Whitepaper · DPP · 2026

Building Digital Product Passports with external data

Digital infrastructure needs, challenges and solutions for turning supplier information into reliable, attributable DPP data – before the passport goes live.

10chapters
8diagrams
~35min read
00Introduction

DPPs don't introduce data exchange. They change what you must do with it.

The EU framework expects product information that is interoperable, machine-readable, structured and searchable. The EU DPP Registry has been operational since July 2026 – yet much of the information behind a passport originates outside the operator that is responsible for it.

0%

of operations and supply chain leaders rank data availability and quality among the top three challenges to scaling AIPwC via WEF · 2025

0%

of COOs identify data quality issues as their most significant data priorityIBM Institute for Business Value · 2025

$5M+

annual cost of poor data quality, estimated by over a quarter of organizationsIBM Institute for Business Value · 2025

0%

of supply chain leaders cite lack of trust or confidence in data as a top hurdleGartner · 2023

0%

report moderate to significant operational impact from duplicates, inaccuracies and no real-time verificationMelissa · State of Enterprise Data Quality 2025

0

data points defined for EV, LMT and industrial batteries in the Digital Batteries Passport guidanceEuropean Commission · Aug 2026

DPP interoperability does not eliminate the cost of data exchange – it determines where that cost is incurred.

01From information to DPP data

One fact. Four representations. One target model.

External information rarely arrives "DPP-shaped". Before it can be exchanged efficiently, it has to become usable within the receiving organization's own data model.

A supplier reports recycled content as
0% PCR PET

A certificate expresses it as a percentage by mass. A supplier DPP exposes it as 0.374 kg/kg. The receiving DPP needs a defined percentage field, linked to a specific material, component or product level. Same fact – not interchangeable without interpretation and mapping.

Macro of recycled PET yarn
ERP · PET-PCR 37.4%DPP · 0.374 kg/kgPROC · RECYCLED_PET = TRUE
Fig. 1 · Heterogeneous sources converge into a target data modelScroll to animate
SUPPLIER ERP PET-PCR37.4% CERTIFICATE Post-consumer recycled PET:37.4% by mass SUPPLIER DPP Recycled content:0.374 kg/kg INTERNAL PROCUREMENT SYSTEM RECYCLED_PET= TRUE Target Data Model
The first layer is not integration

It is data preparation. Plugging a source into a DPP system is only useful when the value has a defined, reliable place in the target model.

01

Identify

What information is actually being provided?

02

Interpret

What exactly does the value mean in the source's context?

03

Structure

Which field or data object should contain it?

04

Map

How does the source representation correspond to the target?

05

Validate

Is the resulting value technically and contextually valid?

06

Associate

Which product, variant, batch or entity does it describe?

Fig. 2 · From external origins to public, regulator and B2B viewsLive data flow
EXTERNAL ORIGINS ORGANIZATION'S DATA INFRASTRUCTURE PREPARATION AUDIENCES Public registers/ standards Certifications/ laboratories Supplier 1DPP Supplier 2DPP DPP platform DPP Public view Authorities /regulators B2Bdownstream orgs
02The DPP data model

Not a flat list of fields. A three-dimensional scope of work.

Which entities are described, which attributes belong to them, and how the resulting information is accessed and governed. The standardized layer defines required data points; the organization configures everything around it.

Fig. 3 · Entities · attributes + evidenceHover a level
PRODUCTSensor Model X
Product identifierManufacturerCategoryDPP type
VARIANT24 V / steel
Model-specific dataMaterials ⎘Components ⎘
BATCH2401
Production dateProduction siteSupplier material batchesTest results ⎘
ITEMSerial #1235
Serial-specific statusRepair history ⎘End-of-life status
Anatomy of an attribute
[ ATTRIBUTE ]
valueunitmeaningscopesource
Roles decide who sees what
Public
Regulator
Partner
  1. What entity does it describe?
  2. What exactly is the attribute – type, unit, range?
  3. When does it apply?
  4. What supports it – declarations, certificates, lab results?
  5. Who can access or update it? (prEN 18239:2025)
03DPP data validation

A value can be syntactically correct – and still be wrong.

Attached to the wrong product, interpreted incorrectly, unsupported by evidence, or contradicted by another source. Validation is a progression from syntax to cross-source reconciliation.

Incoming value · Supplier X
37.4 % recycled PET
→ Material X / batch 1842
  • Format & typewaiting
  • Placement in modelwaiting
  • Meaning in contextwaiting
  • Supporting evidencewaiting
  • Agreement across sourceswaiting
Level0

Syntactic

Is the value in the correct format; is a number actually a number? Highly deterministic, easily automatable, reinforced by the DPP framework's structure and controlled vocabularies.

Fully automatable
Level1

Structural

Is it placed in the correct part of the model? A batch production date must not become a property of the product model. Database constraints, referential integrity, hierarchy and cardinality rules.

