GS1 QR codes and digital watermarks: what each asks of the pack

Quick answer: For checkout and shopper scans, the 2D option most FMCG teams will meet first is a QR code carrying a GS1 Digital Link, one of three formats in GS1’s retail guideline. Digital watermarks do a different job, helping sorting plants identify a pack by SKU and material, and they are past their first industrial trials but short of broad rollout. The two can share a pack.

This piece rests mainly on GS1’s own pages, announcements from the HolyGrail 2.0 partners and material from Digimarc, which sells watermarking and was the technology provider in those trials. Independent test data is thin, and I’ve said where.

What GS1 is asking retailers to do by the end of 2027

GS1’s 2D barcodes at retail point-of-sale guideline lists three POS 2D options for the transition and beyond: a QR Code with GS1 Digital Link URI syntax, Data Matrix with GS1 Digital Link URI syntax, and GS1 DataMatrix. The target, called Sunrise 2027, is that retail checkout systems can read a GTIN from one of these by the end of 2027.

It’s a readiness goal for scanners, not a rule for your pack. In a Packaging Digest report, GS1 US’s Mears stressed that Sunrise 2027 is industry-led and not a mandate. GS1 US’s retailer page says items will need dual marking with a U.P.C. and a 2D barcode during the transition, and that U.P.C. barcodes will continue to be accepted and scanned after the 2027 sunrise date.

If you sell into the EU, the packaging law (PPWR) has its own labelling and marking provisions. It’s summarised at ppwreu.org; check the current text before you decide anything on that basis.

How a Digital Link QR code works

A GS1 Digital Link is a web address built to a fixed pattern. The GTIN sits in the path, optionally followed by batch, expiry or serial number. A checkout scanner reads the identifier out of the string; a phone camera opens the address in a browser. A marketing QR code with a campaign URL is a different thing: an explainer from Advanced Automation says a standard QR code linking to a landing page is not GS1-compliant for Sunrise 2027.

Between the code and the content sits a resolver, a web service that decides where each request goes. GS1’s open-source Resolver Community Edition redirects a consumer who scans the code to the right page on the brand’s website, without changing existing web applications. The same code can send a shopper to a product page and return different data to a business system, and the destination can change without reprinting. That last point matters for the pack: the printed string is permanent, so the address behind it has to stay alive for as long as the pack is on shelf and in homes.

NFC is the other carrier in the beat. Aeroz, which sells resolution services, summarises GS1 as saying the same Digital Link can serve checkout and online information from either a QR code or an NFC tag. I found no published retail-POS guidance that treats NFC as a checkout option, so I’ve left it out of the table.

Free tool: the GTIN validator checks the check digit, which is worth doing before a GTIN is baked into a URI and sent to artwork.

What the code needs from the pack and the line

The GS1 guideline says the Digital Link URI syntax in a retail POS code has to use the uncompressed form. That makes the string longer than the compressed alternative, so the symbol has more modules to print cleanly at a given size. GS1’s General Specifications set the size and placement rules; check them rather than relying on a rule of thumb from a blog.

Contrast matters on dark or metallised film. A software blog summarising GS1’s 2D Barcode Playbook says retail scanners should handle reversed reflectance and ignore non-GS1 codes, which is the scanner side of printing light-on-dark codes. Whether your retailer’s scanners do this is a question for them, not a given.

On the line, the practical items are the same as for any variable or fixed code: a print method that holds quality on your film or carton, and verification of grade rather than a simple “reads/doesn’t read” check. Ask your line integrator what the vision system on your line can grade.

Digital watermarks: a code for the sorting plant

A digital watermark is a pattern spread across the printed surface, invisible to the eye. It can only be read by high-resolution cameras (interpack). Digimarc’s patent filings describe printing it by modulating the package’s own design colours or adding ink, repeated in a tile-like pattern across the surface. Because it’s tiled, any face of the pack can be read, so nobody has to hunt for a symbol.

The same filings note a design constraint: inks with low reflectivity around 660 nm look black to a red LED scanner and can be hard to encode. Digimarc’s own product page says the watermarks embed in artwork with no special inks or modified printing processes; that’s a company claim. An older Labels & Labeling profile adds that the pattern is applied to print-ready, colour-separated art files and that colour correcting the file afterwards could compromise scanning or make the watermark visible. So the step belongs late in prepress, with your repro house told about it.

