Common traceability mistakes: the breaks in the chain that leave a fault unfindable

Batch codes vanish in ordinary ways: printed on outer packaging that buyers discard, printed in ink that fades or rubs off, placed on a removable part that is replaced, or recorded only on a supplier document the seller never filed. By the time a fault appears, the code that would identify the affected units is gone, and the seller must recall a wider range than necessary, or cannot recall precisely at all.

Durable coding means placing the code where it survives normal use and disposal patterns: on the product itself where feasible, in a form that resists wear, and in the sales record as well as on the product. A code that exists in two places, product and record, survives the loss of either. A code that exists in only one place is a single point of failure for the whole trace chain.

Traceability stops at the warehouse because inbound records are kept by shipment, while outbound records are kept by order, and nobody links the two. The warehouse knows which shipment arrived, and which orders shipped, but not which shipment supplied which order. When a fault is traced to a shipment, the seller cannot say which buyers received it, and must contact a wider group or guess. The wall is a linking failure, not a recording failure, because both sides of the wall hold records.

Breaking the wall needs a picking discipline: orders are picked against specific batches, and the batch picked is recorded against the order. That discipline, enforced by the warehouse system or by a simple picking sheet, turns two separate records into one chain. Illustration: a seller of spice jars adds a batch field to its picking sheet, and for the first time can trace a faulty spice lot to the exact buyers who received it.

Mixing batches happens when units from different production runs, supplier lots, or deliveries are stored together, picked interchangeably, or repackaged together without preserving batch identity. Once mixed, the units cannot be separated by batch, and a fault in one batch implicates the whole mixed stock. Mixing is often done innocently, to save space or to top up a shelf, but its traceability cost is severe, because it destroys the very distinction a recall needs.

Preventing mixing needs storage discipline: batches stored separately, labelled, and picked in a way that preserves identity, with repackaging that carries the batch forward rather than discarding it. Where mixing is unavoidable, such as bulk ingredients, the trace record should capture the mix itself, so the affected range can still be defined, even if it is wider than a single batch.

Recording by model tells the seller what type of product was sold, but not which units, from which batches, went to which buyers. A recall or a complaint investigation needs unit level answers: which batch, which buyer, which date. Model level records force the seller to treat every buyer of the model as potentially affected, which is inefficient and alarming, or to guess, which is unsafe. Unit level recording, by batch at minimum, is the standard a trace system must meet.

Unit level recording does not require expensive systems. A sales record that includes batch code, picked from the picking sheet, already provides it. The failure is usually not technology but habit: sales recorded quickly, without the batch field, because the batch field felt like extra work. That extra work, done at sale, saves enormous work during a recall, when batch information is needed urgently.

Supplier blindness means the seller cannot trace a product back to its supplier lot, component batch, or production date, because supplier documents were never collected, never linked to incoming shipments, or filed without batch detail. When a supplier reports a component fault, the seller cannot say which of its products contain that component lot, and must either recall widely or wait for complaints to reveal the scope. Both options are worse than a precise trace.

Fixing supplier blindness needs an incoming discipline: supplier lot recorded against each incoming shipment, supplier documents filed by lot, and component lots linked to finished batches where the product's design makes that link meaningful. That discipline should be part of supplier terms, so suppliers provide lot information routinely rather than on request during a crisis.

Testing a trace chain means running a mock trace: pick a sold unit, or a supplier lot, and try to follow it in both directions using only the records, within a set time. The test reveals breaks quickly: a missing batch, a warehouse wall, a mixed shelf, a supplier document never filed. Each break found in a test is a break fixed cheaply, while each break found during a real fault is a break fixed under pressure, with buyers waiting.

The test should be run periodically and after any change to warehousing, systems, or suppliers, because trace chains break when processes change, not only when they are first built. A chain tested regularly stays intact; a chain assumed intact, without testing, will usually prove broken at the moment it is most needed.