Automatable with a clear model
Level2

Contextual

Does "37.4% recycled PET" describe the PET itself, the finished component, or something else? Vocabularies, semantic mappings and calculation rules help – human supervision often still needed.

Rules + human supervision
Level3

Evidential

Is the document present, valid, unexpired, matched to a known record, digitally verifiable? Finding evidence is not the same as deciding it is sufficient.

Automated checks + review
Level4

Cross-source

DPP, certificate and ERP say 37.4%; procurement says 40%. A system can detect the difference and apply predefined authority rules – it cannot safely infer why the numbers differ.

Detect, then escalate

The more automated the exchange becomes, the more important it is to know what the system can validate – and when it needs to stop and ask.

04The many sources of external data

Source authority is attribute-specific, not system-wide.

An electronics manufacturer with an incorporated battery module relies on information that originates several tiers upstream. Designating one system as "the source of truth" only multiplies duplication.

Fig. 4 · Battery module – where each value originatesRed · internal routes
Material supplier Battery module supplier ERP/MES Declaration DPP ERP/MES DPP Laboratory Certificate Logisticsshipment data ERP PLM / PIM MES DPP Data Model 1 2 3 4 5
Legend
  • 1Material composition (dotted: declaration / evidence)
  • 2Battery specification
  • 3Supplier batch / component ID
  • 4Measured energy capacity (Wh)
  • 5Shipment / custody event

ERP holds the commercial identifier of the purchased module, PLM establishes which module belongs to which variant, MES links it to a production batch. The "quest for the ultimate upstream source" is often pointless – use the source with sufficient authority, context and accessibility.

OriginWhere it was generated

Measured in a laboratory, declared by a supplier, calculated from a bill of materials.

RecordWhere it is stored

The supplier's MES, the manufacturer's ERP – each holding its own representation.

Exchange channelHow it travels

Supplier DPP, portal, API, documentation – every node before the target system.

05Manual → structured → automated

Manual entry isn't the enemy. It's how you learn what to automate.

A manual workflow is a feedback loop: apply the model, encounter exceptions, refine rules, repeat – while stable steps are stacked in the automation backlog.

Fig. 5 · Progression towards automation – a conceptual model, not a mandatory oneLoops back on new exceptions
automation Manual entry Structuredexchange Automatedtransfer Automatedmapping &validation Orchestration routine established? new exceptions?
Interactive · automated routine

When the rule is clear it runs. When it isn't, it stops and asks.

Supplier API↳ route to human review
Retrieve value↳ route to human review
Map attribute↳ unknown identifier · human review
Validate↳ outside expected range · human review
Associate with batch↳ route to human review
Publish↳ route to human review
> ready
06Heterogeneous data

A connection moves data. It doesn't make it usable.

Normalization makes representations technically compatible. Semantic interoperability establishes whether they are conceptually compatible. Mapping turns the relationship into an operational rule.

normalize.runrules v2.4 · traceable
Semantic check · source field → target concept
SUPPLIER FIELDBattery capacity: 85 kWh
Nominal energy capacitykWh
Usable energy capacitykWh
Rated capacityAh
Measured capacitykWh · test
No direct mapping. Request clarification, keep the original as evidence, route to review. An unmappable value is preferable to a confidently incorrect one.

One-to-one

Production date → Production date

One-to-many

37.4% PCR PET → content, material type, methodology

Σ

Many-to-one

component mass + recycled mass → recycled %

0.37437.4%×

Transformation

0.374 kg/kg → 37.4% once meaning and scope are known

Conditional

Destination depends on component type, battery category or regulation

07Source conflicts

Not "which source is correct?" but "are they describing the same thing?"

A battery manufacturer receives recycled cobalt content for a cathode material from three sources. A simple routine sees the values differ – it cannot decide which one belongs in the DPP without rules defined upfront.

Recycled cobalt · cathode material
A · Supplier DPP42%
A′ · Supplier certificate42%
B · Own procurement system45%

Context-first reconciliation checks whether the values describe the same component, batch, period and method before comparing authority.

01Accept or prefer one value
02Retain multiple values, prioritize one for the field
03Flag for review
04Request clarification or correction
08Data decay

Data is not permanently correct once entered.

Facts change during the lifecycle (temporal decay), and representations drift as values move between systems (representational distortion). Each step can be legitimate – and still lose precision, context or meaning.

37.4%
original
0.374 kg/kg
unit conversion
0.37
rounding
30–40%
generalization
35%
derivation
Bar · precision retained relative to the original assertion (illustrative)
Supplier assertion42% recycled Co

Material batch C-2026-041, reporting context 2026.

✓ still valid
Lifecycle eventComponent replaced

Total cobalt in the bill of materials changes.

→ triggers dependents
Derived battery-level %Recalculate

No source expired – the conditions of the calculation did.

needs recalculation

The goal is not to prevent change. It is to make it visible, attributable and traceable.

