Skip to content

Who has to own this, and what happens when a field turns out to be wrong

The company that placed the product on the market owns every field it publishes, and the accuracy standard sitting on those fields carries no materiality threshold and no reasonable efforts qualifier. The instrument that runs the passport registry contains no correction procedure, so the only one there is is the one the business writes for itself. Inside that business, the signature that counts is a record of what was checked, against what, on what date and by whom.

Sources as at
28 August 2026
Share
LinkedIn X Email
On this page

The short answer

The company that placed the product on the market owns every field it publishes, the accuracy standard the ecodesign framework sets on that data is unqualified, and the implementing regulation behind the passport registry contains no procedure for correcting or withdrawing a field that turns out to be wrong. Inside the company, the person who can sign is whoever can produce a record of what was checked, against what, on what date and by whom. Two boundaries before the detail. How the estate corrects its own work is a method statement about these pages, and this page is about the reader's governance rather than ours. Which economic operator carries the duty is a question about how duties fall between operators, and this one is about who inside a single business decides.

That combination is what makes the subject awkward rather than obvious. The duty is unqualified and it is continuous, the instrument that runs the register is silent on retraction, and the evidence under a typical field belongs to somebody else and moves without anybody being told. A business can do everything carefully and still be publishing a field whose support was withdrawn in June.

The part that is already settled

Responsibility outside the business is not in dispute. It sits with the economic operator that placed the product on the market, and working out which operator that is for a given product is a question with a settled answer and a page of its own. That question is answered by role rather than by company, and one business can occupy more than one role at once. Most people arrive here having already had that answer and still not knowing what to do on Monday morning.

What they arrive with is narrower and harder. A company is not a person, and the sentence that the company is responsible does not tell anybody whose name goes next to a recycled content percentage that a supplier supplied, that nobody in the room measured and that a document in a folder only half supports.

What the registry regulation says, provision by provision

The implementing regulation behind the passport registry is more precise on this than the summaries of it are, and the precision lives in four separate provisions. They do four different pieces of work. Merging any two of them is how a governance conversation goes wrong, usually in the direction of comfort.

Four provisions of the registry implementing regulation, kept apart, each named by what it does rather than by a number.
The provisionWhat it doesWhat it does not do
Responsibility at the moment of registration Puts responsibility for the accuracy and the completeness of the registered data on the operator, at the point of registration. It does not reach forward on its own. A point in time duty is discharged at a point in time, which is why a second provision exists.
The continuing duty A separate provision requires the data to be accurate, complete and up to date at all times. It sets no interval, no review cadence and no trigger. At all times is a standard rather than a schedule, and nobody is told when to look.
Registration by a third party Allows a third party to register on the operator's behalf, requires that third party to be verified in its own right, and leaves the operator fully responsible. It does not move responsibility to whoever does the keystrokes. Delegating the act delegates the act and nothing else.
Controller of the submitted data Makes the operator the controller of the data it submits. Controller status is a status. The instrument does not turn it into a correction route, and it answers no question about who signs.

One more sentence from the same read belongs next to those four, because it is the one most likely to be misread in the other direction. The automated checks the registry runs confirm structure and completeness, and they are not proof that a product complies with anything. A field can pass every check the register makes and still be wrong about the goods.

Why there are no article numbers above

The instrument was read at article level by one research programme in August 2026, and those four provisions are the substance of that read. The internal numbering has not been checked against the Official Journal PDF, so each provision is named here by what it does. The source note at the foot of the page records that check as outstanding.

Registration is the act that sets all four running at once, and what registration commits you to is set out separately, including what the register keeps and for how long. None of the four is about a field published on a website. They are about the data an operator registers, which is a narrower object and, once registered, a more consequential one.

Current at all times, and no act by which a wrong one is retracted

The framework's accuracy standard is unqualified. Accurate, complete and up to date, with no materiality threshold under which a small error stops counting, and no reasonable efforts qualifier that trades diligence for correctness.

