Circularity Is a Data Problem
You cannot repair, reuse or recycle what you cannot identify. Before circularity is a materials strategy it is an identity, traceability and reverse-logistics system — and most organisations have none of the three.
Circular-economy programmes tend to be framed as a materials question: which polymer, which alloy, which supplier. That framing skips the constraint that actually blocks most of them. When a used product arrives at a facility, the operator has to determine what it is, what is inside it, what condition it is in, and where it should go next. If that takes more than a few seconds per unit, the economics collapse and the unit is shredded regardless of what it was made of.
Circularity is an identity and traceability problem first.
Identity has to survive the product life
A serial number on a peel-off label does not survive a decade in a plant room. Identity has to be attached in a way that outlives the first owner: laser marking, data-matrix codes on the casting rather than the sticker, an RFID tag or NFC chip where cost allows, or a fingerprint derived from the device itself where it does not.
The identifier should resolve to a record without requiring the resolver to guess the manufacturer. That is the practical argument for standards-based schemes over proprietary ones — the party doing the recovery at end of life is almost never the party that sold the product, and will not integrate with a bespoke lookup service for a single brand.
The record must be component-level
A bill of materials at product level tells a recycler almost nothing useful. What determines the route is component-level information: which subassemblies exist, which are known wear items, which contain regulated substances, which are worth recovering intact, and how they come apart.
That last point is where design and data meet. Disassembly instructions are part of the record, and they are only writable if the product was designed to be disassembled. A unit that requires destructive removal of a bonded housing to reach the battery has a very short decision tree at end of life, and no amount of data improves it.
The record also has to be versioned. Products change mid-production, and a record that describes only the current revision will misroute every earlier unit. Bind the record to the serialised unit, not to the product line.
Condition and provenance accumulate
The interesting part of the record is written after the product ships. Repairs performed, parts replaced, firmware revisions applied, operating hours, fault history, environment. That history is what separates a unit worth remanufacturing from one worth harvesting for parts.
Capturing it requires the service network to write back, which is an organisational problem more than a technical one. Field technicians will record what takes seconds and skip what takes minutes. The system has to be designed around that reality: scan the identifier, pick from a short list, done. A form with eighteen fields will be filled with defaults, and defaults are worse than blanks because they look like data.
Provenance also carries regulatory weight. Where recycled content or substance restrictions must be evidenced, an unbroken chain from material to unit is the evidence. Reconstructing it after the fact is not possible, which is why the chain has to be designed before the first unit ships.
Reverse logistics is a routing system
Forward logistics moves known goods from few origins to many destinations. Reverse logistics moves unknown goods from many origins to few destinations, and the routing decision depends on information that is only available once the unit is identified.
Practically, that means the routing decision should be made as early as possible — ideally at the point of collection, from a scan, rather than after the unit has been shipped to a central sort. Every kilometre travelled by a unit that turns out to be unrecoverable is cost with no return. Triage at the edge, with the identity record available offline, is worth building.
The same system has to handle exceptions gracefully, because a meaningful share of returns will be unidentifiable — damaged marking, missing record, grey-market unit. A route for the unknown is part of the design, not a gap in it.
What to build first
Start with serialisation and the resolution service, because nothing downstream works without them. Then the component-level record, bound to the serialised unit and versioned. Then service write-back, designed for a technician holding a scanner in one hand. Reverse-logistics routing comes last, because it consumes all three.
None of this is a pledge, a target or a report. It is an operational system, and it is the part that has to exist before the materials strategy means anything.