Model, batch or item: the passport decision you cannot undo
Every product passport identifies something at one of three levels: a model, a batch, or an individual item. Most guidance treats this as a technical footnote. It is not. It sets your printing cost, your data volume and your ability to answer questions after something goes wrong - and once the product is on the market, the level is fixed.
What the three levels actually mean
- Model - one passport covers every unit of the product. All units share the same technical characteristics and the same model identifier. Cheapest by a wide margin: one code, printed identically on everything.
- Batch - one passport per production run. Units within a batch share characteristics, but you can tell one run from another. The code differs per batch, which means a variable element in printing.
- Item - one passport per physical unit. Every single product carries a unique code. Most expensive to print, to manage and to keep alive.
The rule that makes this irreversible
EN 18219, the standard governing passport identifiers, requires that the level of granularity stays consistent once the product is placed on the market. If a finer level later becomes necessary, you do not simply switch: a new or extended identifier has to be issued, and it has to be linked to the old one so the history is not broken.
The practical reading: you are choosing before you print, and the printed code is the thing you cannot recall. Products already in shops keep pointing wherever they pointed on the day they left the factory.
When to go finer than model
The standard includes non-binding guidance on when item level is worth it, and the criteria are refreshingly concrete. Paraphrased, item level is advisable when the product:
- is designed to be repaired, refurbished or overhauled by replacing parts
- can be upgraded or enhanced by swapping components
- carries software that gets updated
- undergoes regular maintenance, inspection or a statutory audit
- needs wear, operating hours or other usage information recorded
- requires individual identification to serve its purpose at all
- is exposed to a high risk of counterfeiting
- is one you want installed-base data about, for planning
Notice what these have in common: each describes something that happens to one specific unit after it is sold. If nothing on your product happens per unit, you are paying for a capability you will not use.
Batch level answers a different question: not "what happened to this one" but "which run did this come from". Complaints, recalls and raw material sourcing all live at that level. For most textiles, food and cosmetics it is the natural fit.
What the registry does with your choice
Until recently this decision could only be discussed in the abstract. The EU registry is now live, its rulebook is published, and both say things about granularity that change the calculation.
The most granular applicable level wins. Implementing Regulation (EU) 2026/1778 settles the case where several pieces of Union law require registration of the same product at different levels: the passport is registered at the most detailed level required by any of them. So if you were planning model level and a second regulation touching your product asks for batch, the answer is batch, and the choice is not yours to optimise.
Going finer does not free you from the coarser identifiers. Register at item level and both the batch and the model identifiers have to be linked to that passport wherever such a batch or model design exists. Register at batch level and the model identifier has to be linked. Item level is therefore not a way to skip the work of organising your model and batch structure; it is that work plus one more layer.
The identifier you register is a URL, and it is validated. The registry checks it, rejects excessive redirects and any drop to a less secure protocol, and in the current version caps the length at fifty characters. We asked the Commission about that cap on 17 August 2026; the answer is that it reflects this version of the registry, does not fit every GS1 Digital Link form, and will be raised in a future release. The redirect and https rules, on the other hand, are not going anywhere.
This matters most at item level, because that is where the address has to carry a serial number. A digital link path carrying a product code, a variant, a batch and a serial is a long string. If your architecture forces every unit through a redirect or a long qualifier chain, item level is where it breaks first.
One more practical detail, from the batch upload templates the helpdesk sent us: a bulk registration file carries the schema version, the product group, and a list of unique product identifiers - nothing else. No model identifiers, no batch identifiers, no metadata. Whatever structure you keep, the registry wants one resolvable address per passport.
The costs people underestimate
Going finer than needed means variable data printing, a record per unit, and paying to keep every one of those records available for as long as the rules require. The printing cost is visible. The record-keeping cost is the one that surprises people, because it never stops.
Going coarser than needed costs nothing until something goes wrong. Then it costs everything: without batch identification you cannot scope a recall, so you recall the whole product line.
That asymmetry is worth sitting with. Too fine is a steady, predictable overspend. Too coarse is a rare, catastrophic one.
One argument for the decentralised route
There is a wrinkle worth knowing if you are choosing between identifier schemes. With classic supply chain identifiers, changing granularity later means issuing a new identifier. With a decentralised identifier, the level is an attribute of the identifier document rather than of the identifier string, so changing it is an update rather than a reissue.
For a small brand that might do batches now and items in two years, that flexibility is worth more than it sounds. It does not remove the rule about consistency after market placement, but it makes the eventual transition much less painful.
What the rules may decide for you
If the delegated act for your product group prescribes a level, that settles it and this whole discussion is moot for you. Today most groups leave the choice open, and the guidance is explicit that where regulatory requirements allow a choice, a more granular level may be adopted to improve traceability or meet operational needs.
So the honest position: check the rules for your group first, and if they leave you a choice, use the criteria above rather than guessing.
Answer eight questions instead
We built the criteria into a free tool. No account, no email, nothing stored. Tick what applies to your product and it tells you which level fits and why.
Open the identification level tool
FAQ
Can I start at model level and go finer later?
Yes, but not silently. A new or extended identifier has to be issued and linked to the previous one, and anything already printed keeps pointing at the old level.
Is batch level a compromise?
No. It is a fully recognised level, not a workaround. For a large share of consumer goods it is the correct answer rather than the cautious one.
Does item level mean a unique QR on every product?
In practice yes, which means variable data printing. That is the cost line to check with your label supplier before deciding.
What if my product group has no delegated act yet?
Then you have a choice, and you should make it deliberately now rather than by default at the first print run.