Critical to quality metrics
– When you order pizza for delivery, what’s important to you? Well, you probably don’t want to wait too long, and you definitely don’t want cold pizza. Don’t want to wait too long, and don’t want cold pizza, are what’s important to customers. But these are expressed when from the customer’s viewpoint. These are what we call the Voice of the Customer, or VOC. VOC are needs and expectations expressed in the customer’s language. Now, put yourself in the shoes of the pizza restaurant owner. How can you make these words meaningful to your employees as they make and deliver pizzas everyday? You will have to translate them from the customer’s language, into language that your employees can relate to when they perform their work. Put another way, you have to translate the Voice of the Customer, into Critical-to-Quality requirements, or CTQs. What are CTQs? CTQs are the performance characteristics of a process, product, or service, that are critically important to customers. CTQs are measurable, and how good they need to be in order to satisfy the needs and expectations of customers, can be determined and established. Back to our pizza example. From the Voice of the Customer, or VOC, we know that customers don’t want to wait too long, and they don’t want cold pizza. We can translate “don’t want to wait too long” to on-time delivery. And we can translate “don’t want cold pizza” to hot pizza when delivered. So, the Critical-to-Quality requirements, or CTQs, are on-time delivery and hot pizza when delivered. These CTQs can be measured by order-to-delivery time in minutes, and temperature of pizza in degrees Fahrenheit. We can specify how well your restaurant must perform on these metrics, in order to satisfy your customers. In other words, we can determine the specifications and targets for these CTQ metrics. In this example, the CTQ targets or specifications might be delivery time in 30 minutes or less, and pizza temperature doesn’t fall below 90 degrees Fahrenheit. To recap, CTQs are the performance characteristics of a process, product, or service, that are critically important to customers. CTQs are measurable, and we can specify how good they need to be, in order to satisfy the needs and expectations of customers. If you are managing a process, you benefit from learning about CTQs. With CTQs, you know what metrics to monitor, and how well they must perform to satisfy customers. In our example, you want to monitor order-to-delivery time, and temperature of pizza. If you’re doing a Six Sigma project, you will definitely benefit from learning about CTQs. The underlying premise of Six Sigma projects is Y is a function of X. In our example, one Y is pizza temperature, and the other is delivery time. Xs are all those things that affect pizza temperature and delivery time, respectively. Y and CTQs help you focus your measurement, analysis, and improvement efforts. CTQs provide customer focus for your project, and after the project, CTQs provide customer focus for your process on a day in, day out basis.
Variation and the normal curve
– Remember in school when you were told that only a certain percentage of students will get an A, B, or C because the class is graded on a curve? Well, that curve is the normal curve. In this movie, I will discuss variation and the normal curve. Let me illustrate with an example. How long does it take you to travel to work each morning? Maybe an average of 60 minutes. On some days, it’s as short as 45 minutes, while on other days, it may take as long as 75 minutes, and everyday is a little different. It varies. That’s what we call variation. Let’s plot it. Each dot represents each day’s travel time. The dots pile up vertically if travel times are the same. For no reason, it varies, taking you longer on some days, and shorter on other days. The variation is random. As you can see, this natural, random variation forms a bell-shaped curve. This bell-shaped curve is called the normal curve, or normal distribution. Your average, or mean, is in the middle at 60 minutes. Since the majority of the travel time is between 45 and 75 minutes, the bell curve trails off at 45 minutes on the low end, and 75 minutes on the high end. The normal distribution is a symmetrical bell-shaped curve centered at the mean, and the bell’s curve trails off at a distance of three standard deviations from the mean. In our example, these are at 75 and 45 minutes, or plus, minus 15 minutes from the mean. Since 15 minutes is three standard deviations, each standard deviation is five minutes. In any normal curve, the majority of the variation, or 99.73%, lies within three standard deviations of the mean, and approximately 95% lies within two standard deviations, and approximately 68% lies within one standard deviation of the mean. These percentages are universally true for all normal curves. In this example, with mean at 60 minutes and a standard deviation at five minutes, 68% of the time, travel takes between 55 and 65 minutes. 95% of the time, it is between 50 and 70 minutes. 99.7% of the time, it is between 45 and 75 minutes. Using the normal curve, you now know how long it takes you to travel to work most of the time. There we have it. We’ve discussed variation and the normal curve. Why is this important? Because as you learn more about Six Sigma, you will find the normal curve is widely used to depict random variation in processes.
