What to do now, and what to refuse to do now

What the framework actually establishes, and what it leaves open

The Ecodesign for Sustainable Products Regulation is in force. Article 9 establishes the passport and states that information requirements are to provide that products can be placed on the market only if a passport is available in accordance with the applicable delegated acts, and that the data in it is to be accurate, complete and up to date. Article 10 sets the essential requirements for the passport itself. Article 11 governs its technical design and operation.

Read Article 9 carefully and the whole timing question resolves. The duty is conditional on a delegated act existing for the product group. For textiles, none has been adopted, none has been published in draft, and the preparatory work behind one has a milestone whose timing the Commission records as still to be communicated.

Two further things follow from the same provision, and they are the reason this page exists rather than a timeline.

The delegated act decides the content, the carrier, the level, who may write to the record and who may read it. All of that arrives in one document, at one time. Anything a business commits to before that document exists is committed against somebody's guess at it.

And the framework separately sets a floor: requirements apply no earlier than eighteen months from a delegated act entering into force, except in duly justified cases.

The Commission does publish a timing for the textile act, and the words it publishes it under are the point. It appears under a heading calling the timeline indicative, it is described as a planned adoption rather than a date of application, and it carries a note that implementation timelines may evolve as legislative and technical work progresses. A planned adoption plus a floor is arithmetic rather than a deadline, and it is arithmetic performed on a number its own author says may move. This estate publishes the working and not the result. What is adopted, what is in force, what is planned and what has only been signalled is maintained on the status tracker, with the Commission's own framing kept intact.

None of that is an argument for doing nothing. It is an argument against sequencing by a date, because a plan built on a date that moves has to be rebuilt when it moves, and a plan built on reversibility does not.

The four decisions you cannot undo

Each of these is made once, made early, and cannot be walked back once goods carrying it are in circulation.

What is printed on the goods

A carrier printed onto a swing tag or a sewn-in label is a physical object with a print run behind it and a garment in front of it. The garment is in somebody's wardrobe years later. Ink, geometry and the number inside are fixed at the moment the label is made, and nothing about the printed square can be edited afterwards.

The part of that decision that matters least is which symbology is fashionable this year. The part that matters most is what address the code carries, because the address has to keep answering for as long as the goods exist.

Whose domain is in that address, and whose account holds the identifier

Both are ownership questions dressed as configuration, which is why they get decided by whoever is setting up the system rather than by whoever will live with the consequence.

If the domain printed into the code belongs to a supplier, the supplier controls whether anything answers a scan of your garment. If the identifier licence sits on a provider's account rather than yours, the licence is theirs, the annual renewal decision is theirs, and the identity printed on your goods depends on a payment you do not control. Every persistence commitment this estate has read carries a condition, and the condition is usually a renewal.

Neither is revisitable after the goods ship, which puts both in a category with your company name rather than with your software.

Committing to a level by printing one

The framework names three levels a delegated act may choose between, and none has been chosen for textiles. Printing a serialised carrier commits you to the most expensive of the three before anybody has decided it is required, and the commitment is physical rather than contractual.

The counts involved are not a modelling preference either. Under the numbering scheme's own management standard, each style, colour and size variation is a distinct identifier, so one shirt in five sizes and six colours is thirty identifiers whatever anybody thinks about passports. Which of the three levels a passport eventually sits at, and why the gap between them runs to three orders of magnitude on an ordinary range, is set out on how many passports a range needs.

Putting the wrong kind of value into an identifier field

An internal stock code written into a barcode field is the most common identity defect in this subject. It is free to fix today and it is not fixable at all once it has travelled into artwork, into a channel feed and onto a garment, because a corrected identifier passes the arithmetic and may now name somebody else's product. What allocating an identifier commits you to, and why two identical catalogues can face different bills depending on where the business is incorporated, is on identifiers.

The work that expires, which is the only genuinely urgent part

Everything above is about not deciding too early. This is the opposite, and it is why waiting is not free.

Some facts about a product exist only at the moment it is made and cannot be recovered afterwards from anything anybody kept. If they are not captured then, they are gone, and no software bought later reconstructs them.

Which production run a variant came from. Reading upward is easy: from a batch you can always name the model. Reading downward is the direction a claim needs, and from a model record, or from a garment in a customer's hand, you cannot say which run it came out of unless something was captured at the time. A certificate covering one dyehouse, a recycled input used in one run, a finish that met a restriction on four colourways out of six: each is a fact that is true of part of a style and has no home unless the join was recorded when it existed.

