Choosing a carrier that still works in five years

Six words, and why collapsing them is the expensive mistake

These get used interchangeably in proposals and they fail in different ways, at different times, under different people's control.

The first four are taken apart in detail, with what each is not, on what actually happens when somebody scans the code. The three points worth carrying here are that a carrier is not a passport, a passport is not the whole of a product record, and a code that resolves is not evidence that anything behind it is current.

Five things that have to survive, and only one is physical

Longevity gets discussed as an ink and abrasion question. It is five questions, they fail at different rates, and four of them are contractual or organisational rather than material.

The third and fourth are where most long-life programmes will actually fail, and they are the two that look like somebody else's problem at the point of purchase. What the standard leaves open about continuity, and what a contract has to close, is set out on when the link dies.

The architectural consequence is the useful part, and it follows from one property of the list. Only the resolver and the destination can be changed after goods ship. So the design that survives keeps the printed part minimal and stable, holds a small durable map from identifier to link type to target, and puts everything that changes at the target address where it can be rebuilt without anybody touching the map. Putting the volatile logic into the resolver itself moves it into the layer with the longest life and the least tooling.

The three decisions in front of you

What the mark is. Four carriers circulate under one name and only three of them are machine interpretable as identity.

Where it sits. There is a measured placement constraint for retail scanning that most published commentary omits entirely.

Whether it will still be there. Nobody has established whether any digital identity carrier survives a garment's ordinary life, and that is the load bearing unknown underneath everything above.

The four things people call the code

The distinction that matters is between the first three and the fourth. The first three are machine interpretable as identity: a system reading one can extract the product number and, where present, a batch or serial. The fourth is machine interpretable only as a destination. No point of sale system, no customs system and no downstream reader can get a product number out of it, because there is not one in there.

Two practitioner failure modes follow directly from that table and both are common. A brand prints a marketing address and believes it has a compliant carrier, having put no product number, batch or date structure inside it. Or a brand prints correct address syntax and never stands up anything to answer it, so every scan lands on the same destination regardless of what was asked for, which discards the reason the syntax exists. Both surface late, and the second usually surfaces when somebody in a different market asks why they get an English page.

A note on naming. The third row is commonly called a GS1 QR code, which is shorthand for what the code carries rather than a distinct symbology. It is an ordinary QR code with a particular kind of address inside it. What the identifier itself commits you to is a separate decision, set out in identifiers, and what allocating one commits you to.

The placement constraint nobody quotes

Testing published by GS1 found that a two dimensional symbol needs to be within fifty millimetres of the centre of the linear barcode to achieve the target retail scanning rate of forty to seventy items per minute, and that without placement rules high speed retail checkout may be unable to meet its productivity target.

That is an artwork and ergonomics constraint, and it is absent from essentially all published commentary about carriers, which treats the subject as a software question.

For apparel it lands differently than for grocery, because a swing tag and a sewn in label are not a printed pack panel and a garment does not usually cross a belt at seventy items a minute. It still matters, because it is the reason a retailer may care where your mark sits, and because it demonstrates that the transition being discussed as a data question has a physical half that gets decided by whoever lays out the label.

The related question, whether the industry target for two dimensional codes at the till is a deadline, is answered on identifiers, and the answer is that it is a target set by an organisation for itself with no published measurement of its own threshold. Nothing here restates that argument.

Carrier choice is diverging, and it tracks product form

Across documented programmes the choice is not converging on one answer. It sorts by what the object is.

Near field tags appear where the object is large, expensive and handled with a phone already in the hand. Poltrona Frau furniture, BRABUS vehicle interiors, Axopar helm fittings, a plaque on an MCM bag.

Printed codes appear where the item is cheap, printed at line speed or has to be read at a till. Woolworths fresh categories, 7-Eleven Thailand, wine e-labels under the EU regime, Fnac Darty. 7-Eleven Thailand chose Data Matrix over QR on size grounds alone, because the symbol had to fit.

Marks already present in the object get reused rather than replaced. Bvlgari reactivated an engraving first applied in 2020.

Invisible watermarks, of the kind Digimarc and Polytag work with, appear where the mark must not be seen or must survive sorting.

And Another Tomorrow ships both a near field tag and a printed code on the same label. That is a hedge against not knowing which the customer will use, it is an honest answer to a question nobody has settled, and it costs twice.

The cost structure, where the published numbers mislead

Passive tag inlays are quoted at roughly five to fifteen US cents at high volume, and fifteen to twenty five at around a thousand units. Near field and high frequency tags run roughly ten to fifty cents. A printed code has effectively no marginal cost at all.

Those figures describe a chip or an inlay. The decisive point is structural rather than numeric.

A printed code is a change to a printing plate. A tag is a line on the bill of materials, plus a converting step, plus the cost of applying it, and the all in cost of an applied tag in a converted label at apparel volumes is not published by anybody. The gap between the quoted number and the real number is the part of the decision that is missing, and it is missing consistently across every source we found.

