What seven years of a compulsory product register actually produced
Before deciding what a Digital Product Passport will achieve, it is worth looking at the one Europe already built. The energy labelling register holds around two million product models, is reached mainly through a code on the product, and in June 2026 the Commission published its own assessment of it. The register was built and populated. Whether it is used, whether its contents are right and whether it changed any outcome are three separate questions, and the Commission answered only the first.
On this page
- The short answer
- What it is, in one paragraph
- What the Commission found in June 2026
- The label works. That is a different thing.
- The repair information, and what it actually is
- How accurate is a register that big
- What this is evidence of, and what it is not
- Why this is about to matter more
- What to take from it
- Sources
The short answer
The European Product Registry for Energy Labelling is the closest thing to a natural experiment this subject has. It is compulsory, it is public, it is reached through a code printed on the product, and suppliers have been required to populate it since 2019.
Its record, on the Commission's own numbers, is this. It exists and it is full. It is not widely known. Nobody has measured what anybody does after they reach it. And the parts of it that carry repair information are mostly a directory of links to somebody else's website rather than a repository.
That is not an argument that registers do not work. It is an argument about what evidence exists, and the honest summary is that seven years of operation have produced one usage figure and one admission that the behavioural question is unstudied. Anybody citing this register as proof that passports will change outcomes is citing the existence of infrastructure rather than evidence of its effect.
What it is, in one paragraph
Regulation (EU) 2017/1369 requires a supplier, before placing a unit of a new model on the market, to enter information in the register. Article 12 splits it in two: a public part carrying the supplier's identity, the model identifier, the label and the product information sheet, and a compliance part carrying technical documentation, accessible only to market surveillance authorities and the Commission. Information stays in the compliance part for fifteen years after the last unit is placed on the market. Article 16(3) lets a delegated act specify how a customer reaches it, and it is permissive rather than prescriptive: a website address, a dynamic quick response code, an online label link or any other appropriate consumer oriented means.
That two part structure, a public face and an authorities-only layer behind it, is the same shape the passport architecture is being built in, which is set out at where your passport data actually lives.
What the Commission found in June 2026
The energy labelling framework carried a review clause requiring the Commission to assess implementation and report. The report arrived in June 2026, together with an impact assessment for revising the framework. Between them they contain the first official answer to whether this register is working.
On scale. The Commission describes the register as populated with around two million product models. Its own frequently asked questions note the count is cumulative and includes models no longer placed on the market, so it is a count of registrations rather than of products on sale.
On usage. The report states that the database was visited 6.6 million times in 2025, and that it is accessed mainly by way of the codes on labels rather than through the website. Those two sentences sit next to each other, and the second is asserted without a number. There is no published count of code scans anywhere we could find, which is worth pausing on: the Commission operates the endpoint that every scan reaches.
On awareness. The report states that survey data and feedback from national authorities confirm the register is not yet widely known by the public or by public procurement buyers. The impact assessment describes an underutilisation of the registry.
On behaviour, which is the finding that matters most. The impact assessment says that many labels are not seen by a customer until after the purchase decision is taken, since they remain in the box, and then states:
No studies could be identified on the behaviour of customers after purchase to understand what they do with the energy label.
The Commission's own implementation report, quoted verbatim.Seven years in, on a compulsory register with a public interface, the behavioural question is unstudied. That is the Commission's own account, not a critic's.
On the verdict. Its conclusions put it plainly: the register allows for more market transparency and evidence based policy making, but is not yet as widely known or used as the label itself and could be used more to support compliance and market surveillance.
The label works. That is a different thing.
The contrast inside the same evidence base is the most useful part of it, and it is routinely collapsed.
The energy label itself has strong measured uptake. The impact assessment records that buyers of white goods were most likely to have used the energy label when making their purchase decision, at around 90 per cent of those who saw the label. That denominator is the interesting part, because the same document says many labels are not seen until after the purchase decision is taken, since they remain in the box. The European Court of Auditors reported around 85 per cent of Europeans recognising and using the label when making a purchase.
The register behind the label does not have a comparable figure, and the Commission says so.
So the thing that changes a decision at the point of sale is a printed comparative label with a letter on it. The thing that holds the underlying data is infrastructure that mostly nobody visits. Those are two different objects with two different jobs, and a claim that one works is not evidence about the other.
That distinction is the whole of the argument set out at what has to be true before a passport changes a decision, and this register is the closest thing to a test of it that exists.
One number here needs handling carefully. 6.6 million visits in a year, against a register of around two million models, is not a scan rate and must not be converted into one. The Commission publishes no definition of a visit, no unique user count and no split between arrivals from a code and arrivals from the web. What a scan does and does not measure, and why this estate publishes no universal rate, is set out at what a scan actually tells you.
The repair information, and what it actually is
This is where the register comes closest to the circularity claims made for passports, and the detail is instructive.
Smartphones and tablets are the confirmed case. Under the 2023 ecodesign and labelling instruments for that group, a model record carries a repairability class and index, six sub scores covering disassembly depth, fasteners, tools, spare parts, software updates and repair information, and three weblinks: to spare parts availability, to repair instructions for end users and to indicative pre-tax prices.
Read the last three again. They are links. The underlying ecodesign obligation in Regulation (EU) 2023/1670 requires repair and maintenance information to be publicly available at the free access website of the manufacturer, importer or authorised representative. The register holds a pointer to that site. It does not hold the information, and we found no published check of whether those links resolve to anything useful.
