The real test of the traceability system starts when an email arrives from the raw material supplier saying “There’s a problem with lot L2409-157.” Which parts did that lot go into, which customer did those parts go to on which delivery note, and how many minutes does it take you to say so, with what certainty? If the answer is “We’ll merge the Excel files and have it by evening”, the problem isn’t the label, it’s the design. Below we look at the IATF 16949 traceability requirement from the IT side: number generation, label printing, the data model, the recall query and manual records when the system goes down.
How the traceability requirement reaches IT
ISO 9001 clause 8.5.2 requires outputs to be identified and, where traceability is a requirement, unique identification to be controlled. IATF 16949’s supplementary clause 8.5.2.1 expands this considerably: it requires an analysis of internal, customer and regulatory traceability requirements; a traceability plan based on risk level and failure severity; the ability to identify and segregate nonconforming or suspect product; the ability to meet the customer’s and any regulatory response time requirements; and records to be kept in a format (electronic, hard copy, archive) that enables this time to be met. Identifying individual products by serial number where the customer or legislation requires it is also within this clause’s scope.
Each of these statements turns into a concrete IT design decision from the perspective of IATF 16949 IT requirements:
| Requirement | IT equivalent |
|---|---|
| Risk-based traceability plan | Which products are tracked by serial and which by lot |
| Segregation of suspect product | Status field and shipping block |
| Response time | Query performance and a tested procedure |
| Record format | An archive that stays readable throughout the retention period |
| Outsourced products with safety/regulatory characteristics | Recording the supplier lot at goods receipt |
How long and in what format records are kept is a separate design topic, explained in the guide archiving PPAP and quality records. The focus here is the traceability chain in the live system.
Serial and lot number management
The first decision is the level of detail. For products carrying safety or regulatory characteristics where the customer requires serial-based traceability, each part gets a unique serial number. For other products, lot (batch) tracking may be enough; but the lot definition must be clear: the same shift, the same raw material lot, the same tool setup? If this question isn’t written into the plan, every shift supervisor closes the lot however they see fit. The wider the lot, the wider the recall.
These rules for number generation cut off most problems on the shop floor at the source:
- Single source: Numbers are generated by a central service or database sequence. If local generation per line is needed, each line has a fixed prefix and the ranges don’t overlap.
- No reuse: A number that’s scrapped or cancelled is never assigned again; its status stays “cancelled”.
- Meaningless body, meaningful attributes: Information such as date, line and shift can be embedded in the number, but queries must read this information from separate fields. Reports that extract information from the number format break the day the format changes.
- Check digit: Adding a check digit to the number at every point where it’s entered by hand catches typing errors at the source.
A mistake you may come across is two lines being fed from the same local counter and the counter being reset after a PLC replacement. The same number is printed a second time, and nobody notices, until the customer complaint arrives.
Label and barcode systems
The label is the carrier of traceability data in the physical world, and most errors start here. Label templates are controlled documents: they’re designed according to the customer’s label standard (for example GS1-based labels, the VDA label recommendations common in Europe or OEM-specific label guidelines), managed by revision and not released without approval.
Controls IT needs to set up for IT compliance in automotive manufacturing:
- Source of print data: Label content comes from the MES or ERP record, not from operator input. Permission to edit manually in the label software is kept narrow.
- Reprint control: Every reprinted label is recorded with the user, time and reason. Uncontrolled reprinting is the most common reason the same serial number appears on two parts.
- Print quality verification: ISO/IEC 15415 defines print quality grading methods for 2D codes and ISO/IEC 15416 for linear barcodes. If the customer requires a minimum grade, periodic measurements are made with a verifier. A scanner reading the code isn’t evidence of quality.
- Direct part marking: For DataMatrix codes marked by laser or dot peen, the surface and contrast determine readability; the read rate is monitored, and a downward trend becomes a maintenance trigger.
Scan points are the links in the chain: every critical process step, sub-component matching in assembly, packing and shipping. If you can’t move to the next step without a scan (interlock), the “we’ll enter it later” habit ends. Does it end without an interlock? It doesn’t.
The traceability data model
Traceability works in two directions: backward (from which materials, on which machine and with which parameters was this part produced?) and forward (which parts did this material lot go into, and where did they go?). The minimum model supporting both queries consists of these entities:
| Entity | Key fields |
|---|---|
| Unit (part/serial) | Unit no., product code, status, production time |
| Material consumption | Unit no., material code, supplier lot, quantity |
| Mounting | Parent unit no., child unit no., station, time |
| Process record | Unit no., step, machine, recipe version, result |
| Shipment | Unit no. or package no., customer, delivery note, date |
The packaging hierarchy (part → box → pallet) should be modeled as a separate assembly relationship; otherwise the shipping record shows the pallet but you can’t drill down to the parts inside. If a box’s contents change during repacking, the old relationship isn’t deleted; it’s closed. Having timestamps in all systems tied to the same time source is a precondition for building the correct sequence of events; for technical details, see the note timestamp quality in production data. Record immutability and audit trails are the subject of the guide MES and CMMS record integrity.
