Steps in the Measure phase
– The Measure phase is the second phase of a Six Sigma project. It is the M in DMAIC. Let’s start with a basic question. Why bother doing the Measure phase? What is the purpose? The Measure phase is to measure the size and scope of the problem or performance gap. How big, how bad, how widespread it is. Say you are the owner of a pizza chain of 10 restaurants in the local area, and your sales have been declining over the past year. Initial analysis of the year’s complaint data shows that 85% of complaints were related to pizza crust. So Where would you focus your project? Pizza crust, of course. Remember in Six Sigma we use the equation y is the function of x, which is y equals f of x. In this project, the y is the pizza crust. Wouldn’t you want to know how big and bad the pizza crust problem is? And wouldn’t you want to know how widespread it is? For example, is pizza crust problem equally bad across all 10 restaurants? Are some worse than others? Or is it limited to just one restaurant? That is what the Measure phase does, it measures the y. The purpose of the Measure phase is to measure the y in y equals f of x. What are the stops involved in the Measure phase? Before you spend time and resources collecting data and measuring, you need to develop a data collection plan. That’s step one. To help you know where to collect data, you need to understand activity or processes involved that impact pizza crust. You need to map the relevant processes. That’s step two. Before you start collecting data, or using data that has been collected, you need to ask, are the measurements valid? Were they measured and recorded correctly? You need to ensure that a the data and measurement system used is valid. That leads us to step three. Then and only then can you measure how bad the problem is. How bad compared to what? How bad compared to requirements, specifications or targets? In other words, how well are requirements and customer expectations met? How often is it met or not met? How widespread is it, and so on. Use the collection plan developed earlier to guide you in measuring the y, so that all the questions listed on that plan will be answered. So step four is, measure the y in y is a function of x. In summary, here are the steps in the Measure phase. Develop a data collection plan, map relevant processes, validate the measurement system, measure the y. Also often, project teams waste time going in circles, trying to measure everything in this phase. You, on the other hand, will not. These steps will help you plan and execute the Measure phase.
How to map the current process
– If you have problems getting to work on time in the mornings, would it not be useful to understand the activities and sequence of steps that take place each morning? In other words, what does your wake up and get to work process look like? What you need is a process map. A process map is a diagram that provides a visual representation of the process flow, or sequence of activities, or steps that take place in a process from start to finish. Here’s an example of a process map for the Wake-Up-and-Get-to-Work Process. Activities or tasks are represented by rectangles, decisions are represented by diamonds, and the sequence of flow is represented by arrows. And start and end points are represented by ovals. If you run out of space, connectors shaped as circles can be used to continue to the next page. The flow can go from left to right, or from top to bottom. Here are some common symbols for process mapping. Process maps are very useful in a Six Sigma project. During the measure phase we map the current process using a process map. This provides the project team, and everyone else involved, with a visualization and common understanding of what actually happens in the process. Think of it as a photographic snapshot of the actual SS process. A process map enables you to see and understand the process. A picture is worth a thousand words. Also, once a current process has been mapped, the team will also know what’s not happening or what’s different from what should be happening. Therein lies the power and benefit of mapping the current SS process. In our example, you get a better understanding of what goes on each morning. It will help you decide where in the process to collect data and analyze further, or it may show immediately where the delays, bottlenecks, and work-around’s are that is causing you to be late for work. The process map provides a basis for understanding, and further analysis. When you map a current process, I would recommend the following: Involve those who work the process, and those who know the process. They are most familiar as to what actually happens. Decide on the start and end points of the process map. Use a logical orientation. Either from left to right, or top to bottom flow, to capture the sequence. Walk the process as if you are the transaction or work item that is being processed from start to finish. For example, if you are mapping the invoice process, be the invoice. Walk step-by-step as if you are the invoice. Document what happens to you at each step of the way. Do not try to analyze or improve the process at this time. Just capture the as is of what actually happens. You are mapping the current process, not proposing any solutions. When the process map is completed, have it reviewed and validated by those familiar with the process, including those who work the process. You can map any process, including your morning routine. Follow these tips and you’ll make it to work on time.
