Why Are Root Causes Still Lying on the Production Line Despite a Blackboard Full of 8D Actions? —— A Five-Step Method for D5 Action Derivation and Selection

1. D5 Brainstorming Sessions Mark the Beginning of Recurrence

At a power supply company, the charger casing clips began to break in batches after low-temperature drops at the client's site, leading to an 8D complaint. The team worked through to D4 and confirmed three root causes with data: 1) mold cavity fillet wear, with the fillet radius already below the design lower limit; 2) the holding pressure time for injection molding was shortened by about 20% by the operator to meet production targets; 3) the first and last piece inspections relied solely on visual checks, which failed to measure the fillet wear. When it came to D5, where corrective actions were to be determined, the meeting took a different turn: the blackboard filled up with actions—full inspection, stricter outgoing inspection, material change, mold repair, operator training, fines, and new gauges. Each person had their own ideas, and no one agreed with the others. In the end, because it was cheap and quick, "full inspection plus training" was implemented. Two months later, the client returned another batch of goods: root causes 1 and 2 were still lying on the production line. During the post-mortem, the quality manager frankly stated: "Our D5 is not about making choices; it's a brainstorming session—there's no direct connection between the actions and the root causes."

2. Step One: List the "Root Cause Elimination Checklist" to Prevent Actions from Sprouting Out of Nowhere

The input for D5 is not a sudden flash of inspiration during the meeting but the root causes confirmed in D4. First, list all the root causes, each on a separate line, with supporting evidence and nature noted: direct root cause—actions or states that directly cause defects; amplifying root cause—gaps that allow defects to pass through and reach the client. Root causes without evidence should be sent back to D4 for further investigation. This checklist serves as the baseline for every subsequent step: actions must be derived for each root cause, no skipping lines, and no impromptu additions. The company in the example went off track at this step: while they mentioned the three root causes, their focus was on "how to get this batch of goods out," leading to actions aimed at release rather than addressing the root causes.

3. Step Two: Ask Three Questions for Each Root Cause to Derive "Root Cause Elimination Actions"

For each root cause listed, ask three questions in sequence: Can it be eliminated—by changing the structure or parameters, making the root cause non-existent; if not, can it be poka-yoked—by locking parameter permissions or preventing operation until the correct state is achieved, ensuring errors cannot occur; if still not possible, can it be automatically intercepted before it flows out—using specialized inspection tools or SPC alarms. Each root cause should have at least one action directly targeting it. If the only actions that can be written are "strengthen inspection," "strengthen training," or "pay attention to operations," it indicates that the root cause has not been thoroughly understood, and the team should return to D4 for further investigation. The example company, following this method, derived three actions: fillet wear—repair the mold to restore the fillet and include fillet measurement in the mold acceptance criteria; shortened holding pressure time—implement parameter permission locking and automatic alarms for out-of-range values; visual inspection failure—use gauges to measure the fillet of the first and last pieces and record the data. This was the first time the actions were directly aligned with the root causes.

4. Step Three: Use a Root Cause-Action Matrix to Identify Unaddressed Root Causes

Arrange the actions in columns and the root causes in rows, and check each cell. Two issues will immediately become apparent: a completely empty row indicates that the root cause is unaddressed, with no actions; a row with only inspection actions indicates that the focus is on intercepting defects at the end, while the root cause continues to produce defects at the source. Inspection is like a spotlight; it can reveal problems but cannot extinguish the fire. Looking back at the initial blackboard of actions: full inspection and stricter outgoing inspection only addressed the "last line of defense," without changing any root causes, making them temporary containment actions rather than corrective actions; operator training did not correspond to specific root causes, making it ineffective. The matrix clearly shows who is truly addressing the root causes and who is just padding the list, thus ending the debate.

5. Step Four: Rank Candidate Actions Using Three Criteria, Select the Minimum Effective Set

There are often multiple sets of candidate actions. Selection should not be based on gut feelings but on three criteria: effectiveness—how deeply the root cause is eliminated and how much the recurrence probability is reduced; cost—investment, cycle time, and ongoing maintenance burden; risk—whether addressing one issue might introduce new problems. Score each criterion on a five-point scale, multiply by weights, and rank the actions. The selected set must meet the hard constraints: each root cause must have at least one action that eliminates it, prioritizing actions that target the root cause, are cost-effective, and have minimal side effects. The example company ultimately did not choose to change the material—although effective, it was time-consuming and required new material validation. Instead, they selected the set of actions: mold repair with fillet acceptance, parameter locking with alarms, and first and last piece gauge measurement. These three actions covered all three root causes, were cost-effective, and were implemented within two weeks.

6. Step Five: Validate Actions on a Pilot Line, Only Pass if Effective

D5 actions are not complete once they are written into the report. Each action must first answer "to what extent is it considered effective" and be accompanied by quantifiable metrics and timelines: the fillet radius CPK after mold repair should not be less than 1.33, the holding pressure time compliance rate should be 100%, and there should be zero complaints about clip breakage for three consecutive months. Then, pilot the actions in a small scope—one machine, one shift—to confirm their effectiveness and lack of side effects before rolling them out fully. If the validation fails, return to step four and reselect the actions. The example company followed this process: the fillet CPK increased from 0.7 to 1.4 after mold repair, the holding pressure compliance rate was 100% for 60 consecutive days after parameter locking, and there were zero complaints for three months. Only then was D5 truly closed, moving on to D6 for effectiveness verification and D7 for standardization. Remember the bottom line: closing an 8D without validated actions is equivalent to writing the next complaint into the order.

7. One Sentence Summary

The key to D5 is not in the meeting but in the preparation before the meeting—listing root causes, deriving elimination actions, checking for gaps, ranking actions, and validating them on a pilot line. Completing these five steps ensures that actions are truly addressing the root causes and not just listed in the report.

D5 is not about who has the most actions but about who can align actions with root causes—list, ask, check, rank, and validate. Only after completing these five steps should the 8D be closed.

Knowledge code: 5.2.1

Version: v20260906

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