Root Cause Identified, but the Solution Nearly Failed? —— A Case Study of a Six Sigma Improvement Project in an Automotive Parts Company

In the five stages of DMAIC, Define sets the direction, Measure provides the evidence, Analyze determines the understanding, Improve decides the success or failure, and Control ensures sustainability. Many Six Sigma projects do not fail due to a lack of data or root cause analysis, but rather in the "risky leap" from root cause to solution: After A stage identifies the critical factors, the I stage often relies on experience to determine the solution, without assessment, risk evaluation, or pilot testing, leading to a direct full-scale implementation. The result is often a failed solution and a terminated project. This article analyzes a real case: a bearing bore concentricity deviation project in an automotive parts company. The A stage took three weeks to lock down the root causes, but the I stage nearly failed due to a "senior engineer's optimal solution," ultimately relying on a solution evaluation matrix and a pilot batch verification to overcome the challenge.

1. Case: 4.2% Concentricity Deviation, Project Reaches I Stage

An automotive parts company supplies transmission housings to a vehicle manufacturer. Recently, customer complaints have surged due to bearing bore concentricity deviations causing assembly noise. The monthly nonconforming rate has climbed from 0.8% to 4.2%, and the customer has issued a notice for immediate rectification. The company established a Six Sigma Black Belt project with the goal of reducing the nonconforming rate to below 1% within three months.

The initial stages of the project proceeded smoothly. In the D stage, the problem was defined as "concentricity deviation in the precision boring process of bearing bores," with the scope limited to one production line and two precision boring machines. In the M stage, the measurement system analysis was completed, with a GR&R result of 8.6%, indicating a qualified measurement system. The process capability was also assessed, with a Cpk of only 0.68. In the A stage, the team used a fishbone diagram and hypothesis testing to identify three critical factors in three weeks: 1) wear on the precision boring fixture's positioning surface, leading to a drift in positioning accuracy; 2) deviation of cutting parameters from the work instruction, with operators privately increasing the speed and feed rate to boost production; 3) significant hardness variation in the raw material batches, with noticeable differences within the same batch.

With clear root causes and solid evidence, the project team was highly motivated and believed that "the hardest part was over." Little did they know, the real test was just beginning.

2. Initial Solution: The "Optimal Solution" by a Senior Engineer Nearly Doomed the Project

At the first solution discussion meeting in the I stage, a senior engineer with twenty years of experience spoke up first: "The root cause is the fixture wear; everything else is secondary. We should directly replace it with a high-precision imported fixture, solving the problem once and for all."

This solution sounded flawless: it was targeted, thorough, and immediate. The project team was almost ready to approve it. However, the planner in charge of delivery did a quick calculation, and the atmosphere turned cold: the procurement cycle for an imported fixture is six weeks, with a unit price of 380,000 yuan, and the customer's rectification deadline is only three months. The production line cannot be shut down during the fixture replacement, which would require overtime and outsourcing to meet production demands. The delivery risk alone could result in the project "winning on quality but losing on orders."

More critically, this solution only addressed one of the three root causes. Even with a new fixture, the issue of operators privately adjusting parameters and the hardness variation in raw materials would remain, leading to continued concentricity deviations. The team was divided: some suggested "replacing the fixture first," others argued for "managing the parameters," and still others claimed, "the raw material is the supplier's responsibility, and we can't control it."

The crux of the debate was that everyone had a "solution that they thought would work," but there was no objective standard to compare them. This is the most common pitfall in the I stage—having too many solutions and ultimately deciding based on seniority and volume of argument.

3. Solution Evaluation Matrix: Turning "I Think It Will Work" into "Data Speaks"

Under the guidance of the Black Belt, the project team laid out the candidate solutions on an evaluation matrix. They identified four candidate solutions: Solution A, replacing the high-precision imported fixture; Solution B, restoring and solidifying the cutting parameters specified in the work instruction, and installing parameter monitoring and alarms; Solution C, increasing hardness sorting for raw materials and returning batches with deviations to the supplier; Solution D, a combination of fixture regrinding and parameter control.

The evaluation dimensions were set to five: expected effectiveness, implementation cost, implementation cycle, execution risk, and maintainability, weighted according to the project goals—effectiveness 30%, cycle 25%, cost 20%, risk 15%, and maintainability 10%.

The scoring results were surprising: the senior engineer's favored Solution A ranked second, losing to Solution D. The reason was straightforward: Solution A, while effective, had a long implementation cycle and high cost, and only addressed one root cause. Solution D, using regrinding instead of replacement (costing less than 30,000 yuan and a two-week cycle), simultaneously locked down parameter control, addressing two of the three root causes and also plugging the management loophole of operators privately adjusting parameters. Solution C, as a supplementary measure for raw material variation, was integrated into Solution D for implementation.

