21 of 28 8D reports blamed the operator: how an automotive welding plant found the real causes

Quality · 3 min read · 19.08.2026

Client

An automotive Tier-2 supplier of welded assemblies for body modules. Roughly 300 employees, production across three shifts, with the main customer being a Tier-1 supplying multiple OEM brands. We present the study anonymously — it is compiled from several of our projects in this segment and the numbers are rounded so that the specific supplier cannot be identified.

The project was initiated by the quality director after the OEM formally announced escalation proceedings. Management did not want more forms. They wanted to know why 8D does not work even when it is formally complied with.

Challenge

Complaints from the welding plant grew for three consecutive quarters: from 12 to 31 PPM. The OEM announced a move to CS2 status. The plant faced an on-site audit, a downgraded rating in the quality portal, and the loss of share in future projects.

Behind the numbers stood a clear pattern. Internal 8D reports were being closed quickly and formally — the causal chain ended at operator error and retraining. 40% of problems returned within 8 weeks. Meanwhile, quality engineers spent half of their capacity sorting parts at the customer and the rest writing reports. No time was left for real analysis.

Approach

1. Audit of closed 8D reports. We reviewed a sample of 28 reports closed over the previous 12 months. In 21 of them, the analysis ended at human error — without asking why the process allowed a defective part to move on. The output was not hundreds of new findings, but three systemic failures of the method that repeated in every report.
2. A new definition of “proven cause”. With the team we agreed on a rule: a cause is proven only when we can deliberately trigger it and suppress it. Since then, retraining without verified effectiveness has not been enough to close a report.
3. Training on real cases. Instead of a theoretical course, we walked three teams through their first three 8Ds, from complaint to verified corrective action — right at the part, using the 5-Why method and testing hypotheses on the spot. Supervisors and quality staff learned to challenge the first explanation voiced in the room.
4. Management rhythm. A weekly 15-minute review of open 8Ds and a monthly check of recurring errors. One rule without exception: a report without proof of effectiveness is not closed, even when the form is filled in.
5. Recording in the PFMEA and control plan. Verified solutions are reflected in the PFMEA and control plan once a month — a new inspection step, a poka-yoke on the fixture, adjusted welding parameters. The solution stays in the system even after its author moves on to another section.

Results

The turnaround took half a year and required no investment in new technology. What changed was the method, the discipline in following it, and who decides on closure.

• Customer complaints from the welding plant fell from 31 PPM to 7 PPM within 6 months.
• The share of recurring errors dropped from 40% to less than 10%.
• The time from complaint to closing an 8D with verified effectiveness was cut from 6 weeks to 2 weeks.

The escalation to CS2 never took place — the OEM withdrew the threat after a quarter of a declining trend and demonstrated closures. A permanent sorting team at the customer never had to be created, so quality engineers returned half of their capacity to preventive work. The weekly rhythm of reviews remains in place to this day, and the teams run 8Ds without external facilitation. In the meantime, the stamping plant has adopted the same approach.

If complaints keep coming back and reports solve nothing, take a look at how we work on quality and problem-solving projects. An audit of a sample of closed 8D reports will show within a few days where the causal chain is actually breaking down.

More on this topic

40 % Repeat Complaints: How an Automotive Tier 2 Supplier Stopped Closing 8Ds on Paper

Reducing Recurring Defects from 40%: How an Automotive Supplier Stopped OEM Complaints

How to Expand the Supplier Database Without an Increase in Claims