Traceability cannot stop at lot ID—three gates from lot-level to station-level

The auditor only gives fifteen minutes

The client auditor opened the laptop: 'For this finished product batch number, give me the key material batch, key station human-machine info, inspection conclusion, and shipping customer within fifteen minutes.' The quality manager clicked on the 'Traceability' menu—he could find the batch number, but got stuck on the next question: which batch was the material from? Who started the work that day? Are the batch on the shipping label and the production batch the same? On-site, they started exporting three tables and manually stitching them together. Fifteen minutes turned into two hours, and the audit conclusion was written as 'traceability capability insufficient,' not 'the menu has the word traceability.'

Asking suppliers 'Can you trace it?' will always get the answer 'Yes.' This question has no differentiating power. The effective way to ask is with three specifics: How far down can you trace? How long does it take to get results from a single query? Can it support the scope definition of a real recall? Traceability depth is not a switch for a feature; it depends on whether collection points and documents have left records when business events occurred. Moving from batch-level to workstation-level, there are three hurdles; if the hurdles aren't crossed, the traceability page in the system is just an empty shell.

There is another kind of embarrassment: the OEE and production dashboard look very green, but when customer complaints are traced back, they can only be linked to 'a certain shift on a certain day that did a similar product.' Reporting collects the number of times, but does not collect bindings; the route can still skip stations, and the workstation context is even less reliable—traceability and process anti-skip stations, as well as barcode collection, are different angles on the same execution chain.

Why is the batch-level often insufficient?

Accuracy is proportional to the investment. Practical approach: write the consequences of failure and customer requirements as 'three questions that must be answered,' then work backward through the hierarchy. In most manufacturing scenarios, batch-level critical process human-machine binding is already sufficient to support customer complaints and audits. Starting with single-item coding for the entire factory often fails to pass the financial review; meanwhile, key materials end up uncontrolled.

For low-value non-critical uses, single small-batch customizations, and extremely simple product structures, it is not necessary to strictly track at the workstation level; when warehouse batches have not yet been managed, forcing the shop floor to enforce this will only digitize dirty batches. It is recommended to maintain critical material lists (traceable BOM) separately: focus data collection efforts on safety-critical components, regulatory-required components, and materials with high customer complaints, and handle list changes through engineering changes.

The first hurdle: Batch processing of materials — most commonly blocked at the workstation entrance

Mixing the same type of material together and only tracking the total quantity, grabbing from the pile during feeding—then the first question of 'which batch of material was used for the finished product' cannot be answered. The requirements must be: incoming materials must be independently labeled; the entire process of storage, shelving, picking, and feeding must track batches; scanning the material during feeding and binding it to the work order/process; mixing batches should either be prohibited or recorded separately by batch; substitute materials and emergency releases must have independent markings to avoid mixing with normal batches.

If this hurdle is not overcome, nothing else can proceed. The batch is accounted for in the warehouse, but the workshop doesn't scan when feeding materials, effectively stopping at the workstation door. Common failures in skipping steps: only entering the finished product serial number without entering the material batch — this shows a lack of traceability capability, and recalls still rely on guesswork. Another failure is having two sets of codes for warehouse batches and workshop batches, requiring manual translation when querying.

Boundary with WMS: Warehouse receipt, shipment, and storage maintain batches, prohibiting mixed batch picking without records; MES consumes batches at the feeding point and writes into the work order history. Each side exports a table and then manually merges them, which is the standard form of 'tracing in name only, without actual traceability'.

The second hurdle: Human-machine binding — post-event recording is almost ineffective

Know what was done on the Wednesday day shift, but not who or which equipment — the most common gap in audits. Must achieve: workstation login, binding personnel to workstation and process; associating equipment with process, recording start and end of processing; structured shifts, not relying on timestamps and verbal reconstruction.

The difficulty lies in collecting data at the start of work. Supplementary recording afterwards is almost ineffective for review and recall—this is also why tracing projects often get stuck on the discipline of scanning to start work. Collection points should be placed at steps where the status actually changes, and it is forbidden to scan more just to increase the scan rate; if the team leader scans for the whole team and meets the indicators, the significance of binding becomes null.

If the process route allows skipping a station to report completion, the human-machine context will be linked to the wrong station — preventing skipping stations and traceability acceptance should be considered together: it can be collected, and it is collected at a legitimate station. Rework must follow an explicit route; otherwise, isolation and traceability will be broken in 'private rework,' and the stations cannot be clearly identified, making CAPA unmanageable.

When positioning the 'method' in 'man-machine-material-method' (process parameters), it often requires an additional layer of process parameter collection, which is an extra step after the workstation level and should not be confused with the first two hurdles.

