Michelin is a famous French company that produces high-quality tires. Once, they encountered a headache. The defective rate of formula racing tires produced is as high as 25%, and the return rate is very high. Although the company's scientists carefully tested the application of six horses in each process, they still couldn't find out the problem. After many unsuccessful efforts, the company decided to try lean horse to find out and solve this problem, so they invited Anan? Martha came to form an improvement team.

When doing improvement activities for Michelin, Anan's implementation steps are to decompose the five processes of the Six Sigma DMAIC method, namely, definition, measurement, analysis, improvement and control, and carry out three-week improvement activities. The improvement activities in the first week are mainly to define the problem, carry out measurement, and get the measurement results and various data; Followed by a week-long improvement activity to analyze and improve the problems and data; Followed by a week-long improvement activities, do some control work. Three different improvement weeks (all lasting for one week) have their own emphasis, but lean and six sigma methods are always used alternately.

In the first week, Anan led the team to collect historical data first, and then observed it to find some anomalies or problems. This week is the definition and measurement stage, mainly using historical data to carry out the work. According to the experience of previous implementation, in general, when staff collect data at the beginning, the collection process may be sloppy because they don't know the importance of the data. Therefore, at this stage, the improvement team does not simply use historical data, but needs the whole team to carefully observe and verify these data. At this time, lean method is used for observation, while six sigma method is used for digital analysis and verification. After observation and verification, we usually find some problems: either the data is inaccurate or insufficient, or it is not the data we need at all, or it is completely irrelevant to the problem. If so, it will be verified by experiments.

The improvement team found that some of these problems can be found with ordinary six sigma tools, while others cannot. For example, in the production of rubber tires, large-scale mixing equipment, including scales, is used to weigh rubber and other components and then put them into the mixing equipment. In the observation activities of the improvement team (of course, there are historical data at this time), it is found that this scale is sometimes used before calibration. Of course, there is no problem if it can be calibrated regularly. This scale can be calibrated based on the measurable quota of 50 kg, but the improvement team found that when weighing light or heavy items, weighing based on 50 kg will cause the deviation of actual weight, resulting in light items being too light and heavy items being too heavy. Of course, this is also a problem for scale manufacturers. This problem was discovered through the observation emphasized in lean production.

Another problem that has not been found before, but discovered through observation, is that when a worker who has worked in this weighing position for 25 years closes the door and presses the button to weigh, a piece of metal is sucked into the vacuum tube. Because the rubber does not cover all the vacuum tubes, the metal sheet is sucked in, which affects the reading of the scale. The workers said that when they drilled holes in the metal plate with a drill, they never knew that such problems would occur. This shows that the vacuum problem caused the wrong reading. This is a completely random event, and no one knows when it will happen. So the improvement team proposed to change the drilling method or use metal mesh to solve this problem. "This case shows that it is impossible to solve the problem without observation activities and teamwork."

Then the improvement team carried out improvement activities in the second week and continued to analyze the collected data. At this time, the six-sigma method was used, from which many other problems were found and solved, and the return rate decreased by 2%.

This example shows that the effect of integrating lean and six horses into lean horses is not only better than that of lean alone, but also much better than that of six horses, which requires engineers to spend a lot of time on data analysis.