Plan for data collection
– Let’s assume you decide to get on a health and fitness program. What is the first thing you do? Well, you’ll probably want to know what is your current weight, how much you need to reduce to get your desired target weight, and what is your current level of activity, are you in good health, and so on. These are very good questions to ask as you plan to collect data in order to measure where things stand today and to measure progress. Similarly, that’s what you need to do in a Six Sigma project. In this movie, I will discuss how to plan for data collection during the measure phase of a Six Sigma project. Why plan for data collection? Collecting data is not free or cheap. Collecting data requires time, effort, and resources being tied up. All too often, project teams who do not plan for data collection spend too much time during the measure phase going in circles, hunting for data, collecting data, and measuring everything in sight. Perhaps it would be useful to ask a basic question regarding its purpose. Why do we collect data? To obtain information and answers to questions we want answered and that’s where you need to start, with the questions. To plan for data collection, I recommend the following steps. Step one, make a list of specific questions you want answered regarding the Y. Step two, for each question listed, decide how you want the answers presented and displayed. Step three, based on that, decide what tools are needed to make that possible. Step four, for each question, what type of data and how much data are required by that tool. Step five, determine where and from whom the data should be collected. Going through these five steps will result in a data collection plan. For example, with a pizza crust problem for our chain of 10 restaurants, specific questions to be answered include how big and bad is the pizza crust problem, how widespread is it, is the pizza crust problem equally bad across all 10 restaurants, are some worse than others or is it limited to just one restaurant? Those are the types of questions you want answered regarding the Y, which is the pizza crust. That is how you plan for data collection in the measure phase. And the benefits to you are knowing exactly what data to collect, how much data to collect, from where and from whom. By planning correctly, you’ll avoid common pitfalls of spending too much time in the measure phase going in circles, data hunting, and measuring everything in sight. A data collection plan will help you carry out the measure phase effectively and efficiently.
Types of data and graphs
– Throughout this chapter, we have been using the example of a pizza chain with customer complaints about their crust. Let’s return to our example to help us understand the types of data and graphs used in Six Sigma projects. With a pizza crust problem, you would want to know if the thickness of pizza crust is consistent and how many complaints on pizza crust were received by each of the 10 restaurants. There are two types of data, continuous and discrete. Knowing the type of data can help you determine which graphs to use. Continuous data is data that can take on any value on a continuum or continuous scale, such as inches or centimeters on a measuring tape, or pounds or kilograms on a weighing scale. Continuous data can be expressed as decimals or fractions, such as 2.2 meters tall, or weighing 175.6 pounds. Discrete data is data that take on values which are integers or whole numbers. In other words, discrete values. These are usually numbers that you can count. Examples include number of complaints, number of people in each checkout line, the number of on time departures, and the number of late departures for airlines. Proportions can be calculated from these counts. For example, the proportion of M&M’s that are green, the proportion of flights that are on time. Because of categories used such as color of M&M’s, on time departure or not, and store location A, B, or C, discrete data is also called categorical data. Categorical data is a more contemporary name for discrete data. Why is it important to know the type of data? This is because it impacts the type of graphs that can be used. For continuous data, graphs can be used to show variation. For example, a curve showing the variation in the thickness of pizza crust. Graphs and charts commonly used in Six Sigma projects for continuous data are histograms, dotplots, and boxplots. These graphs display the pattern of variation, showing how spread out the measurements are, and where they are centered. For example, here’s a histogram and a dotplot showing delivery times. The Six Sigma project team can see that even though it averages 24 hours, there is a lot of variation in delivery performance. It can take as short as 15 hours or as long as 31 hours. For discrete or categorical data, bar charts and Pareto charts are typically used to contrast the frequency count in each category. The Pareto chart is the most popular chart, as it shows and prioritizes based on how frequently each category occurs. For example, here’s a Pareto chart showing which complaint type occurs most frequently. In this case, complaints on pizza crust stands out most, accounting for 70% of all complaints. Armed with this information, the team can focus their project on pizza crust problems. For both continuous and discrete data, to show trends over time, graphs such as line graphs, time-series charts, and control charts are used. This chart shows delivery performance over time. Knowing which types of data and graphs to use and how to (inaudible) them will help you answer key questions during the Measure phase, such as how big, how bad, and how wide spread the problem is.
