Unit-level identity: one label fingerprint for five supply chain leaks

In short

Unit-level identity means every single unit carries its own verifiable identity. Idanta reads it from the microscopic structure of the label already on the product, so nothing new is printed or attached. One enrollment on the packaging line lets anyone check a unit with a smartphone and see where it was shipped.

I spend most of my time with heads of supply chain, sales and finance. The conversation usually starts with one leak. A brand protection lead wants to stop counterfeits. A channel head wants to know who is dumping stock on a marketplace. A finance team wants to stop paying credit on returns that should never have been accepted. Those sound like three separate projects with three separate vendors. They share one root cause, and this article is about that root cause and what changes when you fix it.

What is unit-level identity?

Most companies can identify a product at three levels. The SKU tells you what the product is. The batch or lot code tells you which production run it came from. Neither tells you which individual unit you are holding. A batch of 100,000 units shares one batch code, so every one of those units looks the same to your systems.

Unit-level identity adds the missing level. Each unit has an identity that belongs to it alone, recorded when it is made and checkable anywhere after that. Your ERP already knows the SKU. With unit-level identity it can also know where each unit was supposed to go, and whether the unit in front of a distributor, a retailer or a consumer is one you actually made.

Why can't barcodes, batch codes or serialization tell units apart?

Barcodes and batch codes were never designed to. They identify a product line and a production run, and that is useful for stock control. They say nothing about a single unit.

Serialization goes further by printing a unique number on each pack. In the United States, the Drug Supply Chain Security Act aims for what the FDA calls "an interoperable and electronic way to identify and trace certain prescription drugs at the package level." Serial numbers are valuable for tracing. Their weakness is that they are printed information. Anyone with a printer can copy a valid code from a genuine pack onto a thousand fakes. A code proves that a number exists. It cannot prove that the pack carrying it is genuine.

That is why we think of label fingerprinting as a serialization alternative where you have no serialization, and as the missing check where you do. You keep the codes you print today. The fingerprint tells you whether the pack carrying them is the real one.

How does label fingerprinting work?

Every printed label has random microstructure: the way paper fibers lie, the way ink bleeds along the edge of a printed character, the texture of a film surface. Idanta uses features as small as about 20 microns. No printer controls them, including the one that printed the genuine label. Two labels that come off the same press, one after the other, look identical to the eye and are different at that scale.

In security research, a physical feature like this is called a physical unclonable function, or PUF. It is easy to measure and practically impossible to reproduce. Most PUFs are built into chips or added as special tags. Here the PUF is the label you already print.

The process has three steps.

Enroll. A camera on the packaging line reads each label and an edge processor beside the line stores a compact fingerprint in about 100 ms, at up to 1,200 units per minute. Labels can also be enrolled before they are applied, or at the label printer.
Ship. When cases leave the plant, a scan links each fingerprint to the distributor, the date and the invoice. That data flows to your ERP and other systems.
Verify. Product authentication with a smartphone: anyone with an ordinary 20 MP phone points the camera at the label. The match comes back with the unit's identity and its history.

Nothing is added to the product, so there is nothing extra to copy. The rest of this article walks through five leaks a batch code cannot stop, and what the same fingerprint does for each.

How big is the counterfeit problem, and how does unit-level counterfeit detection help?

The OECD and the EU Intellectual Property Office estimate that global trade in counterfeit goods was worth about USD 467 billion in 2021, or 2.3% of global imports. That figure is built from customs seizures, so it counts what crosses borders and gets caught.

The harder cases never look suspicious. In November 2025, police in Bengaluru seized 8,136 litres of adulterated ghee, coconut oil and palm oil. The ghee had been packed in counterfeit sachets and bottles of a leading dairy brand, and distributors holding the brand's own licenses sold it to shops at the genuine price. The pack and the price both looked right, and the seller was authorized. Nothing a retailer could see gave it away.

Most anti-counterfeit packaging adds a feature to the pack: a hologram, a special ink or a tamper seal. Those help until someone copies the feature, and the better ones carry a cost on every unit. A label fingerprint works the other way round. A copied label is a new label with its own random microstructure, so it fails the phone check however good the artwork is. The retailer, the distributor or the consumer gets a clear answer at the shelf, and the brand gets a record of where the fake surfaced.

How do you detect channel diversion and gray market sales?

Gray market diversion is a different problem from counterfeiting. The product is genuine. It is sold in a channel or region you did not authorize, usually because prices differ between markets. Stock meant for one country turns up in another, or units sold to a distributor at a discount appear on a marketplace below your minimum advertised price.

Channel diversion detection usually starts with a test buy. You buy the suspicious unit and learn that it is genuine, which tells you very little. You still do not know which of your distributors released it.

