A spirits brand can say how many cases it shipped to Japan last year. It cannot say how many bottles were sold there. A year after dispatch, the data stops at the distributor’s dock. This is not a reporting gap. Brands hold data on products, orders and customers, but no record of individual items once they leave the warehouse. An item-level product data layer, fed by production, logistics and field scans, supplies that record.
Premium brands know their SKUs in detail, their shipments to the day and their best customers by name. Point them at one bottle, bag or flacon twelve months later and nobody can say where it is, whether it was sold in its intended market or whether it is genuine.
Counterfeit trade was estimated at USD 467 billion in 2021, up to 2.3% of global trade (OECD/EUIPO, Mapping Global Trade in Fakes 2025). In the EU alone, fakes represented USD 117 billion, 4.7% of imports (OECD/EUIPO, 2025). No customs statistic captures genuine goods sold outside their allocated markets. Both problems sit in the same blind spot.
The causes are structural and habitual:
None of these is a technology problem. Carriers, platforms and standards have existed for years. What has been missing is a reason strong enough for several teams to invest in the same thing at once. Regulation, resale and diversion now provide it together.
A product data layer is a record per physical item, keyed on a standard identifier, built from events written by whoever touches the item and readable by any system with the right permissions.
The identifier is typically a GS1 Digital Link carrying a GTIN plus a serial number. The events are commissioning on the line, shipment to an intended market, retailer scans, consumer taps, resale and service. It is not a data lake and it is not a dashboard. It is the record the dashboards have never had.
The distinction that matters: a shipment record is logistics knowledge. A verified scan of one specific unit, in one specific place, compared with the market it was shipped to, is evidence.
Every system a brand runs today was built when the unit of business was the SKU or the shipment. Each is item-blind by design.
| System | What it knows | Blind spot |
|---|---|---|
| PIM | What each SKU is: attributes, content, compliance | Any specific unit |
| ERP / WMS | Which batches shipped to which customer | Anything after the distributor receives them |
| CRM | Who the known customers are | Which product they own, unless registered |
| Marketplace monitoring | Who lists suspicious offers online | Whether the physical item is genuine |
| Item-level data layer | Which unit, where it is, what happened to it | None of these gaps |
Five questions expose the gap. How much of the stock allocated to a market sold there? Is this pre-owned item genuine, and who owned it? Which distributor’s shipments surface in unauthorised channels? How many units of a recalled batch are still in circulation? Which passport data applies to this exact unit?
Today these are answered with estimates, test purchases and investigators. With the layer, they become queries.
The process runs from the production line to the field.
The first line usually takes a few months, because that is where standards, carrier, line commissioning and the ERP link are set. Later lines reuse the foundations.
An item-level layer built on copyable codes has a flaw: a counterfeit produces a record that looks genuine. Every fake scan becomes a false data point on where the brand’s products are, and every report built on it is quietly wrong.
Think of a basic code as a door with the key hanging beside it. A cryptographic chip is a door that issues a new one-time code at each approach, and only something already inside can produce the right one.
| Carrier | Can a copy produce a valid record? | Fit for the data layer |
|---|---|---|
| Printed QR or serial | Yes, a photocopy scans the same | Information only, records are not evidence |
| Copy-resistant print (ACM-secured barcode, Diamond Dust) | Hard to reproduce | Good where unit value rules out a chip |
| Cryptographic NFC (NTAG 424 DNA class) | No, every tap yields a new signed message | Highest trust, tamper evidence with the TagTamper variant |
| UHF RFID | Not designed to resist copying | Logistics events inside the controlled chain |
NTAG 424 DNA is ISO/IEC 14443-A, NFC Forum Type 4, with AES-128. For supply chain scanning of cases and pallets, vicinity-class chips under ISO 15693 serve a different purpose. They are not the standard for phone-tap authentication.
Serialised QR wins in several situations, and the honest answer is to use it there.
A printed QR adds roughly zero marginal print cost. The real cost is serialisation and the platform behind it (industry sources, 2026). The ESPR (Regulation (EU) 2024/1781) is carrier-neutral, and GS1 Digital Link QR is emerging as the de facto default. The workable framing is NFC for trust, visible carrier for compliance. Many brands will run both, with the QR as the public entry point and the NFC chip proving the item is real.
