Reconditioning with a New Label and Shipping? —— The Compliance Boundaries and Five-Step Control Method for Rework and Repair under the IATF 16949 Framework

By: QTank Published: 9/25/2026 Views: 15
Current rating: ★★★☆☆ Rate this Equivalent to 8 ratings

A certain automotive parts company was undergoing a second-party audit by a customer. The first two days went smoothly. On the morning of the third day, the auditor stopped at the rework station: an experienced employee was using a file to remove burrs from a batch of stamped parts. There were no work instructions or rework records at the station, and the周转箱 (turnover box) beside it contained a mix of reworked parts and newly molded parts, all labeled as "待处理" (to be processed).

The auditor asked three questions that no one on-site could answer: Why was this batch being reworked? Who confirmed its conformity after rework? Did the customer know about and approve the repaired parts?

That afternoon, the audit report included two additional nonconformities: the rework process lacked executable work instructions and records; the repaired products were not approved by the customer as required and lacked identification and traceability. The quality manager felt aggrieved — "Rework is to ensure delivery, how did it become a nonconformity?"

In the language of the system, rework is never a "trivial internal matter." It is a redefinition of the product status, inspection conclusion, and customer commitment.

1. Rework and Repair: A Difference in Words, Three Legal Responsibilities

ISO 9000 defines these two terms with only a few characters' difference, but this difference results in entirely different product statuses and customer obligations.

Rework (rework): Measures taken to make nonconforming products or services conform to requirements. Note the phrase "conform to requirements" — after rework, the product should fully meet the drawings, specifications, and technical conditions, equivalent to a product that has never been nonconforming.

Repair (repair): Measures taken to make nonconforming products or services meet their intended use. "Intended use" does not equal "specified requirements." After repair, the product may have dimensional deviations, slightly reduced strength, or visible repair marks. It can be used, but it is no longer the product as originally designed.

Concession (concession): Permission to use or release nonconforming products or services. This is a deviation approval that requires authorization and typically customer approval.

The differences in these terms translate into the following hard distinctions on the shop floor:

Dimension Rework Repair Concession
Product Status Restored to full compliance Meets intended use but not fully original requirements Still nonconforming but permitted for use
Customer Obligation Notify customer if required Typically must obtain customer approval Must obtain customer approval
Risk Analysis Risk assessment before rework Must conduct risk analysis, including life and safety Customer determines acceptability
Identification and Traceability Reworked parts must be traceable Repaired parts must be identified and traced separately Must be noted in records with the scope and batch of the concession
Typical Examples Deburring, retightening, cleaning oil Welding and reprocessing, dimensional adjustment after assembly Appearance color difference within limits

Most companies lose control over rework and repair due to the misuse of these three terms: on-site, they are clearly doing repairs (welding, secondary processing, dimensional adjustment), but all records are labeled as "rework," thus quietly erasing the obligation to obtain customer approval.

IATF 16949 elaborates significantly on clause 8.7: rework must have a documented rework confirmation process, reworked products must be reinspected, traceability of reworked products must be maintained, and risks must be considered before rework. For repairs, it goes even further — risk analysis, identification, traceability, and customer approval must be ensured. These are not just documentation requirements; they correspond to three real risks: rework damaging the product, repaired parts ending up where they shouldn't, and customers unknowingly accepting products that deviate from the design.

2. Why "Rework to Ensure Delivery" Can Become a Big Hole

Rework is essentially a "mini production process" that has not been validated. It is not in the control plan, not in the PFMEA, and often not in the work instructions, yet it directly determines the status of the products leaving the factory. This brings four hidden dangers.

Hidden Danger One: Rework can introduce new defects. Secondary heating may alter material structure, repeated disassembly and assembly may damage threads and clips, manual deburring may change wall thickness and flatness, and welding may introduce heat-affected zones. Without a risk assessment before rework, no one has ever asked, "Will this rework turn a functional issue into a safety issue?"

Hidden Danger Two: Post-rework inspection is often downgraded. The first article inspection and in-process inspection for normal production have clear requirements, but reworked parts are often only given a "visual confirmation." An electronics manufacturing company once only performed visual and continuity checks on reworked connectors without retesting the insertion and removal force. Three months later, the customer found batch contact failures during whole machine vibration testing.

Hidden Danger Three: The traceability chain is broken. If reworked parts are mixed into normal batches, once a batch complaint arises, the company cannot distinguish which parts were reworked or how they were reworked, leading to a full batch recall.

Hidden Danger Four: Lack of data loop closure. Rework rates are not included in process performance metrics, the occurrence rating in PFMEA has not changed in three years, and the detection methods in the control plan have not been upgraded. Rework has become a cover for insufficient process capability, hiding the true Cpk and true risk of nonconforming products.

3. Five-Step Control Method for Rework and Repair

Step One: Define the Authority and Criteria for Judgement.

Create a decision table that specifies when to rework, when to repair, and when to scrap or seek a concession, based on specific nonconformities. The decision table should include at least three columns: nonconformity mode, allowable disposition, and approval authority. The key rules are: the authority to judge repairs cannot be delegated to the team, it must be jointly confirmed by process and quality and documented; any repair involving customer-specified special characteristics or safety characteristics must be handled as "customer approval required" before proceeding, not after an incident.

The decision table should also include a "prohibited repair list." Some defects are inherently non-repairable by design — such as safety characteristic fractures, cracks after heat treatment, and seal structure damage. Posting this list at the rework station is more effective than post-incident accountability.

