A collector taps the NFC tag on a second-hand watch in a shop in Lisbon. The page confirms the watch is genuine. It also receives an IP address, a device signature and a rough location, and nobody has decided what happens to them. Privacy in a connected product programme is settled at that first tap: in what the page collects, what it keeps and what it asks. Brands that design for it get data every team can use. The others inherit a liability.
This is practical guidance, not legal advice. Each programme should be reviewed with the brand’s privacy counsel and data protection officer. The focus is Europe, where GDPR sets the strictest common baseline.
A connected product travels. It changes hands. It is tapped by people who never bought it from the brand, in markets the brand may not sell into, by an owner who may be the third or fourth to hold it.
Several drivers make this a design problem:
It means the experience is built so that the first tap works without identifying anyone, and the owner chooses when to become known.
Three decisions define it. What the landing page stores on the phone. How precisely location is recorded. When the owner is asked for details, and what they receive in return.
The same rule applies to the tag. A secure NFC tag carries a product identifier and cryptographic material. It carries no name, no email and no preferences. Personal data belongs in a separate, access-controlled system, linked only if the owner registers.
It is tempting to call a tap anonymous. That is not accurate, and designing on that assumption causes problems later. A tap usually creates three kinds of data.
| Data type | Examples | Personal data under GDPR? | Design response |
|---|---|---|---|
| Product data | Unit identifier, batch, model, authenticity result, tamper status | No, on its own | Keep for the product’s life |
| Technical data from the tap | IP address, device and browser details, precise location if requested | Often yes | Derive country or city, then discard or minimise the raw data |
| Owner data | Name, contact, ownership, preferences | Yes | Collect only with a purpose and lawful basis, store separately |
A scan event holding the unit, the time, the country and the authenticity result is easy to use, share and retain. One that also stores a raw IP address and a device fingerprint is not. Location is the clearest example, and it is also one of the most valuable signals, as scan location analysis shows for diversion detection. Country or city level is enough for that purpose.
Collect the country, not the coordinates. Keep the verdict, not the fingerprint.
The process below uses an NTAG 424 DNA-class chip (ISO/IEC 14443-A, NFC Forum Type 4, AES-128) with Secure Unique NFC (SUN) messaging.
The authentication decision rests on cryptography, not on profiling the person who tapped. That is the privacy advantage of the mechanism.
Different uses rest on different legal grounds. Mapping them early avoids the common mistake of asking for consent for everything, which weakens both the experience and the legal position.
These are typical positions to review with counsel:
Consent, where it applies, must be specific, informed, freely given and as easy to withdraw as to give.
| Stated | Verified |
|---|---|
| One checkbox for registration, marketing and analytics | Registration and marketing asked separately |
| Pre-ticked boxes | Active opt-in, with a clear value for each: certificate, warranty, refill reminders |
| Authenticity result hidden behind consent | Core features work without marketing consent |
| No record of what was agreed | Consent logged with time, wording and channel, withdrawable from any message |
For how consented interactions become usable customer insight, the guide on turning product interactions into actionable customer data covers the next step.
Both can be built cookie-free and both can avoid collecting personal data. The difference lies in how much the brand has to infer.
| Criterion | Secure NFC (NTAG 424 DNA, SUN) | Serialised QR |
|---|---|---|
| Identifier per tap | Changes at every tap, verified server-side | Static, can be photographed and reused |
| Basis of the authenticity verdict | Cryptographic check | Scan patterns and anomaly detection |
| Personal data needed to detect copies | None | Often more: repeat scans, locations, devices |
| Reading | Deliberate tap, phone in contact | Camera at distance, scannable from a photo |
| Unit cost | About $0.45 to $0.65 for NTAG 424 DNA-class labels at five-figure volumes; on-metal adds 7% to 30% (public catalogues, 2026) | Near zero marginal print cost; the real cost is serialisation and platform (industry sources, 2026) |
The distinction that matters: A static code proves that a number exists. A secure tag proves that a physical item was present at the moment of the tap. The second needs less surveillance of the consumer to reach a reliable answer.
For Digital Product Passport access, the logic differs. The ESPR (Regulation (EU) 2024/1781) is carrier-neutral, but a GS1 Digital Link QR is emerging as the practical default for the visible carrier. NFC for trust, visible carrier for compliance. The central EU DPP registry has been live since 20 July 2026 (European Commission), and passport content is product data, so it can be kept apart from owner data.
