GS1 Application Identifiers for Digital Product Passports: GTIN, Batch, Serial, Expiry and Variant
Understand GS1 Application Identifiers for GTIN, batch, serial and expiry in a Digital Product Passport architecture. Separate GS1 encoding semantics from DPP legal requirements and granularity.
GS1 Application Identifiers tell a system what a GS1 data element means. They are the small numeric prefixes behind values such as GTIN, lot or batch number, expiration date and serial number. They are useful in Digital Product Passport architecture because they let one carrier or URI distinguish different identity and attribute layers in a machine-readable way. They are not a DPP checklist. An Application Identifier does not make the underlying fact legally mandatory, and EU DPP law does not turn every GS1 AI into passport data. The right sequence is: establish the legal granularity and required facts first, then decide whether GS1 Application Identifiers are the right encoding language for the implementation.
The short answer
The four Application Identifiers most likely to appear in early DPP conversations are:
GS1 AI
Meaning
DPP relevance
Do not confuse it with
01
GTIN
Trade-item identity and a possible product-identity building block
The universal legal meaning of UPI
10
Batch or lot number
Batch-level qualification or traceability
A new product model or variant
17
Expiration date
Time-sensitive product attribute where relevant
A universal DPP requirement
21
Serial number
Individual-item qualification
A model-level product identifier
The current GS1 General Specifications are release 26.0, ratified in January 2026. GS1 Digital Link URI Syntax 1.7.0 aligns its data attributes with General Specifications release 26.0.
If your DPP architecture uses GS1, these standards provide the semantics. The applicable EU product law still decides what the passport actually has to identify and contain.
What an Application Identifier is
A GS1 Application Identifier, or AI, is a numeric code that defines the meaning and format of the data that follows it inside the GS1 system.
It lets a machine distinguish, for example, a GTIN from a batch number even if both are printed as digits and characters.
In a GS1 element string, the AI is part of the encoded structure. In a GS1 Digital Link URI, GS1 keys, qualifiers and data attributes are represented using the URI syntax defined by the Digital Link standard.
The AI therefore belongs to the data semantics layer. It is not the physical Data Matrix or QR Code itself.
AI 01: GTIN
AI 01 identifies a Global Trade Item Number.
For many retail businesses, GTIN is already the product identity used in barcodes, marketplace feeds and retailer item masters. That makes it an obvious candidate to reuse when designing a DPP identity architecture.
But the legal distinction matters.
Regulation (EU) 2024/1781 uses the term unique product identifier for the persistent identifier connected to the DPP. Annex III includes GTIN under ISO/IEC 15459-6 or equivalent among information that a product-specific measure can select.
This is useful where products that share a trade-item identity need to be distinguished by production batch. It does not automatically mean the DPP itself is legally batch-level.
Three levels must stay separate:
the trade item a buyer recognises;
the production batch a manufacturer traces;
the passport granularity required by applicable law.
A model-level passport can still have batch information associated elsewhere in the product-data estate. A batch-level passport can use a batch qualifier as part of the identity architecture. The presence of AI 10 in a barcode does not decide which legal model applies.
This is the GS1 data element most directly associated with distinguishing one physical item from another where GTIN identifies the trade item and serial identifies the individual unit.
That makes it relevant to item-level passport architectures, lifecycle tracking, repair history and other individual-unit use cases.
Again, the direction of causality matters:
law or business use case creates the need for item-level identity; the serial AI provides a GS1 way to encode it.
Do not serialise an entire product population simply because a 2D code can hold a serial number.
AI 17: expiration date
AI 17 identifies expiration date.
It is a product attribute rather than the primary product identity. GS1's retail 2D programme uses expiry as one example of the additional data that richer 2D carriers can support.
That can create useful retail use cases such as expiry management or sales prevention for expired products where the retailer systems are designed for it.
It does not mean every Digital Product Passport must contain an expiration date. Many product categories have no meaningful expiry date at all.
This is the general rule for GS1 data attributes: encodable is not the same as legally required.
Where is the “variant” AI?
There is no single universal DPP concept called “the variant AI” that replaces proper product identity modelling.