The instrument that runs the register contains no correction, withdrawal, deactivation or supersession procedure. Change is handled through logging, versioning and a registration status whose permitted values the instrument does not define. So the law asks for a passport that is current at all times and supplies no legal act by which a wrong one is taken back.

Two things follow, and both are better stated without decoration. The first is that the correction procedure is the operator's, by default rather than by design: no instrument read for this page writes one, so whatever the business does becomes the procedure. The second is that what happens to a unit already sold when the passport behind it is corrected is not established, and we did not find an instrument that answers it on the routes we searched.

Nothing propagates

Underneath all of that sits one mechanical fact. When a document supporting a published field stops being good, nothing moves. There is no route by which the invalidation of a certificate reaches the field it was holding up, and nothing on either side of the join is watching for one.

Across the scheme documentation read for this estate, six schemes state their own limits in their own documents, and not one of those six operates automatic invalidation propagation. The nearest thing to a counter example is a scheme whose transaction certificates cite the input certificates behind them and carry an invalidation reason code. Even there the act is a certification body's discretionary decision, taken one node at a time, by somebody who has to notice first.

The clearest public test of it is a scheme's own published answer after an accreditation body suspended a certification body. Certificates already issued were still to be treated as valid unless the scheme notified otherwise. That is a defensible position for a scheme to hold. It is also a plain description of a system in which nothing downstream is told anything.

The data model side is no different. The public precedent registered as a source here grades where a value came from. That is useful. It holds no state at all for a value whose support has since been withdrawn.

The evidence lifecycle drawn in four stages, with the return path from a change in the evidence to the published field left empty. THE EVIDENCE LIFECYCLE Where evidence comes from Assertion, then the document that stands behind it What a value will support Scope, granularity and period read against the wording After you publish The field is live. Its evidence carries on moving without it How we know Review, correction and the history of both The evidence moves A certificate lapses. A standard is withdrawn. Two sources split. WHERE PROPAGATION WOULD BE No mechanism occupies this step Nothing read for this page asks that the published field be told.
The evidence lifecycle, with the propagation gap. The four stages are the four questions the evidence family is organised around, read left to right. The step that would carry a change in the evidence back to the live field is drawn as an absence because nothing occupies it: of the six schemes whose published rules were read for this estate, none operates automatic invalidation propagation, as at 28 August 2026.

The document half of this, which is what ends the useful life of each kind of paper and whether anything announces it, is set out in what stops being current, and when. This page is about the consequence for a field that has already been published.

A standard withdrawn in June, and nothing noticed

The worked example is live and it is ordinary. The self declared environmental claims standard has an edition that the catalogue records as withdrawn on 24 June 2026 and replaced by a new one. Every policy, supplier contract and technical pack citing the withdrawn edition now cites a document that no longer exists.

Nothing flagged that. No mechanism between a standards catalogue and a published product field would have. A recycled content field whose supporting declaration is written against that edition is in exactly the position this page describes: the field has not changed and neither has the declaration, and the thing both of them rest on stopped existing in June.

This page does not describe what the new edition contains. It is behind a purchase, it has not been bought and it has not been read here, so anything written about its contents would be invention with a reference number attached. Where the withdrawal is used above, what is being used is the catalogue entry rather than either text.

That is the shape of silent invalidation, and it is worth separating from the failure people expect. Nobody published something false. Something that made a published field defensible was withdrawn, and no channel existed to say so to anybody holding the field.

Four clocks that do not reconcile

Ask the simplest question anybody can ask of a published field. Was this true of this unit at the moment it was sold. Four different clocks bear on the answer and no system holds all four against a single unit.

