One Line Recovered, Other Lines Still Paying Tuition? — Rebuilding Six Sigma Horizontal Deployment at an Auto Parts Company

1. Oil Seal Pressing Damage Cured, but Recurred in Another Workshop Six Months Later

A star assembly line at an automotive parts company had a defect rate for "oil seal pressing damage" as high as 3.2% three years ago, resulting in significant monthly losses from scrap and rework. A Black Belt team went through the entire DMAIC process: in the M phase, they used measurement system analysis to confirm the reliability of the gauges; in the A phase, they used a fishbone diagram and hypothesis testing to identify the root cause as the interaction between the wear of the pressing guide sleeve and the hardness fluctuations of incoming oil seals; in the I phase, they used DOE to find the optimal parameter combination and installed a poka-yoke device. At the project conclusion, the defect rate dropped to 0.2%, and the financial analysis showed an annual benefit of over 2 million yuan, earning the project the title of "Annual Outstanding Project."

However, six months later, the quality director received a complaint from a customer: the same defect in oil seal pressing damage appeared on the same model assembly line in another plant, and an entire batch was rejected during the customer's incoming quality control. Upon reviewing the records, it was found that the defect rate trend on that line was identical to the one before the improvement—none of the project's results had been transferred to the new line. The project team was disbanded after the project's conclusion, and the poka-yoke design drawings were left in the report appendices, with the optimal parameters only existing in the control plan for that specific line. The project was initially selected for the "most painful problem" line, and no one asked, "Which other lines might make the same mistake?" The company paid the tuition twice for the same issue: once for the scrap and rework before the improvement, and again for customer complaints and claims after the improvement.

During the post-mortem, the company calculated that among the 40 Six Sigma projects concluded in the past two years, fewer than 20% had been implemented on more than one production line. About 30% of customer complaints and internal batch defects were related to the same failure mechanisms as the concluded projects. In the project closure reports, the "horizontal expansion recommendations" section was mostly blank, with a few entries stating "recommended for promotion" but lacking responsible persons, timelines, and verification. The problem was clear—it wasn't that the company couldn't do DMAIC, but that once the DMAIC was completed, it was considered done.

2. Results Can't Be Transferred Due to Four Gaps

First, the project closure criteria only apply to the current line. The Y in the project charter is "the defect rate of this line," and the closure review focuses on whether the line's metrics meet the standards. Replicability is never on the list of deliverables. No one requires an answer to what the project leaves behind for the organization.

Second, experience is hidden in individuals. Why key parameters were set, how the poka-yoke logic was developed, and what tricks were used during debugging are mostly tacit knowledge. The reports don't capture everything, and when the people involved move on, the experience is lost.

Third, no one is responsible for replication, and the receiving party lacks motivation. No one is accountable for "transferring the results to other lines." The receiving workshop thinks, "It's not my project," and if it succeeds, it's someone else's achievement, but if it fails, it's their responsibility. So, they simply ignore it.

Fourth, replication is treated as a copy. A few projects have attempted to promote their results, but by directly copying parameters without analyzing the differences between production lines, they failed. One failure became an excuse for more people to claim that "horizontal expansion is unreliable."

3. Incorporate "Replication" into the C Phase of DMAIC

Led by the vice president and the Six Sigma promotion office, the company rebuilt the horizontal expansion mechanism with five actions, with the core idea being: replication is not an "extra task" after project closure, but a formal deliverable in the C phase. Furthermore, the seeds for replication should be planted in the D phase—adding a "list of similar lines/products" to the project charter at the start ensures that the project team knows from day one: this line is just a representative, not the endpoint.

1. Deliver a "Replication Package" in the C Phase. Before project closure, the team must output: a failure mechanism diagram, key factors and optimal parameters, differences in the control plan, poka-yoke device drawings and acceptance criteria, differences in standard work instructions, and training materials, along with a "list of replicable conditions" detailing which conditions are common, which are specific to the current line, and which must be re-verified when transferred to another line.

2. Add "Three Questions for Replication" to the Closure Review. First question: Which other lines, products, and plants have the same failure mechanism? Second question: What are the differences between those lines and the current line? Third question: Who will complete the replication and verification, and by when? If there are no clear answers to these three questions, the project is not signed off, and the benefits are not settled.

3. Lightweight Management of Replication Projects. Horizontal expansion does not require a full DMAIC cycle. The receiving workshop supervisor leads the team, with "seed" members from the original project team providing support. The process follows four steps: difference analysis, localization, small batch verification, and standardization, with a cycle goal of 6 to 8 weeks. The defect rate and Cpk of the replicated line are tracked for 90 days after replication.

4. Link Benefits and Performance. The additional benefits from replication are shared between the original project team and the receiving workshop. This gives the original team the motivation to "help get the project off the ground and see it through," and the receiving workshop includes "actively introducing mature improvements" in the workshop supervisor's annual improvement indicators, addressing the mindset of "it's someone else's project."

5. Build a Case Library Based on Failure Mechanisms. Project closure reports and replication reports are uniformly archived based on failure mechanisms. Before ramping up new lines or introducing new products, the case library is searched. If a similar mechanism is found, horizontal expansion is prioritized before a new project is initiated, avoiding the repetition of old problems on new lines.

In addition to these actions, there is a routine safeguard: the promotion office's monthly meeting lists the progress of each replication project and checks each item. Red-light projects are directly managed by the vice president. This ensures that horizontal expansion has a dedicated "watchdog" and is no longer dependent on individual initiative.

4. Eighteen Months Later: Replication Rate from 20% to 70%

After eighteen months of the new mechanism, the company's replication rate for concluded projects increased from less than 20% to over 70%. The oil seal pressing damage project was replicated to five similar assembly lines in different plants, with cumulative annual benefits growing from over 2 million yuan to over 6 million yuan. Complaints related to similar failure mechanisms decreased by about 60%, and the average cycle for a single replication project stabilized at around 9 weeks.

There were also two instances where the company paid the "tuition" again: one was due to different generations of equipment, where the improvement effect was discounted after directly copying parameters; the other was failing to assess incoming material differences, leading to a much lower improvement rate on the replicated line. Both instances confirmed the same thing—replication is not copying but "localizing with the mechanism in mind," and difference analysis is a crucial step that cannot be skipped.

Summarizing this approach into a methodology, it can be encapsulated in five sentences: Select Accurately—register a list of similar lines at the project initiation, and only replicate projects with the same mechanism and significant benefits; Package Completely—deliver a replication package with a list of conditions in the C phase, ensuring the receiving workshop can implement it without relying on "asking people"; Evaluate Strictly—do not approve the project closure if the three replication questions are not answered; Follow Closely—include replication projects in the monthly meetings, with red-light projects managed by the vice president; Reward Clearly—share the benefits of replication between the original project team and the receiving workshop, giving both parties a clear incentive. These five sentences sound simple, but the challenge lies in turning them from slogans into hard requirements in the project closure process. The cost to the company was minimal: the promotion office appointed a dedicated coordinator, and each replication project added one to two weeks of seed labor from the original project team. The result is that similar problems no longer reoccur in different workshops. This is a cost-effective investment.

5. One Sentence Summary

The unit of benefit for Six Sigma is not the project but the organization—projects only save one line, while a horizontal expansion mechanism saves the entire company. Project closure is not the end; ensuring that the results take root in more production lines is the true last mile of DMAIC.

Projects save one line, mechanisms save an organization—incorporate horizontal expansion into the C phase to ensure each improvement lives more than once.

Knowledge code: 6.1.1

Version: v20260902

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.