The Commission's own description of which product groups carry this information says it applies to some groups whose legislation has been recently reviewed, and names none. We confirmed smartphones and tablets by opening a record. We could not establish which instrument puts dismantling information for recyclers into the register, or find a product group where it appears, which matters because that is the limb most often cited when this architecture is offered to the recycling question. What a recycler can actually use is worked through at what a recycler can actually use.
And the outcome question is unanswered. We looked for any study, official or independent, measuring whether publishing spare parts lists, prices or repair instructions has changed repair rates, spare part availability, spare part prices or product lifetimes in the Union. We looked at the Commission, at its research service, at the Court of Auditors and at the main repair advocacy organisation. We did not find one. The appliance rules have been in application since 2021 and the smartphone rules since 2025. That is a statement about where we looked rather than a claim that no such study exists anywhere.
How accurate is a register that big
Data quality is the question a passport programme should care about most, and here the evidence is real but has to be read carefully.
The largest denominator available is reassuring. An ongoing project examined 25,637 oven models registered in the database and found 204 of those 25,637, which is 0.8 per cent, declaring an incorrect energy efficiency index. That tests one parameter rather than a whole entry.
Smaller checks point the other way. In one exercise, 67 products were checked for registration and 8 of those 67 were fully correct and complete. In a separate pilot across 71 models, 8 of those 71 had errors or were not registered. Those two results are not comparable exercises and are reported here with their own populations for that reason.
The wider surveillance picture carries a caveat that has to travel with it. A concerted action involving 24 national authorities, described by the Commission as a non-representative risk based exercise across six product groups, found 82 per cent of products in stores and 97 per cent in online shops had minor or major issues around label display, declaration of conformity or technical documentation. Risk based means the products were selected because they looked suspect. Those figures do not describe the market and should never be quoted as though they did.
The volume of formal checking is the number that puts the rest in proportion. Member States encoded 3,147 cases of in depth inspections since 2017, of which 57 per cent were non-compliance, against a register holding around two million models. Inspection capacity is also uneven: the Commission reports annual inspection totals under this framework ranging from between 10 and 100 products in some Member States to 24,000 in one and 73,000 in another in 2024.
What this is evidence of, and what it is not
| The claim | What this register supports |
|---|---|
| A compulsory public product register can be built and populated at scale | Yes. Around two million model entries |
| A code on the product is how people reach it | Partly. The Commission says access is mainly by code, and publishes no number for it. On its own figures the code is live for 10 of about 16 labelled product groups |
| Registers get used by consumers | Not established. One visit figure, no unique users, an official finding that it is not widely known |
| Registers change purchase decisions | Not established for the register. Measured for the label, which is a different object |
| Published repair information improves repair | Not established. No study found by us, in either direction |
| Register contents are accurate | Mixed and mostly unmeasured. One large denominator check on one parameter found 0.8 per cent wrong. Everything else is small or risk based |
| Registers support market surveillance | Partly, and thinly resourced. 3,147 in depth inspections since 2017 against around two million models |
How this estate types each of those states, and why an absence is not a negative, is set out at how we know.
Why this is about to matter more
The June 2026 proposal to revise the energy labelling framework includes a technical link between this register and the central element of the future registry under the ecodesign framework. In other words the two systems are to be connected.
That has a consequence for anybody reading claims about passports. When the connection is made, the register with seven years of operating history and the register with none will be discussed as one thing, and the evidence base of the older one will be cited for the newer one. This page exists so that the citation can be checked.
What to take from it
- Ask what a claim is evidence of. Infrastructure existing, infrastructure being used and infrastructure changing an outcome are three claims. This register supports the first, partially supports the second and does not yet support the third.
- Do not accept label evidence as register evidence. The 90 per cent figure belongs to a printed label, and to the buyers who saw one. It does not transfer.
- Treat risk based samples as risk based. Two of the most quotable numbers in this area describe products selected for looking suspect.
- Expect a link directory rather than a repository. The most repair relevant fields in this register are weblinks to somebody else's site, and we found no published check on whether those links still resolve.
- Watch what gets measured next. The behavioural question has been open for seven years and the Commission has just said so in writing. A revision is the moment it could be closed, and whether anybody closes it is the thing to watch. What a filled field can and cannot establish about a product is set out at what a passport field can and cannot prove.
You might want to read next
Sources
-
Relevant provisions reviewed
The scale figure, the 6.6 million visits in 2025, the statement that access is mainly by code, the finding on awareness, the inspection counts and the conclusion. Read at a parliamentary document repository copy, because the EUR-Lex rendering of this document returned no content to us on 28 August 2026. That is recorded rather than hidden, and the document reference should be checked at the Official Journal before republication.
-
Relevant provisions reviewed
The statement that no studies could be identified on customer behaviour after purchase, quoted. Also the underutilisation finding, the 10 of 16 code coverage, the white goods label figure and the oven check with its 25,637 denominator.
-
Relevant provisions reviewed
Articles 4, 12 and 16(3) and Annex I, for what suppliers enter, the two part structure, who reaches which part and what a delegated act may specify as the route to it. Paragraph numbering within Article 12 was read through a retrieval tool and should be confirmed at the source before republication.
-
Reviewed in full
The product groups covered, the cumulative nature of the model count, and the Commission's own unspecific description of which groups carry repair information.
-
Relevant provisions reviewed
The label recognition figure, and the fact that the register was audited before it was operational, so it evidences delay rather than outcomes.
-
Relevant provisions reviewed
The registration accuracy checks with their own populations. Every figure from this source is carried with the Commission's own description of the exercise as non-representative and risk based.
-
Relevant provisions reviewed
One proposition: that repair and maintenance information must be publicly available at the manufacturer's own free access website, which is what makes the register a directory rather than a repository for this group.
Help someone else make sense of product passports.