  • The production date. One instant, fixed by the goods rather than by anybody's records.
  • The claim date. The day the field was published, which is usually later and sometimes much later.
  • The applicable period. The window the evidence itself says it covers, which is a property of the document rather than of the product.
  • The validity of the evidence. The clock running on the document, where the document has one at all.
Four clocks drawn against the life of one unit, with the moment of sale marked and none of the four reaching it. FOUR CLOCKS, ONE UNIT Production date One instant, fixed by the goods Claim date The day the field was published Applicable period What the evidence says it covers Evidence validity The clock on the document itself The moment of sale ONE UNIT'S LIFE None of the four reaches this line.
Four unreconciled clocks. The production date, the claim date, the period the evidence says it covers and the validity of the document itself, drawn against the life of one unit with the moment of sale marked. None of the four is aligned to that moment, which is why whether the field was true when the unit sold cannot be answered from what the four of them record. The validity drawn here is the one year clock a scope certificate carries, from the scheme rules read for this estate as at 28 August 2026.

The scheme documents make the mismatch sharper rather than softer. A scope certificate runs a one year clock and prints on its own face that it provides no proof that any goods delivered by its holder are certified. The transaction certificate, which is the document that does say something about goods, carries no expiry field at all. So the paper that expires is not about your unit and the paper that is about your unit does not expire.

No standard read for this estate reconciles the four, and none of the four is wrong. They are answers to different questions that happen to be written in dates, and the question a reader wants answered sits between them.

What a sign-off actually consists of

A sign-off on a published product field is not a name on a form. When the evidence belongs to somebody else, which is the normal case rather than the awkward one, the only thing a signature can honestly assert is what the signer checked and what the check could not reach.

A sign-off has five parts. What was checked, against what, on what date, by whom and what the check could not establish.

The fifth is the one that gets dropped, and it is the one that makes the record worth keeping. A sign-off that records only the positives is indistinguishable from a sign-off that looked at nothing, because both of them say the same thing about the parts nobody reached. A record that names its own limits can be inspected a year later by somebody who was not there, which is the whole point of writing it down.

Is this field ready for somebody to sign?

  1. 1Does the document behind the field say, on its own face, what it covers?

    • YesCopy that scope onto the record in the document's own words rather than in yours. Two schemes print a disclaimer on the face of a scope certificate saying it proves nothing about goods delivered by its holder, and that sentence is part of the scope.
    • NoRecord that it does not say. Supplying a scope the document never claimed is the commonest way a folder turns into evidence it is not.

    SourceChain of custody scheme rules, scope and transaction certificate specificationsChecked 28 August 2026

  2. 2Does that scope cover this product, this variant, this market and this period?

    • YesWrite the comparison down. It is the step nobody can reconstruct afterwards.
    • NoThe field is not supported by this document. A nearly relevant certificate supports nothing, and the honest output is a stated gap with its reason.
    • Not sureTreat it as unsupported until somebody establishes otherwise, and record which of the four you could not settle.

    SourceChain of custody scheme rules, scope and transaction certificate specificationsChecked 28 August 2026

  3. 3Do you know what ends this document's useful life, and would anybody tell you?

    • It carries a validity periodThat is the only kind of expiry a calendar can watch. A scope certificate runs a one year clock and is the worked case.
    • It carries noneRecord that there is no expiry rather than leaving the box empty. A blank expiry field and an absent expiry are different states, and the transaction certificate that does say something about goods has no expiry field at all.

    SourceChain of custody scheme rules, scope and transaction certificate specificationsChecked 28 August 2026

  4. 4If this field turns out to be wrong, who takes it down and where is that written?

    • A named person, in a written procedureSign. The record now carries what was checked and what happens if the check was wrong.
    • The supplier or the platformAuthorising a third party to register on the operator's behalf leaves the operator fully responsible under the instrument. The name has to be somebody inside the operator.
    • Nobody has decidedThat is the gap this page exists for. The registry instrument writes no correction procedure, so the operator's own is the only one in existence.

    SourceCommission Implementing Regulation (EU) 2026/1778 establishing the Digital Product Passport registryChecked 28 August 2026

This settles what your own record shows and what it admits it cannot show. It does not settle what the field is worth in law, and nothing in this estate certifies, verifies or validates any claim about any product.