For an authenticating tag, meaning one that answers a challenge rather than serving an address, no unit cost at apparel volumes exists in public at all. That is the number a brand pursuing authentication would actually need, and the reason it matters is on the difference between identifying a product and proving it is genuine.

The unknown that sits under all of it

Whether any digital identity carrier survives a garment's ordinary life is not established.

Domestic laundering. Alteration. A shortened hem. A resoled boot. No independent test data for consumer apparel was found in this research.

What does exist points sideways rather than at the question. Washable tags rated to around a hundred industrial cycles are a real product, sold into industrial linen, which is a different environment with a different laundry process. One tagless approach claims testing that simulates twenty years of wear, and that is vendor testimony without independent verification.

This matters more than it first appears, because the entire long life case for a carrier assumes it is still readable years later. Repair, resale, recycling and authentication all place the moment of use somewhere well after purchase. If the carrier is gone by then, the value proposition that justified it never arrives, and nobody would find out, because nobody is measuring resolutions against units in circulation.

It is a question one properly designed study would answer, and it is the single most useful piece of work available to anybody in this category.

Whether one carrier can do two jobs

One carrier may not be able to do two jobs, and there is a live regulatory precedent for that rather than a theory.

In the one European regime that already places mandatory product information behind a code, mandatory information must not be served alongside sales or marketing content, and data about the people viewing the e-label may not be stored. The practical consequence is that a producer cannot host the compliance information on their own marketing site.

That regime is wine. It is not clothing, and the ecodesign framework carries no equivalent prohibition for textiles. A textile business reading it as a rule that binds them has read it wrongly, and the constraint on what may be measured at a carrier is taken apart, with the scope guard intact, on what a scan actually tells you.

What it demonstrates is narrower and more useful than a prohibition. An industry body argues, in published material, that a single code can serve regulatory information and marketing content at once. A European legislator, handed the same question in one sector, decided it could not. That is a live disagreement between advocacy and an instrument, and it is worth knowing about before somebody commits a label design to the single code position.

What this page does not decide

It does not recommend a symbology. The right answer depends on where the item is scanned, by whom and with what, and the four row table above is the decision rather than a preference. A published European standard now covers symbology, encoding, print and production quality, durability and placement, and it excludes secure elements and cryptographic features by its own scope. What it takes in and what it leaves out is on what the six published standards cover.

It does not carry a readiness percentage for two dimensional scanning at retail. The threshold is published, the measurement of it is not, and quoting a vendor's figure would be repeating something with no survey instrument behind it.

It carries no cost for an applied tag, and no cost at all for an authenticating one, because neither is published.

What to do before the artwork goes out

Decide which of the four you are printing, in writing. Most of the failure modes in this territory are somebody assuming they were printing the third row and shipping the fourth.

Check what comes back before the print run, not after. Ask for a well formed address naming an identifier your service does not hold and confirm it answers as not found, with a not found status code rather than a branded page served as a success. The full set of checks worth running against whoever answers your codes is on what actually happens when somebody scans the code.

Settle placement with whoever lays out the label, not with whoever is building the data.

Get the destination's ownership in writing. Who owns the domain the redirect points at, and what happens to it if the supplier stops trading, is the cheapest thing to settle now and the most expensive to discover later.

Treat carrier survival as an open risk and write it down as one. If your programme's value arrives years after purchase, it depends on an assumption nobody has tested. What keeps the address answering after the company behind it changes is on when the link dies.

What would close these gaps

Independent durability testing of any carrier type through normal domestic laundering and alteration in apparel. That closes the largest gap here.

A published all in cost for an applied tag at apparel volumes, from anybody, with the volume stated.

Any published measurement of the two dimensional scanning installed base at retail, by the body that set the threshold or by anybody else with a survey instrument.

Sources

• GS1 retail guideline on two dimensional barcodes
  Standards body guidance

  Read for the placement finding, the fifty millimetre distance and the forty to seventy items per minute retail rate. Registered here with its read depth: the placement testing is described rather than quoted, and no readiness percentage appears because the guideline publishes none.

• Carrier and symbology research across documented programmes, compiled 27 August 2026
  Field research

  Used for the four way distinction between carrier types, for the two practitioner failure modes and for the divergence of carrier choice by product form.

• Tag and inlay pricing from vendor sources
  Vendor pricing

  Indicative only. Cited for the chip and inlay level and specifically to make the point that the applied cost, which is the number that matters, is not published.

• Carrier durability in consumer apparel
  Not established

  Searched and not found. Adjacent industrial linen ratings and one vendor durability claim are described in the body and neither is treated as evidence for apparel.

• Regulation (EU) 2021/2117, electronic labelling of wine
  In force

  Cited for one proposition and no more: that in at least one European regime placing mandatory information behind a code, that information may not be served alongside marketing content and viewer data may not be stored. It is wine, it is not clothing, and nothing here extends it to textiles.

Sources as at 30 August 2026.