1007/2011 on textile fibre names and related labelling, consolidated text
  Art. 5(1), Annex ICELEX 02011R1007-20180215In forceRelevant provisions reviewed

  Read at the closed list of permitted fibre names and the provision governing their use. It is on this page for one consequence: composition is a property of the finished product, checkable against the result of analysis by a method the regulation prescribes. The analysis provision itself was confirmed and not read in full by us, so this page states that a prescribed method exists, quotes none of it and states no tolerance figure.

  View official source

• Joint Research Centre, Study on DPP content for textile apparel products under ESPR, 13 May 2026
  ProposedReached through a secondary reproduction, primary text not read

  The preparatory study informing the textile decision, dated 13 May 2026. Its granularity section was read at its own address in an earlier pass. The two propositions this page takes from it, that no laboratory test at scale determines recycled from virgin origin and that the chain it describes between raw material and retail runs to at least fifteen nodes, reach us through a testing house summary rather than from the study's own file, so the read is recorded as second hand and the page says so in the sentences that use it. It is pre-decisional by design and it is never cited here as law.

  View official source

• ISO 22095, chain of custody, published abstract
  Published standardPublished abstract read, full text behind a purchase

  The chain of custody standard, of which only the published abstract has been read. One sentence of that abstract is quoted on this page and nothing else is taken from it. The five chain of custody models the standard governs sit behind the purchase, this build did not buy it, and every freely available account of those models we found was a vendor page, so none of them is named here.

• Published traceability depth data, a peer-reviewed upstream pilot and a documented certificate fraud
  Case recordReached through a secondary reproduction, primary text not read

  Three separable items, each carried at the strength it survives. A brand whose central proposition is full traceability publishes its own depth figures, and the shape of them is carried here with that brand's own framing while the figures themselves are not restated. A peer-reviewed pilot that reached the farm and still concluded that transparency beyond the second tier of suppliers remains a significant challenge. And a documented fraud that defeated a certified chain using forged transaction certificates. The fraud is carried as a case and never as a rate. The most quoted fraud rate in this territory is an indicative figure more than a decade old and this page does not publish it.