Who has trialled watermarks, and what they found

HolyGrail 2.0, run by AIM (the European Brands Association) and powered by the Alliance to End Plastic Waste, is the main body of evidence. In its 10 April 2025 release, trials at the Hündgen Entsorgung recovery facility in Swisttal, Germany, reached about 90 percent or higher sorting efficiency on post-consumer rigid household packaging, single pass. The modules combined near-infrared and digital watermark detection. Aldi, Arla Foods, Netto Marken-Discount, Haleon, Hochland, Mondelēz, Orkla and Procter & Gamble had watermarked SKUs placed on the German and Danish markets. One outlet, Innovation News Network, reported 93 percent sorting accuracy for the same trials.

The Ellen MacArthur Foundation adds that the add-on modules were developed with machine vendors Pellenc ST and Tomra and detected watermarks at rates above 96 percent in the control test. The programme has moved on to HolyGrail 2030, which focuses on market adoption programmes.

Flexible film is harder. The flexibles trial targeted separating hygiene-grade LDPE films from food-grade PP films, with single-pass results below.

HolyGrail 2.0 flexibles trial, single-pass results. Dec 2023 (reported as above these values): Detection efficiency 95, Sorting efficiency 85, Purity 70; Feb 2024 (baled, aged streams): Detection efficiency 89.9, Sorting efficiency 75.1, Purity 88.1 (percent)
HolyGrail 2.0 flexibles trial, single-pass results
Source: HolyGrail 2.0 flexibles trial results as published by Petnology

The December 2023 assessments averaged above 95 percent detection, above 85 percent sorting efficiency and above 70 percent purity. For the more contaminated, baled and aged streams in February 2024, detection was 89.9 percent, sorting efficiency 75.1 percent and purity 88.1 percent. The trial team judged the December results more indicative of a single sorting step in practice. Our calculation: 85 minus 75.1 leaves a gap of at least 9.9 percentage points in sorting efficiency between the cleaner and the dirtier stream, using the December figure as a floor.

The options side by side

Option What it carries Read by Needs from pack and line Status, 13 Nov 2026
UPC/EAN 1D barcode GTIN only Every current POS scanner A printed symbol in a set place GS1 US says it stays accepted after 2027
QR code with GS1 Digital Link GTIN plus optional batch, expiry, serial, as a web address Phone cameras; 2D POS imagers Uncompressed URI, quality print, a resolver that stays live One of three GS1 retail options; Sunrise 2027 is a target, not a mandate
GS1 DataMatrix GTIN and data elements in GS1 element-string syntax 2D imagers Small, dense symbol printed to spec One of three GS1 retail options
Digital watermark Product and material identity, tiled across the surface Sorting-line cameras; some POS and phone readers (Digimarc) Artwork step after colour separation; ink choice affects detection Industrial trials done for rigid packs; flexibles less proven

What is still unproven

On the 2D code side, the open question is retailer readiness: the target is a date for scanners, and my sources give no per-retailer status. A pack that carries a Digital Link also carries a promise to keep the destination live, which is an operating cost that nobody’s pack spec shows.

On watermarks, the evidence comes from partner-run trials with the technology provider inside the project. The rigid-pack result was single pass; the initiative expects two-pass sorting, common in plants, to do better, but that is expectation, not a published result in my sources. MCC’s HolyGrail summary says watermarks could potentially support future EFSA approvals for recycled food-contact material, and quotes Arla’s sustainable packaging lead noting that competing technologies, including AI-powered image recognition, are evolving in parallel. Nothing I found shows a sorting plant outside the trials running watermark detection routinely.

What this means for you

  • Packaging engineers: if a retailer asks for a 2D code, get the exact format (QR or DataMatrix, Digital Link or element string) in writing, then check size and placement against the GS1 General Specifications and your print method.
  • Brand and design teams: ask who owns the domain behind the code and who maintains the resolver. A self-hosted option exists in GS1’s open-source resolver; a vendor-hosted one needs an exit clause.
  • Prepress and print: if you’re considering watermarks, ask the provider which of your spot colours and dark inks cause detection problems, and agree where in the file workflow the pattern is applied.
  • Sustainability and procurement: treat watermarks as a pilot for rigid HDPE, PP and PET packs sold in markets where sorting plants are in the programme. For flexibles, ask for trial data on your film type, and for a quote with your supplier.
  • Quality teams: add code grading to the line’s checks before anything ships with a new code on it.

Sources

Header image: illustration, AI-generated.