Initially, the senior engineer was unconvinced, but upon seeing the scoring basis for each column in the matrix, he nodded in agreement: "It's not that my solution won't work, but it's too slow and too expensive. Let's start with Solution D, and consider the fixture replacement as a second-phase option once the batch is stable." A decision that could have been made based on seniority was thus brought back to a rational path by the matrix.

4. Risk Prediction: FMEA Helps the Team Avoid the Second Pitfall

With the solution determined, the project team was ready to proceed. The Black Belt stopped them: "We've chosen the right solution, but we need to ask—will it bring new problems?"

The team used the FMEA approach to thoroughly review Solution D: How much can the positioning accuracy be restored after regrinding the fixture? What if the regrinding is excessive? Will the operators adapt to the parameter monitoring alarms, or will they turn them off due to frequent false alarms? Who will handle the additional workload from raw material sorting?

The most valuable risk identified was the uncertainty in the positioning accuracy of the regrinded fixture. If the regrinding amount is not well controlled, the accuracy could be worse than the worn state. Based on this, the team added a preventive action: after regrinding, the positioning accuracy must be tested and found合格 before the fixture can be put back into service, and this test will be included in the daily inspection routine. The parameter monitoring alarms were decoupled from production targets to prevent operators from turning them off to meet production quotas—this addressed the management loophole at the "human" level.

In hindsight, this step saved the project a second time: if the risk analysis had been skipped and the solution implemented directly, the likelihood of excessive regrinding leading to worse accuracy was high, and the project would have faced the awkward situation of "the fix being worse than the problem."

5. Pilot Batch Verification: The Final Gate Before Full-Scale Implementation

After implementing Solution D, the project team did not immediately switch the entire line. Instead, they followed Six Sigma guidelines and conducted a pilot batch verification: producing 300 pieces on one precision boring machine and continuously tracking the results for three days, with half the production during the day shift and half during the night shift.

The verification results were a relief to everyone: the concentricity deviation rate dropped from 4.2% to 0.3%, and the process capability Cpk improved from 0.68 to 1.67. The data from both shifts was stable, with no significant variations. The team also used a control chart to confirm that the process was statistically under control, ruling out the possibility of "good luck."

Looking back at the initial solution, the differences were stark: if Solution A had been implemented directly, 380,000 yuan would have been spent, and the six-week delivery risk would have been a significant burden. Moreover, the root causes of parameter control and raw material sorting would have remained unaddressed. The pilot batch verification, using 300 pieces and less than a week's time, helped the company avoid hundreds of thousands of yuan in ineffective investment and irreversible delivery losses.

6. Closing the I Stage: Not "Implementation," but "Solidification"

After the pilot batch verification passed, the project team officially switched the entire line, but the work in the I stage was not yet complete. The Black Belt listed a "closing checklist": updating the control plan to include parameter monitoring alarms in the daily inspection routine; revising standard operations to include new parameters and inspection requirements in the work instructions; training and assessing operators in both shifts; and incorporating the regrinding accuracy test into the equipment inspection standards to prevent "fixing and breaking again."

A month later, the project was reviewed: the concentricity deviation nonconforming rate stabilized at around 0.15%, the Cpk remained above 1.5, and customer complaints were zero. More importantly, the management loophole of operators privately adjusting parameters was closed, and similar issues did not recur on other production lines. This is the true closing of the I stage: the improvement measures become organizational memory, not a one-time "campaign."

7. Three Lessons Learned

First, identifying the root cause does not equal having a solution. The A stage identifies "which factors are at play," while the I stage must answer "how to change these factors." There is a critical assessment step between root cause and solution, and skipping it allows experience to take precedence.

Second, the solution evaluation matrix, FMEA risk prediction, and pilot batch verification are the three essential gates in the I stage, none of which can be omitted. The matrix helps decide "which solution to choose," FMEA identifies "new risks," and the pilot batch verification confirms "whether it will work." Only after passing all three gates is it worth scaling up.

Third, the endpoint of the I stage is not "implementation," but "solidification." Switching the solution is just the beginning; the control plan, standard operations, training and assessment, and poka-yoke mechanisms must follow to ensure that the improvement results are truly secured. Many projects fail in the final stages due to a lack of solidification, leading to a rebound in defect rates after three months, and everything returning to the starting point.

Identifying the root cause but failing in the solution is the most common reason for Six Sigma project failure; the solution evaluation matrix, FMEA risk prediction, and pilot batch verification are the most cost-effective insurance in the I stage.

Knowledge code: 6.1.1

Version: v20260823

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 improve their quality capabilities.