Which Nonconforming Items to Address First for Effective Improvement? — A Practical Case Study of a Power Supply Company Using a Pareto Chart

A power supply manufacturing company produces switch power supplies for home appliance brands, with three SMT lines and two assembly lines, producing 300,000 units per month. In the first half of the year, the customer complaint rate increased for three consecutive months, and the largest customer issued an ultimatum: if the return rate does not decrease, new project assignments will be off the table. The quality department piled up the customer complaint forms and production line nonconformity records from the past three months on the conference table, listing over a dozen defect types: poor soldering, capacitor leakage, shell scratches, stripped screws, incorrect labels, short circuits in wires, and damaged packaging.

Quality Manager Zhou decided: soldering is a long-standing issue in electronics factories, so start with poor soldering. Three specialized teams spent three months adjusting the reflow soldering oven curves, and the poor soldering rate indeed decreased by 30%, but the customer return rate barely budged, dropping by only a few tenths of a percentage point. The team was confused: was the direction wrong?

The direction was not wrong, but the mistake was in "focusing on a single point." It wasn't until Quality Engineer Wang reclassified the data by defect type and drew a Pareto chart that everyone realized: while poor soldering was the top issue, it only accounted for 28% of the total. The next three issues—capacitor leakage, stripped screws, and shell scratches—combined to form the real majority. By addressing the most obvious issue, they overlooked the others that collectively accounted for over half of the defects. This article will explain how a Pareto chart can transform the question "which to address first" from a debate in the meeting room to a fact on a chart.

1. Why Relying on Intuition to Set Priorities Often Fails

The human brain's judgment of "which problem is the most severe" is almost always skewed by three factors: the most recent (recency effect), the most noisy (rework and line stoppage), and the one the leadership is most concerned about (customer complaints). After a batch of poor soldering complaints last month, the entire factory would have felt that poor soldering was the top issue. However, if the time frame is extended to three months and the data from customer complaints, internal nonconformities, and post-sales returns are combined, the conclusion often differs.

The principle behind the Pareto chart is the Pareto Principle: a small number of key causes contribute to the majority of the results. In the 19th century, economist Pareto discovered that about 80% of the land in Italy was owned by about 20% of the population. Later, quality management pioneer Juran applied this rule to industrial settings, calling it the "vital few and trivial many." The value of the Pareto Principle is not in calculating exact percentages but in forcing the management team to answer three questions: how many categories of problems are there? What percentage does each category represent? Which categories cumulatively account for 80% of the issues? Once these answers are clear, the decision on "which to address first" is no longer determined by the loudest voice.

2. Case Study: Five Steps to Identify What to Improve

Let's return to the power supply company. Wang followed the standard five steps to create a Pareto chart.

Step 1: Standardize Classification, Merge "Same Mechanism" Defects. This is the most time-consuming and critical step. After pulling out the data from the past three months, Wang's first version had 19 defect codes: capacitor-related issues alone had "leakage," "bulging," and "electrolytic capacitor failure," while soldering-related issues had "poor soldering," "insufficient solder," and "solder bridging." He worked with the process and maintenance teams to review the defect code dictionary: any defects with the same failure mechanism but different manifestations were merged into one category; any defects with different mechanisms but incorrectly classified were reclassified. After merging, the 19 categories were reduced to 12. Without this step, the Pareto chart would be a "diluted version" where the major issues are broken down into several small bars, making none stand out.

Step 2: Standardize Statistical Criteria, Combine Three Sources. Wang reclassified and combined the data from customer complaints, final inspection nonconformities, and post-sales returns using the same set of defect codes, with a time window of three months. The reason for combining three sources is that different sources show different defect structures: final inspection mainly catches soldering issues (due to strict online AOI monitoring), customer complaints have a higher proportion of shell scratches and incorrect labels, and post-sales returns focus on capacitors and wires. Focusing on any single source would result in a "one-sided" Pareto chart.

Step 3: Sort and Accumulate, Draw the Standard Pareto Chart. The 12 defect categories were sorted in descending order of frequency, and the percentages were accumulated. The left vertical axis represents the number of defects, and the right vertical axis represents the cumulative percentage. A reference line was drawn at 80%. When this chart was presented, the meeting room fell silent. Over three months, there were 1,268 defects: 356 poor soldering defects (28.1%), 287 capacitor leakage defects (22.6%), 194 stripped screws (15.3%), 133 shell scratches (10.5%), and 82 incorrect labels (6.5%). The top five categories cumulatively accounted for 83%, crossing the 80% reference line; the remaining seven categories only accounted for 17%, forming a typical long tail.