None of that makes the field true. It makes the field accountable in the ordinary sense: somebody can reconstruct how it came to be published, on what basis and who would take it down. Where a business cannot answer those questions about a field it has already published, the field is not signed off, whatever the form says.

What to do in the next week

This is a procedure rather than a project, and it can be run against a single field first to see how long it takes. The field worth testing it on is the one that would be most expensive to be wrong about rather than the one with the tidiest folder.

  1. Establish where the fact is currently asserted. A published field is rarely in one place. The same percentage tends to sit in a product page, a marketplace feed, a tender response, a supplier contract and, once registered, in the register. Write the list before deciding anything, because the list is usually the surprise.
  2. Decide whether it is still supported. Take the document, read what it says it covers on its own face and hold that against the wording of the field. A document written against a withdrawn edition of a standard is a change of support even though nothing about the goods moved.
  3. Decide who takes it down. Name a person rather than a function. Authorising somebody else to register or to publish leaves responsibility with the operator, so the name has to be inside it.
  4. Record having decided. The date, the evidence, the decision and what the decision could not establish. Where the law supplies no procedure the record is the procedure, and it is the only artefact that will still exist when the people change.

One absence belongs alongside all of that, stated as an absence. No first-person account of a business amending live product information because upstream evidence changed was found by any of the research programmes behind this estate. We looked and did not find one. That is a fact about our search rather than about the market, and it does not tell us whether the problem is unserved or whether nobody minds it enough to write about.

You might want to read next

Since you have read this, these may answer the questions that usually come next.

Sources

Four of these rows carry a text somebody read and one carries a catalogue entry rather than a text. The difference matters on this page more than most, because the worked example in the middle of it is a document nobody here has bought.

  • Commission Implementing Regulation (EU) 2026/1778 establishing the Digital Product Passport registry
    CELEX 32026R1778In forceRelevant provisions reviewed

    The four provisions in the table above come from this instrument, read at article level by one research programme in August 2026. The internal article numbering has not been checked against the Official Journal PDF, so no article number is published on this page and each provision is named by what it does instead. That check is outstanding and it is the one thing that would change how these four are cited. The same read establishes the two absences the page rests on: the instrument carries no compliance deadline of its own and no correction, withdrawal, deactivation or supersession procedure.

    View official source

  • Ecodesign for Sustainable Products Regulation (EU) 2024/1781, consolidated text
    CELEX 02024R1781-20240628In forceRelevant provisions reviewed

    Here for one thing only, which is that the accuracy standard on passport data is unqualified: accurate, complete and up to date, with no materiality threshold and no reasonable efforts qualifier attached to it. Two independent reads of this instrument attributed the passport provisions to different articles, so this page names no article of it.

    View official source

  • ISO 14021, self-declared environmental claims, edition withdrawn 24 June 2026
    Withdrawn 24 June 2026Official source confirmed, detailed review pending

    The live worked example of silent invalidation. What is used here is the catalogue entry, which records the edition as withdrawn on 24 June 2026 and replaced. That is a fact about the catalogue rather than about any text. The replacing edition is behind a purchase, has not been bought and has not been read, so nothing on this page describes what it contains.

  • Chain of custody scheme rules, scope and transaction certificate specifications
    Scheme rulesRelevant provisions reviewed

    Carries the scope certificate disclaimer and its one year clock, the transaction certificate with no expiry field, and the six schemes that state their own limits in their own documents. Three of these documents carry effective dates in late 2026 and early 2027 and both changes alter what is said above. The scheme statement about certificates issued before a certification body was suspended is a public answer rather than a rule, and it is described here rather than named, because no address for it is registered.

  • Battery Pass Consortium data model
    IndependentRelevant provisions reviewed

    Read as one of the public precedents this estate positions against. Used here for a single sentence: it grades where a value came from, and it holds no state for a value whose support has since been withdrawn.

    View official source

Worth sharing?

Help someone else make sense of product passports.

LinkedIn X Email