From "Audit Props" to "Improvement Engine" — A Practical Case Study of Implementing ISO 9001 Improvement Clauses

1. Why Do Improvement Clauses Often Remain in Documents?

ISO 9001:2015 dedicates Chapter 10 to "Improvement," which includes three core requirements: 10.1 General, which requires companies to identify and select improvement opportunities and take corresponding actions; 10.2 Nonconformity and Corrective Action, which mandates corrective actions for nonconforming products to prevent recurrence; and 10.3 Continuous Improvement, which requires the continuous improvement of the quality management system's (QMS) suitability, adequacy, and effectiveness. The clauses themselves are not complex, but in reality, many companies' improvement clauses exist only on paper: procedure documents mention "continuous improvement," management reviews annually emphasize "continuous improvement," but on the shop floor, improvements still rely on the quality manager "putting out fires," the boss making decisions on a whim, or customer complaints forcing action.

Why does this "disconnection" occur? The fundamental reason is not that the standards are difficult to understand, but that companies treat improvement as an "audit requirement" rather than a "business necessity." Before audits, companies rush to update records, fill corrective actions with "strengthen training," and after improvement projects are initiated, no one follows up. Behind these phenomena lies a lack of a closed-loop mechanism, data support, and resource assurance. To implement improvement clauses, what is needed is not just slogans but a system that runs from "identifying problems" to "preventing recurrence."

Below, we will use a complete case study from a manufacturing company to analyze the entire process of transforming improvement clauses from "props" to "engines."

2. Case Background: The Improvement Dilemma of a Parts Manufacturer

A certain automotive parts manufacturing company produces chassis connectors for vehicle manufacturers, with an annual output value of about 600 million yuan and over 1,200 employees. The company obtained ISO 9001 certification eight years ago and later received the IATF 16949 certificate. The system documentation is complete, and all external audits have been successfully passed. However, beneath the surface, problems have been accumulating: the client PPM (parts per million) has hovered around 800 for three consecutive years, far exceeding the industry benchmark of 200; the internal scrap rate remains high, with the electroplating process alone incurring losses of over 400,000 yuan per month; even more troubling is the recurrence of similar customer complaints—issues with a certain bracket breaking were reported four times in a year, each time resulting in a "correction, apology, and a batch of replacement goods," only to be forgotten afterward.

At the year-end management review, the quality director laid out a stack of data: 186 corrective action requests were issued throughout the year, with 143 of them citing "improper operation" or "insufficient training" as the cause, and the measures listed as "strengthen training" or "increase inspections." Less than 10% of these actions were verified for effectiveness or implemented across the board. The general manager then asked, "We have passed so many audits, why do the same problems keep recurring?" This question marked the beginning of the company's improvement transformation.

3. First Step: Transform Corrective Actions from "Form Filling" to a Closed Loop

The company's first move was to tackle the most concentrated issue of "bracket breaking" according to the requirements of ISO 9001 Clause 10.2, and to walk through a complete corrective action process.

In the past, their corrective actions followed a "three-step" approach: customer complaint, write an 8D report, and send it to the customer. The "root cause" section of the report was often filled with "improper operation by employees" based on experience. This time, the quality team approached the problem with a "problem-solving" mindset and did four things: First, they clearly defined the problem—not just "bracket breaking," but "a certain batch of products showing cracks in the heat-affected zone during whole-machine durability testing, with a defect rate of 3.2%," and they retained samples of the failed parts; Second, they used a fishbone diagram to investigate from five dimensions (people, machines, materials, methods, and environment), and combined this with on-site data to verify the root cause, which was ultimately identified as welding parameter drift—after two hours of continuous production, the current setting of the welding machine would gradually decrease by about 8%, exceeding the process window; Third, they implemented technical measures, installing real-time current monitoring and automatic alarm systems, and revising the work instructions; Fourth, and most crucially, they verified and expanded the measures: after implementation, they tracked four batches, only closing the corrective action request when the defect rate dropped below 0.1%, and they also inspected all similar welding machines, finding that two others had the same parameter drift issue, which they addressed simultaneously.

This single corrective action took six weeks, a stark contrast to the previous "two-day report writing." However, it brought about fundamental changes: the company experienced the benefits of "finding the true cause and getting it right the first time." Subsequently, the company formalized the requirements of Clause 10.2 into a "five-step nonconformity disposal method"—containment, cause analysis, correction, verification, and horizontal deployment—and clearly defined the responsible person, timeline, and deliverables for each step. Corrective action requests were no longer the sole responsibility of the quality department; the process, equipment, and production departments also had to sign off on them.

4. Second Step: Use Data to Guide Improvement

After establishing the closed loop, the company found the next bottleneck: "What to improve?" In the past, improvement topics were proposed by the quality manager based on intuition, focusing on appearance one day and dimensions the next, without clear priorities. Therefore, the company, following the requirements of Clause 10.1 "Determine and Select Improvement Opportunities," established a data-driven mechanism for selecting improvement topics.

They collected data from three sources: First, customer complaint and return data, analyzed using a Pareto chart by defect category; Second, internal scrap data, statistically analyzed by process and defect type on a monthly basis; Third, process inspection data, incorporating the SPC monitoring results of key dimensions into the monthly review. The first Pareto chart in the month surprised everyone—while everyone assumed the most common issue was electroplating appearance defects, the data showed that "thread processing dimension out of tolerance" accounted for 37% of the defects, and this issue only made up 12% of customer complaints, indicating a significant internal loss that was largely overlooked.

With data in hand, selecting improvement topics became uncontroversial. The company determined the quarterly improvement priorities: first, tackle thread dimension out of tolerance, then address electroplating defects, and simultaneously monitor the rising issue of assembly interference in customer complaints. The quality department also established an "improvement opportunity list," incorporating findings from customer audits, internal audits, processes with insufficient process capability, and employee suggestions. This list is reviewed quarterly, and projects are scored and ranked using "occurrence frequency × impact × improvement difficulty" to ensure that improvement resources are always directed to the most worthwhile areas.