What a document actually covered. A certificate arrives, is filed, and the scope printed on its face is never written down beside it. A year later that scope is a matter of somebody's memory. Writing the coverage onto the row when the document arrives takes a minute and cannot be done later without reading every document again.

What is attached to the fabric. Trims, coatings, laminates and the elastane fraction are known at the point of design and forgotten within a season. Nobody currently asks a brand to record them, and they are the material every downstream process turns on.

What you asked, of whom, on what date. Without a request log an absence cannot be typed at all. There is no way to tell a fact nobody has looked for from a fact somebody looked for and did not find, and those two license completely different sentences. It is the cheapest record to start keeping and the only one whose past cannot be recovered.

That is the honest cost of waiting, and it is not a licence fee. Most of what you defer here you can do later at broadly what it costs today. These four get more expensive every season they are left, and some of what they capture stops being recoverable altogether. Their value does not depend on which fields a delegated act eventually names.

The obligation that arrives with the passport and nobody prepares for

One provision is worth knowing about in advance because it creates a workflow rather than a data set, and workflows take longer to build than fields do.

Article 10 requires the economic operator placing a product on the market to give dealers and providers of online marketplaces a digital copy of the data carrier or the unique product identifier, so that they can make it accessible to customers who cannot physically handle the product. It requires that copy, or a webpage link, to be provided free of charge and in any event within five working days of a request.

That is a service level, not a field. It applies once passports exist for the goods concerned, and it lands on whoever answers the phone to a wholesale customer or a marketplace. A business selling through its own store will barely notice it. A business selling through fifty stockists will need a route from an incoming request to a correct answer in five working days, and that route is a process, an owner and a record rather than a metafield.

Nothing requires it yet. It is on this page because it is the one future obligation whose shape a business can usefully know now, and because building a request-and-answer process is exactly the kind of work that survives whatever the delegated act says.

The afternoon's work that survives every outcome

Five things. Four cost an afternoon each and none requires buying anything.

Find out which markets you place goods on, and record it once for the business. One answer settles the applicability question underneath several duties at a stroke. Shipping settings are not that answer: a shop that will post to Belgium has not thereby placed goods on the Belgian market.

Find out which role you occupy in each of them. Manufacturer, importer, distributor and dealer are defined terms carrying different duties, one business can occupy different roles in different markets, and the role decides the duty and does not change when an act arrives.

Look at what your own product data actually describes. This is the largest of the five and the one that changes every later estimate, because most of the difficulty here is not collecting new values but establishing what the values you already hold are about. Where they live, and the four questions that decide whether one is usable, is worked through on the product data you already have.

Structure the attributes that are already governed. Fibre composition is regulated now, by an instrument in force, independently of any passport. It is also one of the harder values to reconstruct later and one of the easier ones to fix today.

Write down what you cannot answer, and why. A recorded absence with a reason is a usable artefact that somebody else can pick up. A blank cell is a question nobody knows has been asked.

A sixth item costs more than an afternoon and still belongs here, because it does not get cheaper by being deferred. Asking your suppliers once, properly, is a different exercise from sending a questionnaire, it produces a record of having asked even when nobody answers, and the one statutory route compels rather less than most people assume. It is set out on getting evidence out of suppliers.

What to refuse to do yet

The shorter list, and the one a vendor will not write for you.

Do not print anything that assumes an answer. A serialised carrier commits you to a level nobody has chosen. A code carrying somebody else's domain commits you to their continued existence. Printed decisions do not get revised.

Do not buy a platform against a requirement that has not been written. The fields, the carrier, the level and who may write to the record all arrive together in one document. A system selected before that document exists is selected against a guess. Data structured for its own sake survives whatever the act says; a system chosen to satisfy a guess may not.

Do not commission a full life cycle assessment because a passport might want a number. Nothing requires an environmental figure on a textile product today, and the proposal that exists permits secondary data and default values rather than requiring measurement. The composition and the weight underneath any such figure are worth fixing first, because every modelled figure is built from one and divides by the other.

Do not buy item-level identity as compliance. It may be a good idea for anti-counterfeit work, for warranty or for a service you want to offer, and those are real arguments to be made on their own merits. What item-level identity is not, today, is a requirement, and a proposal presenting it as one is describing a decision nobody has taken.

Do not treat a marketplace as the deadline. Nothing read in any platform or marketplace rule requires passport data as a condition of listing, and the largest supplier manual this estate has read describes a passport as expected rather than obligatory across three versions of the document.