In many commercial catalogues, a sellable size-colour combination has its own GTIN. That means the variant distinction is expressed through the GTIN allocation itself rather than by adding a generic “variant” field to one GTIN.
Other GS1 Application Identifiers can describe specific characteristics or qualifiers in defined use cases, but do not use an arbitrary extra AI to repair a catalogue whose variant identity is wrong.
The practical question is:
Does this variation create a different trade item, a production batch, an individual item or merely another descriptive attribute?
Get that answer right before encoding anything.
Primary key, key qualifier and data attribute are different roles
GS1 Digital Link makes the distinction especially visible.
At a high level:
a primary identification key identifies the main entity, such as a trade item;
a key qualifier narrows that identity, such as a lot or serial where the GS1 rules permit it;
a data attribute provides additional information about the identified entity.
Those roles matter because a serial number used to identify an individual item is not semantically the same as an expiration date attached to it.
This is why a long list of AI numbers is a poor DPP data model. The implementation should first decide which facts identify the object and which merely describe it.
How AIs appear in GS1 DataMatrix
GS1 DataMatrix uses Data Matrix under GS1 rules and can encode GS1 element-string data using Application Identifiers.
For example, a carrier can encode a GTIN plus batch or serial information in a defined GS1 structure. The scanner and downstream software need to understand the GS1 syntax and parse the data elements correctly.
The visible square does not tell you which semantics are inside it. A generic Data Matrix and GS1 DataMatrix can look similar while behaving differently in systems.
The AI value 01 tells a conforming system that the following value is a GTIN. Qualifiers and permitted data attributes can then be represented under the Digital Link rules.
The benefit is that the same GS1 identity becomes a web identifier. The physical carrier can be QR Code or Data Matrix, provided the chosen implementation follows the applicable GS1 and product requirements.
EN 18219:2026 is the harmonised European DPP standard on unique identifiers. Its public scope covers unique product, economic operator and facility identifiers, including global uniqueness, persistence, syntax, granularity, interoperability and openness. It accommodates product identity at model, batch or individual-item level and includes issuing-agency and self-issuing approaches.
That is a broader standards question than GS1 Application Identifiers.
GS1 AIs tell you how defined data elements work inside the GS1 system. EN 18219 is part of the DPP harmonised standards route for identifier requirements. The applicable product law sits above both and decides whether the product needs a DPP and the relevant product-specific detail.
We have reviewed EN 18219's public BSI scope, not the full licensed standard. Do not infer from this page that a particular GS1 AI is mandated by EN 18219.
A practical mapping from DPP granularity to GS1 identity
This table is an implementation orientation, not a legal prescription.
DPP/traceability need
Possible GS1 identity shape
What still has to be established
Product/trade-item identity
GTIN using AI 01
Whether GTIN is the chosen compliant UPI route
Batch distinction
GTIN + AI 10 lot/batch
Whether the passport or process is legally batch-level
Individual-item distinction
GTIN + AI 21 serial
Whether item-level identity is required or useful
Expiry-aware process
Identity + AI 17 expiry
Whether expiry is legally or operationally relevant
The mistake is to read the right-hand side backwards and say that because GS1 has an AI for a fact, the DPP requires that fact.
What product-data teams should store
Do not store only a single “barcode” string.
A governed data model should be able to separate at least:
identifier scheme;
identifier type;
identifier value;
scope or granularity;
batch/lot where relevant;
serial where relevant;
expiry where relevant;
encoded syntax;
carrier type;
source and allocation evidence.
That makes it possible to generate different carrier representations without changing what the product actually is.
It also makes errors visible. A SKU in a GTIN field, a batch number treated as a new product or a serial copied across many items becomes much easier to detect.
What POS teams need to know
GS1's current retail 2D guidance is designed around GTIN plus optional additional data such as batch, expiry or serial where the use case requires it.
That can unlock richer POS and supply-chain behaviour, but only if the scanner and host can process the data.
A retailer can choose to use the GTIN and drop unsupported extra data. That is another reason not to assume everything encoded in a 2D barcode automatically becomes part of the transaction or DPP.
