ESPR is the Ecodesign for Sustainable Products Regulation, and it is the law that creates the Digital Product Passport. It entered into force on 18 July 2024 and it applies to almost every physical product sold in the EU. There are plenty of legal summaries of it. There are almost none that answer the question a product data team actually has: which attributes, at what granularity, held in which system. That is what this piece covers.
What ESPR actually is
Regulation (EU) 2024/1781 replaces the old Ecodesign Directive, which only covered energy-related products. ESPR extends the same machinery to virtually all physical goods, with a short exemption list that includes food, feed and medicinal products.
It is a framework, not a rulebook. ESPR itself sets almost no product requirements. What it does is give the Commission the power to set them, product group by product group, through delegated acts. Annex I lists the parameters those acts can address. Durability, reliability, reusability, upgradability and repairability. Presence of substances of concern, energy and water use, recycled content and recyclability. Environmental impacts, including carbon footprint.
Every one of those parameters becomes attributes on your product records. That is the whole story, compressed. The rest of this article is what that costs.
Why ESPR does not tell you what to do yet
Two mechanisms control the timing, and both are widely misreported.
First, the delegated acts. Nothing binds you until the act covering your product group is adopted. As of August 2026, none of the priority ESPR product groups has an adopted delegated act. Every published field list you have seen for textiles or furniture is a proposal.
Second, the grace period. Article 4 says a delegated act cannot apply earlier than eighteen months after it enters into force. There is an exception for duly justified cases. So an act adopted in 2027 typically bites in 2029.
The Commission adopted its first working plan (COM(2025) 187) on 16 April 2025. It sets indicative years for adoption, not for compliance.
| Product group | Indicative adoption year |
|---|---|
| Iron and steel | 2026 |
| Aluminium | 2027 |
| Textiles, with a focus on apparel | 2027 |
| Tyres | 2027 |
| Furniture | 2028 |
| Mattresses | 2029 |
| Repairability, including a score (horizontal) | 2027 |
| Recycled content and recyclability of electrical and electronic equipment (horizontal) | 2029 |
A mid-term review is scheduled for 2028, so these can move. Some passports arrive under other laws entirely. Battery passports come from the batteries regulation and apply from 18 February 2027. Construction products and toys have their own instruments and their own timetables.
The part nobody writes about: what ESPR does to your data model
This is where the legal summaries stop and the real work starts. We break the requirement into four attribute families, because they behave differently and belong in different places.
Which attributes
Identity and actor data. A persistent unique product identifier, model and batch identifiers, a facility identifier and a commodity code. Plus the legal manufacturer as an entity, not a brand string. Most catalogues hold some of this. Almost none hold it as controlled references. If your commodity codes and your taxonomy disagree, fix that first, because the registry stores both. Our notes on product classification cover the usual mess.
Composition and substances. Material breakdown as percentages, recycled content by material, and substances of concern. The textiles proposal asks for substances identified, located within the article, and quantified by concentration. Note “located within the article”. That single phrase breaks the flat attribute model most PIMs run.
Performance and lifecycle. Durability and reliability test results, repairability, spare parts availability, carbon and environmental footprint figures. These are derived from testing, not authored by a content team.
Evidence. Declarations of conformity, test reports, chain of custody certificates. Documents, with versions and validity dates, linked to the product record rather than filed by supplier.
The second family is the one that hurts. Composition is not a product-level attribute. A dining chair has a frame, a foam, a cover and a fixing kit, and each has its own material breakdown. You need a component structure with repeatable attribute groups, not four more columns. Anyone who has modelled a furniture catalogue properly already knows how much that changes.
At what granularity
The Digital Product Passport registry implementing regulation (Commission Implementing Regulation (EU) 2026/1778, in force 6 August 2026) is explicit. Passports register at the granularity the applicable law specifies: model, batch or item. Where you register at item level, you must also link the batch and model identifiers.
Most catalogues have exactly one level. That will not hold, and the reason is simple. Different attributes change at different rates.
Durability class and care instructions are model level. They are stable across a production run. Recycled content percentage is batch level, because a supplier switching resin changes it without changing the SKU. Carbon footprint is batch level for the same reason. Serial number and health status are item level, which is why battery passports need serialisation and a T-shirt does not.