Defects per million opportunities
– Let’s imagine that you are the CEO of a company with two divisions. One makes markers and the other makes notebook PCs. The marker division reported their quality is at two defects per unit. The PC division also reported a quality of two defects per unit. Are the two divisions performing at the same quality level? The answer is no. In this movie, I will discuss a more accurate way for calculating performance using the concept of defect opportunity and a metric called DPMO, or, defects per million opportunities. Let’s continue with that example. How many things can possibly go wrong in a marker? Perhaps five, such as the pen can leak, the cap doesn’t stay on, the ink color is wrong and so on. These possibilities for defects in each marker are called defect opportunities. So let’s assume that there are five defect opportunities per unit for the marker. Now how many things can possibly go wrong in a notebook PC? The notebook PC is a lot more complicated than a marker. There are many more opportunities for defects. Looking at the number of components and functions in a notebook, the number of defect opportunities is definitely more than five, much closer to 1,000. So for the sake of illustration, let’s say there are 1,000 defect opportunities per unit. To summarize, a marker has five opportunities for defects while a notebook PC has 1,000 opportunities. Let’s compare that report of two defects per unit. Say 1,000 units were produced in each division. That means there were a total of 2,000 defects in those 1,000 markers and 2,000 defects in the 1,000 notebook PCs. Defects per opportunity, or DPO, is the total number of defects divided by the total number of opportunities. For the marker, the numerator is the total number of defects which is 2,000 and the denominator is the total number of opportunities, which is five per unit multiplied by 1,000 units, which equals 5,000. When we divide 2,000 by 5,000 we obtain 0.4 defects per opportunity, or 0.4 DPO. To convert to DPMO, we multiply by one million and we obtain 400,000 DPMO. For the notebook PCs, the total number of defects is 2,000 also. The denominator is 1,000 opportunities per unit multiplied by 1,000 units produced for a total of one million opportunities. When we divide 2,000 by one million, we obtain 0.002 defects per opportunity, or 0.002 DPO. To convert to DPMO, we multiply by one million and it is 2,000 DPMO. The performance of the two divisions are very different. Based on our example, 2,000 DPMO is very different from 400,000 DPMO. Using DPMO as a measure of quality is more accurate than using defects per unit. So back to our CEO example. Comparing both divisions by defects per unit is misleading. The PC division is performing at a much higher level of quality than the marker division. We can only arrive at this conclusion by using DPMO and that’s why we use DPMO in Six Sigma.
Learn Sigma levels
– What is Sigma Level? What do people mean when they say, “this process is performing at Three Sigma,” or “that’s a Six Sigma process”? In this movie I will discuss Sigma Level, how it is used, and what it means to be at Two Sigma, Three Sigma, or Six Sigma. Let’s start with what Sigma Level is. Sigma Level is a performance metric used to indicate the quality level of a product, process, or service. The higher the Sigma Level, the better the performance. A Six Sigma Level or performance means there are no more than 3.4 DPMO, or defects per million opportunities. That’s equivalent to 99.99966% good. Here’s a table showing the different levels of Sigma. Most products, services, and processes operate between Three and Four Sigma Levels. As you can see, the Sigma Level scale is not linear. Improving from Two Sigma to Three Sigma is a five times order of magnitude in the reduction of defects. But improving from Three Sigma to Four Sigma is ten times order of magnitude for reduction of defects. But why bother with Sigma Levels? Sigma Level provides a common yardstick to compare the performance of different products, processes, and services. For defects that can be counted, the Sigma Level is based on DPMO, or the number of defects per million opportunities. DPMO takes into account differences in complexity, and differences in the number of defect opportunities in each unit or confection. For example, the number of defect opportunities in paycheck and the number of defect opportunities in a car are very different. For performance against specification limits, such as delivery within 24 hours, Sigma Level is the number of standard deviations between the mean and the specification limits. The graph depicts a curve showing the delivery times from two warehouses, A and B. A has a wide, normal curve because it has more variation, and B has a narrow, normal curve because it has less variation. It is more consistent in its delivery performance. Even though both A and B have the same average or mean, B performs better. Why? You’ll notice that B’s curve does not touch the specification limit at all, while A does. You will also notice that B has four standard deviations between its mean and the specification limits, while A’s mean is only two standard deviations away. B is performing at Four Sigma, and A is performing at Two Sigma. So now you should know what it means when someone says they have a Four Sigma process or a Two Sigma process, and that a higher Sigma Level is much better. The higher, the better.