The query in a recall scenario
Scenario: a supplier has reported a problem with a specific raw material lot. You need to list, by customer, the units that consumed material from that lot, the parent units those units were assembled into and, ultimately, the units shipped. As the assembly relationship is multi-level, a recursive query is essential. The example below is written in PostgreSQL syntax; in SQL Server, the RECURSIVE keyword isn’t used and UNION ALL is required in the recursive part:
WITH RECURSIVE affected AS (
-- 1) Units that consumed material from the suspect lot
SELECT c.unit_no
FROM material_consumption c
WHERE c.supplier_lot = 'L2409-157'
UNION
-- 2) Parent units these units were assembled into (multi-level)
SELECT a.parent_unit_no
FROM assembly a
JOIN affected e ON a.child_unit_no = e.unit_no
)
SELECT s.customer_code, s.delivery_note_no, s.unit_no, s.ship_date
FROM shipment s
JOIN affected e ON s.unit_no = e.unit_no
ORDER BY s.customer_code, s.ship_date;
It isn’t enough for the query to give the right result; its duration is measured too. If there are no indexes on the child_unit_no, supplier_lot and unit_no fields in the assembly and consumption tables, the query takes minutes on data accumulated over years. The first place to look in the console is the query plan: if you see a full table scan, an index is missing. If old data moved to an archive isn’t in the live database, the procedure must also state which archive is queried and how.
Manual records during a system outage
If the line doesn’t stop when MES, the label server or the network goes down, traceability has to continue on paper. Records improvised during an outage usually end up incomplete and can’t be transferred into the system afterwards. The manual fallback is designed in advance and tried in a drill. An untested procedure doesn’t work. For wider outage scenarios, the guide IATF 16949 cyber attack contingency plan gives this procedure its framework.
Manual fallback checklist:
- Serial number blocks set aside in advance for each line and held in “reserved” status in the system, with spare labels printed from these blocks
- Printed record forms with fields for process step, material lot, operator and time; forms ready at the head of the line
- An outage report recording the start and end time of the outage and the lines affected
- A separate stock area or shipping block for units produced during manual recording
- When the system comes back, entering the data, having the entry checked by a second person and flagging the records as “manual entry”
- Cancelling unused reserved numbers and reconciling the blocks
Checklist
- Has the traceability plan been prepared by product family, with the serial/lot decision written down with its risk justification?
- Have customers’ response time expectations been extracted from CSRs and entered into the plan?
- Is number generation central, with collisions and reuse prevented?
- Are label templates revision-controlled, and are reprints recorded?
- Is barcode print quality verified against the standard the customer requires?
- At critical steps, is moving to the next step blocked without a scan?
- Is the packaging hierarchy in the data model, and can you drill down from a pallet number to the part?
- Is the recall query written down, tested and timed?
- Are manual fallback materials ready at the head of the line, and is the latest drill record up to date?
If you need to assess the traceability chain together with MES, label servers and the OT network, we cover these controls in a single assessment under IT/OT compliance at manufacturing sites. The primary source for label and barcode data structures is the GS1 standards.
Frequently asked questions
Is lot-based traceability enough for IATF 16949?
The standard doesn’t require a serial number for every product; it requires the level of traceability to be planned according to risk and failure severity. If the customer or legislation requires unique serial numbers, serial-based tracking is mandatory. In other cases, lot-based tracking can be acceptable; but the broader the lot definition, the larger the scope of a recall.
Who sets the recall response time?
The response time requirement comes from the customer or the relevant legislation; on the customer side, it’s usually written in customer-specific requirements or the supplier manual. The organization’s job is to write this time into the traceability plan and prove through periodic drills that it can meet it.
Why should label reprints be controlled separately?
Uncontrolled reprinting leads to the same serial number appearing on more than one part, or a scrapped part’s number being stuck on a good part. Recording every reprint with the user, time and reason makes this risk visible and shows the reliability of the traceability chain in the audit.
How are records entered manually during a system outage distinguished?
Records entered when the system comes back are stored with a “manual entry” flag, the outage report number and details of the person who entered and the person who checked them. That way, in a complaint or recall, the source and reliability of these records can be assessed separately.