Two duties that do already reach a clothing business

Neither is a passport duty and both get lost in conversations about one.

A prohibition on destroying unsold apparel, clothing accessories and footwear is already in application, made under the same framework. It is graded by the size of the business, which is the one place in this architecture where size changes what is owed, and it is the most plausible origin of a belief this estate meets constantly: that small businesses will be carved out of the passport. Nothing in the passport provisions carries a size test of any kind.

A second instrument reaches any business making environmental claims to consumers in the Union, with no size relief at all, and its date is close. For most brands the work it creates is a copy and packaging change rather than an evidence project, and it is considerably cheaper to do before it starts than after.

Both dates, and which instruments are in force, adopted or merely indicated, are kept on what already applies today. This page carries no date of its own, deliberately, because a page about sequencing is the easiest place in an estate for a date to be read as a deadline.

Four questions any quotation should be able to answer

Each has a checkable answer rather than a persuasive one, and each works against anything this business puts in front of you as well as anybody else's.

Which level does the price assume, and which document says so. The unit being counted is styles, sellable variants or garments, and those differ by three orders of magnitude on the same range. A quotation that cannot say which it counted is not a quotation.

Which of these charges recur, and which stop being optional once goods carrying a code are in circulation. The published costings this estate has looked at price a first year. Very little of the real work has the shape of a first year.

Whose account holds the identifier licence, and whose domain is printed on the garment. Both answers should be yours. If either is not, the reason should be written down before anything is printed.

Which half of this quotation is data work. Tidying, deduplicating and attributing product data is not a new market and has had a published price for years. A reader being quoted for a passport programme is partly being quoted for that, and asking which half is which is fair and answerable.

What would change this page

A draft or adopted textile act, at which point the reversibility analysis is replaced by a requirement and this page is rewritten around it rather than defended.

A published requirement from a marketplace or a large buyer making passport data a condition of supply, which would be the first genuine commercial deadline in this subject and which this estate has looked for and not found.

A decision on the level at which a textile passport sits, which is the single input that most changes what is safe to print.

Sources

• Ecodesign for Sustainable Products Regulation (EU) 2024/1781, consolidated text
  In forceRelevant provisions reviewed

  CELEX 32024R1781. In force. Read in full at the Official Journal HTML rendering, 28 August 2026. Read for four propositions: that Article 9 makes the passport duty conditional on a delegated act adopted for the product group; that a delegated act specifies whether a passport is established at model, batch or item level; that Article 10 requires a digital copy of the data carrier or the unique product identifier to be provided to dealers and marketplace providers free of charge and in any event within five working days of a request; and that requirements apply no earlier than eighteen months from a delegated act entering into force, except in duly justified cases. Article numbers are cited on the strength of the direct read, corroborated by a Commission implementing decision published in the Official Journal which names Articles 10 and 11 as the articles carrying the passport requirements.

  View official source

• European Commission, Digital Product Passport, textile apparel
  Institutional statementReviewed in full

  Institutional material. Read at the Commission's own page, 28 August 2026. Behind one proposition and its framing: that the Commission publishes an implementation timeline for the textile act which it labels indicative, describes as a planned adoption of the delegated act, and footnotes as liable to evolve as legislative and technical work progresses. The date itself belongs to the status tracker and is not stated on this page. The page is also the source for the Commission's own statement that textile-specific passport requirements will be defined through the same future delegated act that introduces the ecodesign requirements.

  View official source

• Joint Research Centre product group project plan and document list for textiles
  Pre-decisional researchReviewed in full

  Pre-decisional. Project plan and document list read at their own address. Behind one proposition: that no draft delegated act for textiles has been published for consultation, that the third preparatory milestone closed in March 2026 and that a fourth milestone on policy scenarios and elements to be included in the passport is listed with its timing still to be communicated.

• Commission working plan COM(2025) 187
  IndicativeRelevant provisions reviewed

  Indicative. Relevant provisions reviewed. Names textiles as a first priority and gives an indicative adoption year. An indication in a programming document rather than an adopted instrument. The year itself is stated on the status tracker and appears nowhere on this page.

  View official source

• GS1 GTIN management standard
  Scheme rulesRelevant provisions reviewed

  Scheme standard. Relevant provisions reviewed. Read for the allocation rule whose worked example makes each style, colour and size variation a distinct identifier. It binds because catalogues and selling channels apply it, not because any instrument requires it. The general specifications governing whether an identifier already printed onto goods may be reused are unread here, so this page asserts nothing about reuse in either direction.