Step 4: Interpret the Chart to Set Focus, A-Class Defects Are the Main Battlefield. By cutting at the 80% cumulative line, the five categories—poor soldering, capacitor leakage, stripped screws, shell scratches, and incorrect labels—were classified as A-Class, the "vital few." This chart corrected two misjudgments: first, capacitor leakage was severely underestimated—before merging, it was split into three codes, each of which seemed insignificant, but after merging, it jumped to second place with 287 defects, and leakage involves safety (some batches had bulging and smoking at the customer site), making its severity much higher than appearance issues; second, 194 stripped screws indicated that assembly issues were as significant as soldering, and focusing only on soldering was like abandoning half the battlefield.

Wang also drew a "cost Pareto chart": he added up and sorted the rework hours, scrap costs, return shipping fees, and compensation amounts for each defect category. Comparing the two charts provided a more comprehensive conclusion—capacitor leakage, which ranked second by quantity, ranked first by cost because each leakage required the entire unit to be scrapped and air-freighted for replacement; while incorrect labels, which had a significant number of defects, only cost a few dollars per defect and ranked last by cost. The final priority should be determined by combining both the "quantity" and "cost" dimensions, as focusing on a single chart can lead to neglecting other important factors.

Step 5: Take Action and Redraw, Keep the Pareto Chart "Dynamic." Manager Zhou established three specialized teams based on the A-Class defects: the soldering team focused on poor soldering (adjusting reflow soldering curves, redesigning stencil openings, with a trial period of 2-3 weeks), the incoming quality control (IQC) team focused on capacitor leakage (conducting specialized audits of suppliers and tightening incoming inspection), and the assembly team focused on stripped screws (switching to electric tightening and re-verifying torque parameters). After three months, the Pareto chart was redrawn: the total number of defects decreased from 1,268 to 512, and the three major categories—poor soldering, capacitor leakage, and stripped screws—decreased by 68%. More importantly, the shape of the chart changed—wire short circuits and damaged packaging, which were previously buried in the long tail, rose to the top and became the new A-Class. The team did not "rest on their laurels" but directly incorporated the new Pareto chart into the next quarter's improvement plan.

3. Results: Return Rate Decreased from 2.1% to 0.6%

Six months later, the company's customer return rate decreased from 2.1% to 0.6%, and the average number of monthly customer complaints dropped from 31 to 9. Just the savings from return shipping fees, rework hours, and customer compensation amounted to approximately 2.8 million yuan over six months. The largest customer reassigned new projects to them. The B and C-Class defects in the long tail were not neglected: appearance issues like scratches and incorrect labels were addressed with low-cost measures such as poka-yoke racks and locking label printer parameters, preventing the "whack-a-mole" effect. Manager Zhou summed up in the review meeting: "In the past, we were 'treating the symptom,' addressing whichever issue was the loudest; now we draw the chart first, then take action, ensuring our efforts are focused where they matter most."

4. Three Practical Lessons, More Important Than Charting Techniques

Lesson 1: Standardizing Classification is the Lifeline of the Pareto Chart. Splitting the same mechanism into multiple codes dilutes the major issues, while mixing different mechanisms into one code creates false major issues. Spending half a day to clarify the defect dictionary before drawing the chart is more valuable than any charting technique.

Lesson 2: A Single Data Source Results in a "One-Sided Chart." Focusing only on internal inspection data misses customer complaints about appearance issues; focusing only on customer complaints misses high-frequency minor defects on the production line. Combining data from customer complaints, internal inspections, and post-sales returns provides a complete view of the problem structure.

Lesson 3: The Pareto Chart is "Dynamic," Not a One-Time Exercise. After improvements show results, the chart must be redrawn: the original A-Class issues will shrink, and the B and C-Class issues in the long tail will rise to become the new A-Class. Updating the Pareto chart monthly ensures that the focus of improvements follows the data, not intuition.

No matter how many nonconforming items there are, start by drawing a Pareto chart: standardize classification, combine three sources, and lock in A-Class by the cumulative 80%, letting the data decide which to address first.

Knowledge code: 5.2.4

Version: v20260829

Author: Quality Think Tank

Quality Think Tank is dedicated to providing systematic knowledge, methodologies, and practical tools for quality management professionals, helping companies continuously improve their quality capabilities.