The rule we give clients is short. Put an attribute at the lowest level at which its value can legitimately change. Decide that before anyone buys software, because it determines whether you need a lot layer at all.
Getting this wrong is expensive in a specific way. If you hold batch-variable data at model level, you are publishing a number that is wrong for most units. That is worse than publishing nothing.
Held where
The honest answer is that no single system holds a passport. It is assembled. The split we recommend looks like this.
The PIM holds identity, classification, model-level attributes, published content and the access-rights view. It is the system of record for the passport as a published object, which is the role it is actually good at. This is what PIM solutions for DPP means in practice, and it is a narrower claim than most vendors make.
The ERP, MES or quality system holds batch data: lot genealogy, production facility, test results, batch-variable composition. This data is created by operations and moves on a manufacturing cadence, not a merchandising one. Trying to master it in the PIM fails.
Supplier-declared data comes in through a validated intake, not email. Composition, substances of concern and evidence all originate outside your business. If your suppliers currently send PDFs, that is a supplier data onboarding programme. It is the longest single lead time in the whole compliance effort.
Documents live in a document store, but linked to the SKU with version and expiry, not filed by supplier and date.
The anti-pattern is obvious once stated. Buy a standalone passport platform, point it at a spreadsheet, and you have a QR code that resolves to data nobody maintains. We see the same shape in every rushed compliance project.
The data carrier is the easy part
Article 10 requires the passport to connect through a data carrier to a persistent unique product identifier. The carrier must be physically present on the product, its packaging or accompanying documentation. Data has to be machine-readable, interoperable and based on open standards.
That question is now largely settled by standards work. CEN-CENELEC’s JTC 24 has published a family of DPP standards. They cover unique identifiers, data carriers, data exchange, storage, APIs, interoperability, access rights and data authentication. Commission Implementing Decision (EU) 2026/1736 of 14 July 2026 cites the harmonised versions in the Official Journal.
So the carrier is a solved problem with a known answer. The data behind it is not. We would not spend a single planning session on QR codes before the granularity question is settled. More on the sequencing in our DPP fundamentals material and on the Digital Product Passport overview.
What ESPR means for a UK business
Three things, and only the third one is urgent.
ESPR obligations follow the product, not the company. If you place goods on the EU market, you are in scope when the delegated act for that group applies, wherever you are registered. If you sell through an EU importer, that importer carries the legal obligation, but the data has to come from you. You will be contractually on the hook long before you are legally on the hook.
Great Britain has no equivalent requirement. The government opened a call for evidence on digital product records policy on 27 July 2026, which closes on 21 September 2026. No UK scheme has been decided. Northern Ireland is different: EU DPP rules apply there under the Windsor Framework.
The urgent one is commercial. Your EU customers and marketplaces will start asking for these fields as a condition of listing. They will do it ahead of any legal deadline, because they are managing their own exposure. That is the date that will actually hit your team.
What is confirmed and what is not
Confirmed as of August 2026: ESPR is in force. The registry exists in law and Member States must appoint national administrators by 18 February 2027. The DPP standards are published and partly harmonised. Battery passports apply from 18 February 2027. The ban on destroying unsold clothing, clothing accessories and footwear applied to large companies from 19 July 2026.
Not decided: every product-specific field list under ESPR. No delegated act has been adopted for textiles, furniture, tyres, aluminium or mattresses. Anyone quoting you a definitive attribute schema for those categories is quoting a proposal.
That distinction matters for how you spend. Build the structural things now, because they take years and hold their value whatever the delegated acts say. Identifiers, granularity, classification and supplier intake are all useful on the day they are finished, regardless of Brussels.
Where to start
- Decide your granularity model (model, batch, item) per category. This is the decision everything else depends on.
- Move composition out of description text into structured, component-level attributes.
- Fix the disagreement between your commodity codes and your taxonomy.
- Score your suppliers on how much of the composition and evidence data they already send as data.
- Do not buy a passport platform until the four points above have answers.
We run a fixed-scope readiness audit against a live catalogue and hand back a scored result with a remediation sequence. Thirty minutes on a call is enough to work out whether it applies to your categories. Talk to us, or read the Digital Product Passport overview first.