Unit-level traceability sounds like bureaucracy until the first field issue. Then it is the difference between recalling forty units and recalling four thousand.
The case for serial traceability is a containment calculation. A field failure arrives with a serial number. Without traceability, the suspect population is everything you shipped since the last known-good point — often months of production. With it, the population is the units that share the failing unit’s component batch and process window. We have seen this arithmetic turn a four-thousand-unit recall scare into a forty-unit exchange.
What traceability means in practice: each unit carries a serial applied at the start of assembly, and that serial accumulates a record — the work order, the BOM revision, the component batches consumed from kitting, the test results with measured values, the operator and date. The label is the visible part; the linked data is the product.
The cost side is honest but modest when built into the line. Serialisation adds a label and a scan; batch capture adds a scan at kitting; test-data logging is configuration on a rig that exists anyway. The real cost is discipline — a line that sometimes skips a scan has approximate traceability, which fails precisely when precision is needed.
Component batch capture is where the value concentrates, because components are where field failures usually originate. When a capacitor manufacturer announces a suspect date code, the question "are we affected, and which units?" is answered from the kitting records in an afternoon. Without them, the answer is either "all of them" or an untruth.
Traceability also changes supplier conversations in quieter ways. Warranty claims resolve on data instead of assertion: the returned unit’s test record shows it left the line within limits, or it shows a marginal value that should tighten the limit. Both outcomes are cheaper than the argument, and both improve the product.
Our advice on scope: match depth to consequence. Full unit-level trace with component batches for products where failure is expensive or safety-relevant; batch-level trace for the rest. What we counsel against is the middle path of serial labels without linked data — it advertises a capability that will be tested, once, at the worst possible moment.
— GANI Manufacturing engineering team