Measurement system analysis
– In the morning when I step on a bathroom scale, I am x pounds. When I visit a doctor’s office that same day, the nurse weighs me and I end up at more than x pounds, which weight is correct? So which weighing scale should I use as I train for the Olympics? In this movie I will discuss the importance of ensuring that data is valid before using it. In Six Sigma projects, this is done using a technique called Measurement System Analysis, or MSA for short. MSA is done early during the measure phase so that any data to be collected and used is valid. The Measurement System includes the entire system including the measuring instrument, the operator, and the procedures used to collect, measure, and record data. To be valid, measurements should be accurate with no bias, repeatable, and reproducible. Let’s go back to my weight example. To be accurate with no bias, the bathroom scale should be centered on zero when there is no one standing on it, although the temptation is to offset it by negative two so that I feel better that I’m two pound lighter. In other words, the equipment should be calibrated. To be repeatable means that if I weigh myself twice, or even three times on the same scale, I should get the same readings. The measurement is said to be repeatable. To be reproducible means that my weight at home should be the same as my weight in the doctor’s office. If that is the case, the measurement is reproducible. However, in my case, my weight measured by the nurse is not the same as my weight at home so the measurement was not reproducible. When the measurement system analysis, or MSA, is carried out, 20 to 25 known items are selected and in a blind study, each item is measured twice by at least two operators. Any inaccuracy, repeatability or reproducibility problems with the measurement system can be determined. And if problems are corrected, and the follow up MSA is carried out, to verify that the measurement system is good, and data collected is valid. MSA has to be done before any data is collected or used. MSA is applicable to both types of data, it is applicable for continuous data such as weight or transaction time, and it is also applicable to discreet or categorical data such as the classification of defects or complaints into reason codes. You need to perform a measurement system analysis to ensure that data is valid before it can be used, otherwise, your project will be side tracked chasing the wrong problem, analyzing incorrectly, or drawing the wrong conclusions. You definitely do not want to be a victim of invalid data. As the old saying goes, garbage in garbage out.
Process capability and sigma level
– Imagine, you own an online retail company that promises delivery within three to four hours. And average delivery time reported for this month is 23 hours. This is a second month that it is 23 hours. And everyone in the company is celebrating, because 23 is less than a 24-hour maximum. But wait. Why are there still so many complaints regarding late deliveries? Hmmm. In this movie, I will discuss how to measure performance in terms of process capability and Sigma level. What is process capability? Process capability refers to how capable or how well the process performs against it’s target requirements or specifications. Most of us have driven into garages with height restrictions, where there is a horizontal bar that hangs at the entrance. Cars drive right in without slowing down, because cars are not tall and are very capable of meeting the maximum height allowed. But big trucks or SUVs always slow down at that entrance, just to make sure they don’t hit that bar, because many are not capable of meeting the maximum height restriction. Similarly with any process, how capable is the process in meeting specifications? In our delivery example, if we plot performance for the past 100 days, it looks like this. Yes, the average or mean is 23 hours, but there is a lot of variation around that 23-hour average. So much so, that there are many deliveries that exceeded the 24-hour limit. That 24-hour limit is also called the upper specification limit. Why upper? Because any delivery more than 24 hours is considered defective performance. From the graph, we can see why there are so many complaints about late deliveries. Now that there is an upper specification limit, is there a lower specification limit? No. Not in this case. Since there is no requirement to not deliver it too early. Notice, that 16% of deliveries are late. We say that the defect rate is 16%. So, 16% of the time, the company is not capable of delivering on its 24-hour guarantee. Looking at it from the other side, 84% of the time it can. We say that the yield is 84%. To express this in terms of Sigma level, a Sigma conversion table can be used. A 16% defect rate is approximately a 2.5 Sigma level of performance. There you have it. Process capability. In summary, process capability can be expressed as a defect rate, a yield rate, or Sigma level. In the measure phase, a key benefit of measuring baseline performance in terms of process capability, is that it enables you to know just how often, how well, or how badly performance meets specifications. Specifications which you agreed and promised to customers. Another benefit is that improvement results at the end of the project can be measured against this baseline to show just how much was improved. We cannot improve what we don’t measure.