Step Two: Write an Executable Work Instruction for Each Type of Rework.

Generic work instructions for rework have no value. Effective rework instructions must clearly specify six things: scope of application (nonconformity mode and product family), rework steps and process parameters, required tools and gauges, qualifications of rework personnel, environmental and protective requirements for rework, and inspection requirements after rework.

Each rework instruction corresponds to a specific nonconformity mode. A certain automotive parts company identified 12 common nonconformities and wrote rework instructions for each, reducing the rework nonconformity rate from 6.4% to 1.8% in the same month — the process of writing instructions exposed many steps that relied on "feel."

Step Three: Set Control Points Before and After Rework.

Before rework, conduct a risk analysis: assess the impact of the rework process on product function, life, appearance, and subsequent assembly, and determine whether rework might turn a "repairable defect" into an "undetectable hazard." Defects with unacceptable risks should not enter the rework process.

After rework, conduct a re-inspection, which should be executed in three levels based on the depth of the rework impact:

Level Applicable Scenario Inspection Requirements
One Surface and cleanliness rework, no impact on dimensions and function Recheck affected items according to original inspection specifications, retain records
Two Rework involving dimensions, assembly, and electrical performance Recheck all items according to original inspection specifications, redo functional tests if necessary
Three Repairs involving welding, heat treatment, and material state changes Recheck all items and verify performance or reliability, obtain customer approval

Reinspect according to the "original inspection specifications" rather than the "post-rework confirmation requirements." This should be clearly stated in the documents; otherwise, the field will always default to the first level.

Step Four: Ensure Proper Identification and Traceability.

Three actions are essential: use dedicated turnover tools and labels for reworked parts, with the label noting the product, batch, defect, and rework date; do not mix reworked parts with conforming parts before rework is complete; reworked parts should only be in the "rework in progress" status, not "conforming pending warehousing"; rework records must be traceable to the original nonconformity records, the rework executor, the post-rework inspection conclusion, and the release approver.

The identification of repaired parts should be more prominent — their product status is "customer-approved deviation," and the label should include a repair mark or the customer's approved number. A certain automotive parts company once shipped a batch of welded repaired parts without identification, leading to a full production line shutdown when the customer found abnormal weld seams. The batch was ultimately treated as nonconforming, with costs far exceeding the value of the parts.

Step Five: Integrate Rework Data into the System.

Rework rate, repair rate, and rework re-nonconformity rate should all be included in the process performance dashboard, broken down by product family and process, and reviewed monthly for trends. High rework rate processes should trigger three actions: re-evaluate the occurrence and detection ratings in the PFMEA, re-evaluate the control methods in the control plan, and assess whether additional poka-yoke or tooling modifications are needed.

Rework losses should be accurately recorded as internal failure costs in the quality cost, not just as scrap. The combined cost of rework labor, tool wear, secondary inspection, and delayed delivery is often two to three times the scrap cost. Calculating this is not for blame but to provide a basis for decision-making on whether to improve the process or continue reworking.

4. A Reference Improvement Process

A certain automotive parts company (stamped and welded parts, monthly production of about 200,000 units) had a first-time yield of around 91% for a long time, with a rework rate of 7.5% and a repair rate of 1.5%. The rework loss was about 180,000 yuan per month, but 30% of customer complaints were still related to reworked parts.

After the customer audit issued nonconformities related to rework and repair, the company took four actions over three months: first, they developed a rework and repair decision table and a prohibited repair list; second, they wrote rework work instructions for 12 common nonconformities and completed the risk analysis and customer approval process for three types of welding repairs; third, they conducted a full re-inspection of reworked parts according to the original inspection specifications and added a seal integrity test for welded repaired parts; fourth, they integrated rework data into process performance and implemented poka-yoke modifications for the two processes with the highest rework rates.

Three months later, the rework rate dropped to 3.1%, the rework loss to about 80,000 yuan per month, and complaints related to reworked parts were eliminated. More importantly, the rework data exposed process capability bottlenecks: the occurrence ratings in the PFMEA for two processes were increased, and online inspections were added to the control plan, changes that would never have been triggered if rework rates were not tracked.

5. Five Common Pitfalls

Pitfall One: Recording repairs as rework. Misusing terms means waiving the obligation to obtain customer approval, which is the easiest for auditors to catch.

Pitfall Two: Using a generic rework procedure for everything. Rework is a process activity that must have process parameters and inspection requirements; a generic procedure document cannot replace work instructions.

Pitfall Three: Storing reworked parts with conforming parts in the same batch. Once the traceability chain is broken, any issues will require a full batch handling, doubling the cost.

Pitfall Four: Only performing visual confirmation after rework. Rework affects the product status, not just the surface. The inspection scope should be determined by the depth of the rework impact.

Pitfall Five: Only tracking yield, not rework. Counting rework as "conforming" makes the yield data look good but hides all process risks.

6. One Sentence Summary

Rework is not just fixing a product; it is running a nonconforming process again — only with work instructions, re-inspection, identification and traceability, and data loop closure is it truly safe.


Rework restores conformity, repair is a customer-approved deviation, and misuse leads to loss of control.

Knowledge code: 2.1.2

Version: v20260925

Author: QTank QTank is dedicated to providing systematic professional knowledge, methodologies, and practical tools for quality management practitioners, helping companies continuously improve their quality capabilities.