6σ- Key to Quality Disputes

Effective Communication Empowered by 6σ- Key to Quality Disputes

Lai Chih-Chung
Plant 1

6σ provides a mode of logical thinking and statistical analysis tool that solves the problems lay down a firm foundation where fair and effective communicate can take place. 

 

In July 2011, an American customer rejected one batch after another of samples for a certain product that had been shipped continuously without any issues. This was a nasty surprise! The customer insisted that there were problems despite the best effort of our quality assurance department to proving the quality compliance. What went wrong? We had to clarify the situation and find out the reason why.

All efforts put in to avoid supply interruptions

If we could not resume deliveries soon, the customer would suffer from shortage on the production lines. We immediately applied quality management tools of Six Sigma to investigate the possible causes. We also performed "hypothesis testing" on the test data of both parties (see knowledge box). Historical data were collected from the products we shipped in the past and tested by statistical analysis. Several discussion meetings were held in the meantime in hope of finding the root causes as soon as possible.

Our statistical analysis results showed the dispute may caused by slight differences in operating conditions, such as dyeing condition, type of fabric, equipment, or the operating method of the personnel, between both parties. However, no one could be sure before making item-by-item comparison and clarification. We sent our ideas across to the customer with a complete study report. After the customer learned that Everlight went this far to conduct such a sponstaneous study on all possible causes, the customer agreed to select the standard again and continue with the shipping of goods. This temporarily resolved the crisis of potential stock-out at the customer's production line.

Siren went off again while trying hard to close the gap

While the incident ended temporarily, and we shipped out the products stably again, we kept in our mind to minimize the difference between the dyeing process conditions. Through various channels, we continued to collected information on the operating condition at our customer's location and investigated the differences of the inspection system between us.

In August 2012, another bomb went off – the customer rejected several batches of products in a row! As the inventory at our customer's factory was about to be depleted, we were extremely anxious. Fortunately, the sales department brought back the standard from our customer. Together with the results from our statistical hypothesis testing, we found out that the standard of our customer had degraded and lead to misinterpretation. We also provided the statistical analysis results and explanations to the customer, and keep close and seamless communication with the customer by e-mails.

Statistical analysis opens up the door of communications

In early 2013, the customer agreed to a face-to-face discussion on this recurring incident. Although at first both sides were only stating their opinion, no consensus was achieved. Everlight presented the statistical data as the foundation of communication with great sincerity and determination in solving the problem, the customer responded with a gesture of goodwill. In the end the customer selected another standard, and adopted the dyeing process and condition of the customer's as the common inspection standard to minimize the testing differences.

We learned from the incident that Six Sigma provides a mode of logical thinking and statistical analysis tool that solves the problems lay down a firm foundation where fair and effective communicate can take place. This approach not only helps resolving the crisis wherein the customer runs out of raw material twice but also realizes Everlight's brand value of "reliability and client-centric". This incident makes the customer more assured and confident with Everlight than ever.

Hypothesis testing

A method or technique with limited data (maybe from sampling) to infer the overall situation of a study. It is mainly used to infer data and target value (or imaginary value) or between two sets of data whether there are significant differences.