Published converter prices for NTAG 424 DNA-class labels cluster around $0.45 to 0.65 per unit at five-figure volumes. On-metal variants add 7 to 30%. Above 50,000 units, pricing is quote-only (public catalogues, 2026).
Standard non-secure tags of the NTAG213 class list at about €0.19 to 0.23 at 1,000 to 10,000 units (public catalogues, 2026). The gap between the two is the price of a record that cannot be forged.
For premium goods, that difference is small next to the value of knowing every record comes from a genuine item.
No app is needed. iPhone XS, XR and later read tags in the background. Android reads natively with the screen on. Read reliability reaches 99% in Selinko deployments.
Two physical constraints matter at design stage:
Placement and form factor are tested during line commissioning, not after launch.
The mechanism is constant. The threat and the question change by category.
Wine and spirits. Allocated stock leaks between markets and genuine bottles get refilled. Intended market at shipment, plus tamper-evident tags in the closure, traces diversion to a shipment and flags refills by venue.
Leather goods and fashion. Resale happens without the brand. Item records with ownership certificates that transfer on resale let the brand see the secondary market and keep the relationship. Clothing, footwear and leather goods account for 62% of seized counterfeits (OECD/EUIPO, 2025).
Cosmetics and fragrance. Stock appears in unauthorised retail and marketplaces. Consumer scans at point of use, compared with intended market, identify the leaking channels.
Watches. Service and warranty records scatter across boutiques. One record per watch, with authenticated service events, puts warranty, service and resale on a single history.
Cigars, furniture and premium audio. High unit value, a gray market or a long service life make the same case: one verifiable identity per item.
The central EU DPP registry has been live since 20 July 2026 (European Commission). Textiles and apparel are indicatively first, with a delegated act expected around 2027. Application starts at least 18 months after each act, so enforcement lands around 2028 to 2029. Batteries have their own passport from 18 February 2027. In France, AGEC is the consumer-information lever.
Passport data is usually managed at model level because nothing finer exists. An item-level layer lets a brand attach the right data to the exact unit.
Brands that build digitisation for commercial reasons get compliance as a by-product. Brands that build it for compliance alone miss the commercial return.
A layer used by six teams needs rules, or it becomes six copies of the same data with different definitions.
40+ brands · 12 years · €1.5B protected
Selinko, part of Toppan, gives premium brands the item-level layer their other systems lack: secure identity on the line, intended market at shipment, and verified scans in the field. The questions your teams cannot answer today can become queries.
A product data layer is an item-level record keyed on a unique identifier, such as a GTIN plus serial. It is fed by production, shipment and field-scan events, and it tracks each unit through its life. It sits beside PIM, ERP and CRM, which only describe products, batches and customers.
No. A PIM describes what each SKU is: attributes, content and compliance data. It holds nothing about any specific unit, such as where it was shipped, who scanned it or whether it was resold. Item-level data needs an identity on each physical item and events captured after it leaves the brand.
It links every item to an intended market at shipment, then compares that market with where authenticated scans occur. A genuine unit scanned outside its intended market, traced back to a shipment and distributor, is evidence. Diversion disputes then rest on records rather than on plausibility.
Four elements: a secure carrier on each item, commissioning on the production line, intended market captured at shipment through an ERP or WMS link, and a reason for the field to scan. One product line is enough to start, ideally where diversion or counterfeiting pressure is highest.
A copyable carrier lets counterfeits create records that look genuine, which corrupts every report built on the data. A cryptographic NFC chip such as NTAG 424 DNA signs a new message at each tap, and the platform verifies it. Only genuine items can generate valid records.
The first product line usually takes a few months, because standards, carrier, line commissioning and the ERP integration are set then. Later lines reuse those foundations, so adding one is mostly a question of form factor and content rather than new infrastructure.
One accountable owner, usually in digital or operations, should own data quality, standards and access. Brand protection, supply chain, marketing, quality, client services and compliance all use the layer, so they share definitions and work under role-based permissions.
Talk to us