5. Third Step: Incorporate Improvement into Routine Management with Resources and Evaluation

The topic selection mechanism solved the question of "what to improve," but the next challenge was "who will improve, how long will it take, and what to do if it stalls." In the past, the company's improvement projects were mostly advanced part-time by quality engineers, who handled customer complaints during the day and wrote improvement reports at night, often dragging projects for six months. This time, the company, in line with the requirements of Clause 10.3 "Continuous Improvement," formally integrated improvement into the management system.

The specific practices included three aspects: First, the project initiation system: each quarter, management approves 3-5 key projects from the improvement opportunity list, clearly defining the project leader, cross-departmental members, goals (quantified where possible, such as "reduce the thread dimension out of tolerance scrap rate by 50% within three months"), completion time, and budget, which takes effect after the general manager's signature; Second, resource assurance and regular meetings: key projects hold a 30-minute progress meeting each week, and any resource issues such as equipment modifications or funding are resolved on the spot by the vice president in charge, ensuring projects are not delayed due to lack of approval; Third, performance evaluation: the completion status of improvement projects is included in the monthly performance evaluation of relevant departments and leaders, and departments with a completion rate below 80% for two consecutive quarters must explain the reasons during the management review.

In the first quarter after the mechanism was established, significant changes were observed: out of the 5 key projects initiated, 4 were completed on schedule, and 1 was delayed by a month due to equipment delivery. The phenomenon of "initial enthusiasm followed by neglect" largely disappeared. More importantly, middle-level managers began to proactively propose improvement topics, as they realized that successful improvement projects could positively impact departmental performance and personal promotions.

6. Fourth Step: Standardize and Poka-Yoke to Lock in Results

Whether improvement results can be sustained is a common pitfall for many companies. This company also experienced "recurring issues after initial resolution": a dimension problem solved at the beginning of the year resurfaced at the end when a new batch of operators was hired. To address this, they made "preventing recurrence" a hard requirement for closing improvement projects, aligning with the requirements of Clause 10.2 "Prevent Recurrence."

The specific practices included three aspects: First, standardization: after each improvement measure is completed, the relevant work instructions, control plans, or inspection standards must be updated, and the affected positions must be trained during pre-shift meetings, with training records as essential documentation for project acceptance; Second, poka-yoke: any measures that rely on "increased inspection frequency" or "enhanced responsibility" must be redesigned, prioritizing physical error-proofing methods such as jigs, limiters, and alarms—after installing automatic detection and shutdown devices for the thread process, operators could not release out-of-tolerance parts even if they were negligent; Third, change management: any process, equipment, or jig changes related to improvements must be managed through a change management process, assessing risks before changes and verifying effectiveness afterward to prevent "fixing A and breaking B."

The effects of this approach were evident in the data six months later: repeated customer complaints for similar issues dropped to zero; more importantly, when auditors issued nonconformities, the company no longer scrambled to "update records" but could present a complete "analysis—action—verification—standardization" evidence chain.

7. Outcome Review: Data Speaks After One Year

From the start of the improvement transformation to the twelfth month, the company's key indicators showed significant changes: the client PPM dropped from around 800 to 220, approaching the industry benchmark; the internal scrap rate decreased by 61%, reducing losses by about 550,000 yuan per month in the electroplating and thread processes alone; the average closure cycle for corrective actions shortened from 46 days to 12 days; and the horizontal deployment rate increased from less than 10% to over 90%. Over the year, the direct financial benefits from improvements exceeded 6 million yuan, while the total improvement investment was less than 800,000 yuan.

Reviewing this transformation, several lessons can be learned: First, the starting point for improvement is not tools but mechanisms—the company did not rush to implement Six Sigma or black belt training but first established the basic closed loop of "nonconformity → corrective action → verification → horizontal deployment," gradually introducing tools as the loop became operational; Second, leadership must focus on resource allocation—the general manager personally approved project initiations and attended monthly progress meetings, which was more effective than any motivational speech; Third, data must become the "common language" for improvement—when everyone stopped debating "where I think the problem is" and instead focused on "which item to improve first" based on the Pareto chart, the improvement cycle entered a positive loop; Fourth, preventing recurrence is more important than solving the problem—all results must be locked in through standardization and poka-yoke, otherwise, improvement becomes a game of "whack-a-mole."

Of course, the transformation was not without challenges: initially, the quality department tried to take on all improvements, leading other departments to view improvements as "helping the quality department," which created significant resistance. This was resolved by changing to a "problem-owning department leads, quality department supports" model. Additionally, setting overly ambitious initial targets led to a project being revised mid-way in the first month, which was later adjusted to more realistic, phased goals, accelerating the improvement process.

8. Conclusion

The improvement clauses of ISO 9001 are not meant for auditors but are a set of underlying principles for the company's self-evolution. The three requirements in Chapter 10 essentially address three questions: the willingness to improve (10.1), the ability to stop bleeding and prevent recurrence when problems arise (10.2), and the continuous enhancement of the QMS (10.3). This case study demonstrates that implementing improvement clauses does not require advanced theories but rather a system that builds a closed loop, leverages data, provides adequate resources, and locks in results. When improvement shifts from "audit props" to a "driving engine," the QMS truly begins to create value for the company.

Improvement is not a performance for audits, but a step up from every nonconformity.

Knowledge code: 2.1.1

Version: v20260803

Author: Quality Think Tank

Quality Think Tank is dedicated to providing systematic professional knowledge, methodologies, and practical tools to quality management practitioners, helping companies continuously enhance their quality capabilities.