The third hurdle: time windows and shipping mapping

The timestamps for material picking, work reporting, inspection, and shipping come from different sources, with some being as precise as a day and others only to the 'morning'—when raw material issues arise, if the granularity is insufficient to determine 'which finished products are affected,' the only option is to broaden the scope, directly increasing costs.

It is essential to ensure: key actions have system timestamps (preferably down to the minute); equipment start and stop times can be cross-verified with manual work reports; shipping batches and production batches are clearly mapped, so that recalls do not break the chain at the shipping stage. Re-labeling shipments with batch numbers unrelated to production batches is a typical breaking point where recall authority is lost—shipping documents must include production batch/serial numbers and be aligned with CRM/WMS outbound records.

Time-window level traceability is a capability achieved after overcoming all three hurdles; it is not something you can get by buying a single module. Retention periods are configured according to customer requirements as system rules (commonly up to one calendar year after discontinuation, and often longer for automotive parts); deletion or export upon expiration comes with audit logs — relying on people to remember 'keep auto parts for two more years' will inevitably lead to omissions.

Forward search and reverse search: two abilities, one output chain

In inspection (Raw materials → Finished products): An anomaly was found in a batch of incoming materials, affecting finished products and shipped customers, used for containment and recall. Reverse lookup (finished product → raw materials)A certain finished product failed, and the materials, processes, personnel, and inspection conclusions were checked, used for customer complaints and determining responsibility.

Both need to achieve a single query chain. It requires manual joining of multiple tables, taking several hours, which has no practical significance on the customer's site or audit site. Judgment criteria: timing from entering the finished product batch number to outputting the complete reverse trace chain; if it often exceeds five minutes, the model and data collection need to be optimized, rather than adding manpower.

Conducting a desktop drill every quarter is more effective at exposing breakpoints than watching vendor demos: randomly select a finished batch (or a returned batch) and perform a reverse check within five to fifteen minutes; then select a key raw material batch and verify within a time limit whether it affects finished products and whether they have been shipped; record the checkpoints—whether there’s a shortage of material, manpower, shipping mapping, or cross-system table export. If the drill fails, do not claim externally that 'workstation-level traceability is already in place.'

How to link JEDP-MES traceability and quality

JEDP-MES product traceability emphasizes: connecting batches/serials across raw materials, processes, quality inspections, and shipments, enabling bidirectional identification of people, machines, materials, and methods; the goal is to advance from batch-level to workstation-level, supporting timelines for customer complaints and recalls. On the production execution side, barcodes/RFID collect start and finish times, personnel, output, and process parameters at key steps; process quality incorporates first-article inspections, patrol inspections, sampling inspections, and defective rework/scrap/quarantine into the execution chain. When combined with JEDP-QMS, inspection standards, defect libraries, CAPA/8D, and incoming/final product inspection evidence are primarily managed by QMS, while MES is responsible for triggering actions and writing results back to workshop execution.

Identification rules must be unified: batch numbers/serial numbers, flow cards, containers, and finished product labels should not conflict. Material feeding, start of work, completion, and inspection events should be recorded in the same history. When key components are not scanned and bound, tracing can only reach to date and production quantity — the list of tracked components should be supplemented first, then discuss the big screen. Phase one connects processes, reporting, and key station collection points; subsequent stages will conduct timed reverse-check exercises for batches/serial numbers. Pursuing error-proofing and traceability without managing process routes will only digitalize chaos.

Suggested progression sequence: Batch key materials and bind to work orders → Scan key processes to start work (human-machine) → Link inspection and non-conforming orders to batches → Synchronize timestamps and map to shipments → Advance sequence/item as needed. The indicators of effectiveness are: the ability to trace materials, the ability to answer common audit questions, and a manageable recall scope — not just having the words 'one item, one code' appear on the menu.

This quarter, track time first, then talk about one item one code

First, write down the three most critical tracing questions and make a gap list against the three checkpoints. Then, time it again: How long does it take to trace back a real customer complaint batch number—if it exceeds the time limit, prioritize fixing the collection rather than adding reports. It's equally necessary to check proactively: when a raw material batch has a problem, can you list the customers who have received shipments within the same time frame—if you can't, the authority to command a recall is not in your hands.

Turn the checkpoints from the two rehearsals into the next-phase collection list; this is closer to usable capability than buying another traceability big screen. A big screen cannot solve the problem of materials not being scanned. Cross the obstacles first, then talk about rolling out one item-one code across the entire factory—the sequence is reversed, and serial numbers will just become more expensive batch labels. When batches cannot even be linked, adding serial numbers only hides the breakpoints in more detail; only after overcoming the three obstacles does the serial-level become an enhancement rather than a lifeline.

• MES manufacturing solution
• Scan code to report work collection point
• Process Anti-Skip Station and SOP
• QMS quality solution
• Workshop Work Reporting and Process Traceability Self-Test