10 Years of Quality Management: My 5-Step Improvement Method to Save You Time and Effort
What do quality management professionals fear the most?
The worst fear is that after working hard on an improvement plan, it fails to be implemented. Or even if it is implemented, the results are not significant, and the problem resurfaces a few months later.
I have seen too many quality professionals trapped in the vicious cycle of "identifying issues → issuing reports → driving improvements → issues rebounding → re-identifying issues."
Today, I will share a proven 5-step improvement method, distilled from my practical experience across multiple industry projects. No beating around the bush, just straight to the point.
Step 1: Define the Problem, Don't Rush to Conclusions
Many improvement projects fail at the first step—the problem is misdefined.
For example, the production line reports that "the welding defect rate has increased from 2% to 5%." Most people would immediately think of "changing the welding wire" or "adjusting the parameters."
But what is the problem?
Is it "unstable welding parameters," "insufficient skills of new employees," or "changes in incoming material batches"?
These are three fundamentally different causes, each requiring a completely different solution.
What to do?
Use a "problem description template" to force yourself to define the problem accurately:
What happened? → Welding defect rate increased
Where did it happen? → Line 3
When did it start? → Last Wednesday, day shift
How significant is the impact? → Defect rate increased from 2% to 5%
What is the difference from normal? → Previously stable at around 2%
Key principle: During the problem definition phase, focus on "facts" rather than "causes." Don't ask "why," but rather "what."
Only after completing this step can you proceed to the root cause analysis.
Step 2: Root Cause Analysis, Use Tools Instead of Guesswork
Once the problem is clearly defined, the next step is to find the cause.
A common pitfall is "empiricism"—senior engineers rely on their experience to say "it must be the XXX issue," and the team follows suit.
Experience can provide direction, but it cannot replace verification.
I recommend three time-tested tools:
Fishbone Diagram (Cause and Effect Diagram) Expand along six dimensions: "people, machines, materials, methods, environment, and measurement," to ensure no potential root causes are overlooked. When drawing the diagram, each branch should be questioned to the third level to truly identify the "root cause."
5Why Analysis Continuously ask "why" five times to get to the root of the problem. Note: 5Why does not necessarily mean asking five times; it means asking until you find a "root cause that can be acted upon."
Hypothesis Verification Table For each potential root cause, answer: What data do I need to verify this hypothesis? If the verification fails, what is my alternative hypothesis?
A real case:
A certain electronics factory had a consistently high defect rate for PCBA welding. The team, based on experience, concluded it was a "soldering flux issue" and planned to change suppliers.
After using a fishbone diagram, they discovered that besides the soldering flux, "welding temperature profile" and "PCB incoming material storage conditions" were also potential root causes.
Further verification revealed that the true root cause was that the PCB incoming material was not used within the specified time after being opened.
Without systematic analysis, changing the soldering flux would not have solved the problem and would have unnecessarily increased costs.
Step 3: Develop Countermeasures, Design Poka-yoke Mechanisms
After identifying the root cause, a core principle when developing countermeasures is: prevention is better than detection, and detection is better than training.
What does this mean?
Let's illustrate with three real cases:
Case A: Relying on Training "Improper operation by welders → Strengthen training," Effect: Improved for two weeks, then reverted to the original state.
Case B: Relying on Detection "Possible incorrect assembly → Add a 100% inspection step," Effect: Identified missing parts, but increased rework costs.
Case C: Relying on Prevention "Possible incorrect assembly → Design a physical positioning structure that prevents incorrect assembly," Effect: Impossible to make a mistake.
This is the core concept of poka-yoke.
When developing countermeasures, prioritize them in the following order:
- Elimination — Remove the step that causes the problem
- Substitution — Replace with a more reliable process/material
- Prevention — Design physical/logical mechanisms to prevent errors
- Simplification — Simplify operations to reduce the probability of errors
- Detection — Add inspection steps
- Training — Use as a supplementary measure
In most cases, we tend to start from level 6. Experts start from level 1.
Step 4: Pilot Verification, Use Small Data to Prove Big Value
Once the countermeasures are developed, don't rush to implement them across the board.
One principle: Pilot first, then roll out.
The purpose of piloting is to verify the effectiveness of the solution with the least cost.
Three key points for piloting:
Select Representative Samples — Choose a shift, a production line, or a product model. Avoid selecting the best or worst scenarios; the samples should be representative.
Set Clear Acceptance Criteria — What are the comparison metrics before and after the improvement? What is the target value? When is the acceptance time point?
Record Deviations in Execution — What execution deviations were encountered during the pilot? Recording these deviations is very valuable for subsequent rollouts.
Data Presentation:
I once helped a client optimize their incoming quality control (IQC) process. During the pilot phase, we selected Class A materials (15% of the total incoming materials), and the results were excellent—inspection efficiency improved by 40%, and the miss rate decreased by 60%.
However, during the full rollout, we found that the conditions for Class B and C materials were completely different from those for Class A materials, requiring adaptation and adjustment. If we had rolled out without piloting, it would likely have failed.
Step 5: Standardization and Horizontal Rollout
After the improvement plan has been verified as effective, the most important and often overlooked step is standardization.
What is standardization?
It's not just writing a document and letting it gather dust on a server. It involves embedding the improved methods into formal work processes to ensure they are consistently executed.
Three levels of standardization:
Level 1: Documentation Update work instructions, inspection specifications, and process documents. Record the changes in the version updates.
Level 2: Training All relevant personnel must complete training and pass the assessment. Training records for critical operations should be retained.
Level 3: Systematization Embed control requirements into the system. For example, set parameter lock ranges in the MES system, with automatic alerts for out-of-range values, rather than relying on operators to check voluntarily.
After standardization, horizontally roll out to other similar processes or products.
Note a common mistake: When rolling out horizontally, don't copy blindly; do "adaptation"—different processes have different conditions and may require appropriate adjustments.
Summary: The Complete Map of the 5-Step Method
Step 1 [Define the Problem] → Describe facts, don't look for causes
↓
Step 2 [Root Cause Analysis] → Use tools to verify, don't rely on experience
↓
Step 3 [Develop Countermeasures] → Prioritize prevention, use training as a fallback
↓
Step 4 [Pilot Verification] → Test in a small scope, let data speak
↓
Step 5 [Standardization] → Embed into processes, roll out horizontally
This method is not complicated, but few people implement it fully.
Because each step involves fighting human nature—we are eager to draw conclusions, lazy to verify, want to roll out quickly, and overlook standardization.
However, quality management is a field where "slowness is speed." The more time you spend on the first two steps, the fewer rework issues you will encounter later.
Next time you encounter a quality issue, make a commitment to yourself: define the problem clearly before taking action.
This is more important than any tool.
This article is original content. Contact the author for reprints.
Knowledge code: 2.1.1 Version: v20260501 Author: Quality Think Tank