A serialised QR wins in several honest cases.
Many programmes combine the two: a visible QR for access and compliance, an NFC tag for products where trust and resale matter.
Resale is where connected products differ most from ordinary digital services. The design has to protect both people.
The item history can say “transferred twice since 2025” without saying who owned it. That is usually all a buyer needs. Ownership certificates built this way keep the owner’s identity private while the transfer remains verifiable.
Partners need clear roles. A typical split: the brand is the controller of programme data, the platform provider is a processor acting on its instructions under a data processing agreement, and a resale platform verifying items is a separate controller for its own customers. Each arrangement should be confirmed with counsel.
The simplest rule is to share results, not people. A partner needs to know the item is genuine and perhaps that it has been transferred. It does not need to know who owns it.
Rights requests become routine when the architecture supports them. Owners can ask for access, correction, erasure, portability or an end to marketing. Keeping product events and personal data in separate structures, linked only through registration, lets a brand erase a person without destroying the product’s history. That history may be needed for authentication, warranty or a passport long after the owner has gone. The same separation of layers is described in the connected product architecture guide.
Outside the EU, personal data needs a transfer mechanism, such as an adequacy decision or standard contractual clauses. Regimes such as CCPA/CPRA, PIPL and LGPD need local review. Minimising personal data at capture is the most effective protection across all of them.
A data protection impact assessment is required when processing is likely to create a high risk to people, for example large-scale monitoring or systematic profiling. Programmes that combine location over time, ownership and marketing attributes at scale can approach that line. Running one early forces purposes, data flows, retention and partner roles onto paper while they are cheapest to change.
| Sector | Privacy consideration |
|---|---|
| Wine and spirits | Age gates and alcohol marketing rules vary by market; coarse location is enough for diversion analysis |
| Watches and jewellery | High-value owner data needs strong security; owner identity must never appear in resale |
| Cosmetics and fragrance | Replenishment patterns are behavioural data, so base journeys on consent |
| Leather goods and fashion | Passport data is product data; keep owner data out of passport views |
| Cigars and furniture | Long ownership cycles and gifting make transfer and retention rules central |
Physical constraints matter too. Standard tags detune on metal, so watch cases and metal closures need on-metal tags with a ferrite layer. That is a hardware choice, but it affects privacy indirectly: a tag that reads reliably removes the temptation to fall back on weaker, more data-hungry methods.
40+ brands · 12 years · €1.5B protected
Selinko, part of TOPPAN, designs connected product programmes so the first tap is useful without identifying anyone, and owners choose when to become known. Authentication and diversion analysis run on product data; certificates and marketing run on consent.
A scan is personal data when it can identify a person. The product identifier and authenticity result alone are not. The IP address, device details and precise location sent with the tap often are. Minimising these at capture, for example keeping country or city only, reduces the exposure.
Consent is needed for some uses, not all. Marketing emails and tracking cookies generally require it. Authentication and fraud prevention can rest on legitimate interests, and ownership certificates on contract. Each use needs its own lawful basis, reviewed with counsel, rather than one blanket consent request.
It can, but a well-designed programme does not do so. A secure tag such as an NTAG 424 DNA holds an identifier and cryptographic keys. Personal data stays on a server under access control, linked to the item only if the owner registers. This also allows erasure without touching the tag.
Unlink it and never display it. The previous owner’s details are separated from the item, kept only for a defined retention period such as a warranty window, then deleted or anonymised. The item history records that a transfer occurred, without revealing who the earlier owners were.
Yes. Partners should receive results, not identities: whether the item is genuine and whether it has been transferred. Role-based access, written agreements and clear controller and processor roles keep staff scans separate from owner records. Resale platforms should exchange verification results, not owner identity.
A DPIA is required when processing is likely to create a high risk, such as large-scale monitoring or systematic profiling. Programmes combining location over time, ownership and marketing attributes can approach that threshold. Running one before launch is useful even when not strictly required.
Usually not. Passport content describes the product: materials, origin, repair and recycling information. It becomes a privacy question only if owner data is mixed into passport views, which good design avoids. Keep product data and owner data in separate structures.
Other regimes apply, including CCPA/CPRA, PIPL and LGPD, each with its own rules on notice, consent and transfers. Personal data processed outside the EU needs a transfer mechanism. Local review is advisable, and minimising personal data at the first tap lowers the number of cross-border questions.
Review your programme