One product record, from the mill to the repair bench

This note is a draft. Items in square brackets are still to be confirmed.

I grew up in a knitwear factory that supplied luxury houses. The thing I learned first, before I could have named it, is that a luxury product is a set of claims. This yarn, this gauge, this construction, this finish, this care. The house makes those claims to its client. The factory either makes them true or it does not. Everything in between, from the tech pack to the window display, is a handover of those claims from one function to the next.

Later, developing product at Burberry and at adidas, I sat on the other side of the same handovers, and watched the claims change shape as they travelled. Not through dishonesty, but through the ordinary friction of seven functions each keeping their own version. Faster communication spreads that inconsistency; it does not correct it.

[One real handover, anonymised: a yarn or construction claim that changed between the tech pack and the campaign, or a care label that did not match the lab result. What was claimed, where it changed, what it cost. Two or three sentences.]

An illustrative failure

A campaign describes a coat as suitable for heavy rain. The product record, the one the technical team keeps, says water resistant. Those are different claims. A client who read the campaign buys the coat, wears it through a storm, and asks a service assistant how to care for it. The assistant, or the chatbot the assistant has been given, invents care instructions, because the care restrictions were never attached to the record it can see.

Nobody in that chain lied. Marketing used the warmer word, development used the accurate one, and service filled a gap. Now put generative AI on the campaign copy, a recommendation engine on the product and an assistant on the care question. Each system works from whichever version of the claim it was pointed at. The technology did not cause the failure. It made the failure faster and harder to trace.

What technology may carry, and what it may not

The chain runs from creative intent through product development, sourcing and production, merchandising, marketing, retail and clienteling, to aftercare. Development converts the idea into specification, materials, construction and care requirements. Production substantiates them or sends back the ones that cannot be met. Merchandising decides which products reach which location. Marketing translates the proposition into imagery and language. Retail explains it to one person. Aftercare tests whether the promise survives use.

Each link involves a decision made by a person, and the next note is about keeping it that way. What technology can legitimately do is carry trustworthy information between the links. That requires one thing the chain usually does not have: a shared, versioned product record beneath every function. Material, construction, fit, origin evidence, availability, care restrictions, repair eligibility, and the owner of each claim by name, so that when the campaign says heavy rain, someone is responsible for the gap. Client data stays in a separate record with separate rules, and never inside a public product passport.

The last link

Burberry describes its own model as source, design, make, sell, with aftercare inside the definition of selling. In the year to March 2026 it reports 38,110 products repaired, more than 41,000 repaired or refreshed in all, and 385 stores across 32 countries and territories offering aftercare. A retail concept that ends at the till has left out the last link of the company's own model.

Ownership is where the promise is tested hardest, in three moments. Wearing and maintaining, where the record has to hold the care restrictions the client was actually promised. Repair or alteration, where a craftsperson needs to know what the item is made of and whether it is eligible. Transfer, where an authentication workflow needs records that move with the item, where the market allows.

The loop most houses never close runs from the repair bench back to the design table. The repair bench knows which seam fails and which lining pills. Product development rarely hears it. One record, with repair eligibility and repeat faults attached to the item, is how it would. That is the first place I would look for a return on a product record, before any passport or assistant: fewer repeat faults in next season's specification.

Why this is a development problem first

The temptation is to buy the passport, the clienteling tool or the assistant, and let the record follow. In my experience it works the other way round. Technology only carries what it is given, and a house whose development function already keeps one truthful record and hands it on intact has something reliable to carry. A house that does not will buy systems that give confident, wrong answers.

So before the vendor conversation, ask a development question. For one product, can every function see the same specification, the same care restriction and the same repair eligibility, with a named owner for each? If not, that is the first change. It costs less than any system, it is the simpler change the first note kept pointing at, and without it every system you buy inherits the gap.

---

Sources