When a defect reaches a customer, the first questions are which other products are affected and where the cause came from. A factory without traceability answers with a guess and recalls far more than it needs to, or not enough. Customers, especially in automotive, medical devices and food, require it, and some industries have legal requirements. This skill designs traceability that can answer those questions in hours, sized to the factory's systems.
Write in Simplified Chinese unless asked otherwise. Industry and customer requirements differ: food production under the Food Safety Law and its rules, medical devices under the UDI (Unique Device Identification) requirements, automotive under IATF 16949 and customer-specific requirements, and others. These are named for orientation and marked 需核实; the plant confirms what applies to its products with its quality and regulatory teams. For writing up a specific complaint, use cn-8d-report.
Required Inputs
Ask for these if not provided, otherwise infer and label the assumption:
- Products and industry: what is made, for whom, and any regulatory or customer requirements known
- Process flow: from incoming materials to shipment, including outsourced steps
- Current systems: paper travellers, Excel, ERP, MES, WMS; barcode or QR capability
- Volumes: units per day, lot sizes, number of SKUs
- Pain: the last time a trace was needed and how long it took, or a customer audit finding
- Budget and timeline
Output Structure
1. Traceability objectives and scope
What questions the system must answer and how fast (for example: given a customer serial number, find the material lots and process data within 2 hours; given a bad material lot, find every finished product and customer within 4 hours).
2. Traceability units and coding rules
Unit level (single piece, batch, box, pallet) chosen per product with reasons; the coding rule (for example product code + production line + date + shift + serial number), label type (barcode, QR, DPM laser mark), and where it is applied.
3. Data collected at each process step
| Process step | Collected content | Method | Record location | Responsible person | Incoming lot and supplier, work order, equipment and moulds or fixtures, key parameters, operators, first-article and in-process inspection results, rework and scrap, packing and shipment to customer. Includes the link between material lots and work orders (the most common gap).
4. Forward and backward traceability paths
- Backward (反向): from a finished product or customer complaint back to materials, process and people
- Forward (正向): from a suspect material lot or process event to every affected product and customer Each as a step-by-step query in the current system, showing where it breaks today.
5. Record retention
What is kept, in what form, for how long (the product life plus a margin, or the period required by the customer or regulation, subject to verification), and backup.
6. Recall drill
A mock recall drill: the scenario, the start and stop time, who takes part, the target time, and what is measured (completeness of the lot list, time to answer). Run at least yearly or as the customer requires.
7. Gaps and phased implementation
| Gap | Risk | Measure | Phase | Cost estimate | Phase 1 with paper or Excel where needed, phase 2 with scanning, phase 3 with MES integration; quick wins first.
Quality Checks
- The target questions and response times are stated
- Trace unit and coding rule are defined per product
- The material lot to work order link is captured
- Both forward and backward paths are written as concrete queries
- A mock recall drill has a scenario and target time
- Regulatory and customer requirements are marked as requiring verification
Anti-Patterns
- Buying an MES before defining what to trace. Software without a data model gives fast answers to the wrong questions.
- Lots so large the trace is useless. A lot that is a month of production recalls a month.
- Mixing lots at a step without recording it. Shared bins and silos break the chain.
- Never testing. The first real recall is a bad time to find the gaps.
- Outsourced steps left out. The subcontractor's lot data must link back.
Example Trigger Phrases
- "We make automotive wiring harnesses, and the customer requires traceability to every batch of raw material. Help me design a traceability plan."
- "After a quality problem, we searched for three days and still didn't know which batches were affected. How do we establish traceability?"
- "Help me design a recall drill."
- "How do we implement one-item-one-code traceability in a food factory?"
- "Design a batch traceability plan for a Chinese electronics factory."