09Granularity and identity

What exactly does 42% describe?

A supplier reports that 42% of cobalt in a cathode active material batch comes from recovered sources. The manufacturer's DPP is scoped by model, batch or individual battery. Three legitimate moves – not interchangeable.

Cathode active material powder in a lab dish
RECOVERED COBALT · BATCH C-2026-0410%
SUPPLIER SCOPE MANUFACTURER SCOPE Cathode active material Batch C-2026-041 Recovered cobalt Battery Model X Production batch 2401 Battery B12345 42% stays at material-batch scope used intraceable – meaning unchanged BOM cobalt massmethodology Battery-level value (new)= f(inputs, method) · traceable

Keep. If no battery-level value is required, the 42% figure remains at material-batch level – original scope preserved.

Same unit ≠ same identity85 kWh vs 85 kWh

One is the rated capacity specified for Model X, the other a measured result for battery B12345.

Context that mattersScope metadata

Model, variant, batch or serial; site; period; measured, calculated, declared or specified.

When there's no valid moveDon't invent precision

Retain the original and request more data rather than publishing a more precise-looking number.

10Burden or privilege?

Who bears the cost of interoperability?

If every company builds its own integration and mapping, the same work is duplicated across the network. Shared interfaces, schemas or intermediaries turn repeatable work into reusable infrastructure.

5 suppliers · 4 manufacturers
20integrations to build & maintain
20mapping sets
Supplier-provided interface+
UPSIDEExpose data once, serve many customers consistently.
DOWNSIDEMore work for the supplier; risk of a lowest-common-denominator model.
Downstream mapping+
UPSIDEEach manufacturer controls interpretation against its own model.
DOWNSIDEThe same supplier data is mapped repeatedly by many customers.
Shared intermediary+
UPSIDEExchange, transformation or validation done once for many.
DOWNSIDEDependency and lock-in: rules, governance, commercial model.
Industry schema or common model+
UPSIDEFewer bilateral mappings, easier supplier onboarding.
DOWNSIDEAgreement takes time; internal mapping is still required.
DPP platform as exchange layer+
UPSIDECollection, transformation, validation and access in one place.
DOWNSIDEPlatform dependency; semantics and responsibility still needed.
Bilateral integration+
UPSIDETailored closely to two organizations' needs.
DOWNSIDEScales poorly: every relationship adds another integration.

Getting to set your own schema is an "early bird" privilege. Integration capability can morph into infrastructure rather than a one-off compliance expense.

11Access and the B2B layer

Access is not the same as redistribution.

Manufacturer X may use V's confidential data to compile its DPP – but not publish it or pass it unchanged to Y. Technology can't always enforce contractual limits, but it can represent them and make permitted flows auditable.

Fig. 6 · Source data vs derived data across three operators▲ access conditions
Operator VOperator XOperator Y DPPV permission B2B sourcevalue evidence derivedvalue check DPPX ▲▲▲▲▲ ▲ V's records stay at X
Example · data-level access metadata
Recycled-content valueManufacturer X
Supporting production recordsX + authorized authorities
Onward sharing with defined partnersPermitted
Public disclosure of the recordsNot permitted

Source data → validation / calculation → derived result → authorized disclosure. Each stage can carry its own access conditions, enforced by role-, organization- and relationship-based rules, controlled APIs and audit records.

12From passport creation to data infrastructure

The DPP is only the visible part of a larger system.

As products, suppliers and exchange relationships grow, the question shifts from "How do we create our DPP?" to "What infrastructure keeps the information behind it reliable as everything changes?"

01Structured creation & validation
02Integration
03Orchestration
04Ecosystem reuse
05Continuous maintainability
01Data modelsentities · attributes · scope
02IntegrationsERP · PLM · MES · supplier DPPs
03Transformationsnormalize · map · calculate
04Validationsyntax → cross-source
05Orchestrationretrieve → reconcile → publish
06Access controlpublic · regulator · B2B
07Provenancelineage · evidence
08Monitoringfreshness · schema versions · escalations
DPPA reliable, living passportvisible layer
Conclusion

Standards can define how information should be represented and exchanged. They cannot eliminate the need to establish what it means, where it came from, whether it is still valid, and who is responsible for it.

Identify what a source value actually describes
Structure and normalize heterogeneous information
Map different representations and meanings
Validate values, context and supporting evidence
Resolve or escalate conflicting information
Preserve provenance as data moves between systems
Distinguish product, component, batch and item scopes
Control who can access or redistribute information
Monitor information as sources and requirements change
Whitepaper · 2026

Building Digital Product Passports with External Data

Digital infrastructure needs, challenges, and solutions
Free download · PDF

Get the full whitepaper

10 chapters on data models, validation, source conflicts, data decay, granularity, B2B access and the infrastructure that keeps a DPP reliable.

It's on its way.

Check your inbox for the download link. Want to discuss your DPP data infrastructure? Talk to Lionwood →