With unit-level identity, the fingerprint was linked to a distributor, a date and an invoice when the case left your plant. One scan of the test buy names the source. That turns a suspicion into evidence you can act on: enforce the contract, adjust allocations or stop shipments to the distributor who leaked the stock.

How do you stop returns fraud?

Returns are a bigger leak than most finance teams expect. Appriss Retail estimated that retailers took back USD 706 billion of merchandise in 2025, and that about USD 100 billion of it, 14.2%, was preventable loss from fraud and abuse.

The patterns repeat. A buyer returns a counterfeit and keeps the genuine unit. A used item comes back as new. Goods bought cheaply on the gray market are returned through your authorized channel for a full refund. In B2B channels, a distributor sends back fakes or out-of-region stock for credit.

Returns fraud prevention gets much simpler when every unit can answer one question: did we ship this unit, and to whom? A scan at the returns desk or the warehouse shows whether the unit is genuine, whether it was sold through that channel and whether it has come back before. You refund, credit or replace only what you actually shipped.

How does warranty abuse happen, and how does unit identity stop it?

Warranty abuse is the quieter cousin of returns fraud. A counterfeit unit fails and is claimed under your warranty. A unit sold in one region at a lower price is claimed in another region with a richer warranty. A serial number sticker is moved from a dead unit to a live one. In each case you absorb the cost of a product you never sold, or never sold there.

Checking the unit's identity at the point of claim closes most of these. A counterfeit has no record. A unit claimed outside its region shows where it was shipped. A sticker can be moved, but the label microstructure underneath stays with the unit it was printed for.

Why are recalls so blunt, and what makes a recall targeted?

When a quality problem appears, most companies can trace it back to a batch at best. So they recall the whole batch, often from every market it reached, because they cannot tell which units are affected or where those units went.

A targeted recall needs two things a batch code cannot give you. You need to know which units shipped to which customer, and you need to be able to line up complaint units against production time. With each unit linked to its route at dispatch, you can group the units behind the complaints, find the shift or line that produced them and pull only those units from the customers who received them. That protects patients and consumers faster, and it keeps good stock on the shelf.

What do India's drug traceability rules require?

India already required a barcode or QR code on the packs of its top 300 drug brands. In June 2026, the Health Ministry notified the Drugs (Seventh Amendment) Rules, 2026 (G.S.R. 506(E)), which extend the requirement to all vaccines, all anticancer drugs and all narcotic and psychotropic drugs from 1 July 2027, and to all antimicrobials from 1 July 2028.

These India drug traceability rules make the code mandatory. They do not stop the code from being copied, and a cloned QR code on a fake pack will scan as valid. Idanta works with the barcode or QR code you already print, so you keep your compliance workflow and close the cloning gap.

Where do connected packaging and first-party data fit?

Every verification is also a moment of contact with the person holding the product. When consumers scan to check that a product is genuine, the brand learns where and when it was bought, and can respond in that moment with offers, rewards or product information. Connected packaging usually means printing a new code or adding a chip. Here the label is the connection, so the first-party data comes without new packaging and without a retailer in between. In markets where counterfeiting is common, that scan is also how a brand delivers deals and promotions to the customers who just confirmed they bought the real thing.

Why does one identity solve all five leaks?

Look back at the five leaks and the question each one asks of a unit. Is it genuine? Where was it supposed to go? Did we ship it to the person returning it? Is it covered by the warranty being claimed? Which units are affected by this problem? Every one of those questions is answered by the same record: a fingerprint captured when the unit was made, linked to where it was shipped.

That is the economic case for unit-level identity. Brands have usually bought a separate fix for each leak: a security feature for counterfeits, an investigation service for diversion, process controls for returns and an insurance mindset for recalls. One enrollment on the line serves all of them, using the label you already print. Idanta is proven in multiple pilots and commercial deployments, and works today on paper labels, polymer film labels, cartonboard, flexible packaging and printed aluminum tins.

If any of these leaks is on your list this year, I would be glad to show you what a scan tells you about your own product.

Frequently asked questions

Do we need new labels, inks or tags?

No. Idanta reads the microstructure of the label you already print. Nothing is added to the product.

What do people need to verify a product?

An ordinary smartphone with a 20 MP or better camera.

How fast is enrollment on a packaging line?

Up to 1,200 units per minute, with about 100 ms of processing per unit at the edge.

Does it replace the QR codes we print for compliance?

No. Idanta works with the barcode or QR code you already print, so you keep your compliance workflow and close the cloning gap.

Which materials does it work on?

Paper labels, polymer film labels, cartonboard, flexible packaging and printed aluminum tins today.