Business Success Using Six Sigma tools: DMAIC
DMAIC for Breakthrough Success
DMAIC stands for Define, Measure, Analyze, Improve, and Control. This process is used to create breakthrough improvements in quality: Fewer defects, lower costs, and improved performance of business operations. The five steps of DMAIC are used at two levels: In the overall program; and on each specific Six Sigma project.
But does DMAIC really produce breakthroughs? Sometimes. Read on to learn how to make sure it does.
DMAIC: Diagnosis Before Action
DMAIC in Six Sigma
DMAIC is the core methodology used in Six Sigma, both at the executive decision-making level, and within each Six Sigma project. Everyone in an organization that adopts Six Sigma should understand DMAIC - in fact, universal education in Six Sigma is a key principle of Six Sigma itself. If you're not clear about DMAIC, here is your chance to make sense of it all.
Other articles about DMAIC are usually written by Six Sigma experts, and what they want to do is sell you on Six Sigma. I'm an independent consultant that helps companies decide whether Six Sigma is the way to go, or whether there's a better solution. I'm writing this article because DMAIC is important - even though it isn't always right - and I want you to understand it.
If you want more background on Six Sigma, you can read What is Six Sigma in Operations Management? You might also read an Overview of Leading Continuous Improvement Tools by Global-Chica. If you want to evaluate whether a Six Sigma initiative is a good choice for your business, be sure to check out Will Six Sigma Help Operations Management in Your Company?
What Does DMAIC Stand For?
DMAIC is an acronym that means:
- Define the problem.
- Measure what is happening.
- Analyze the root cause or causes of the problem.
- Improve production operations to eliminate the problem.
- Control production operations going forward to ensure that the problem doesn't come back.
Click on each word for a complete definition. But before we get into details, where does DMAIC come from, and how does it work inside Six Sigma?
Where Does DMAIC Come From?
In the photo on the left, you'll see that DMAIC is similar to the medical processes of diagnosis and treatment. A doctor:
- Defines the illness and sets the intention to cure it.
- Measures the patient's condition with a checkup and tests.
- Analyzes the illness with differential diagnosis to ensure he knows the cause of the illness.
- Improves the patient's condition with medicine or surgery.
- Controls the patient's recovery with follow-up checkups, and, as needed, medical tests and rehabilitation.
DMAIC and medical procedures are similar for a very simple reason. Both are adaptations of the scientific method. Science moves forward through:
- formulation of hypotheses
- testing and modification of experiments
The result is that, through science, we get a better understanding of why things work the way that they do.
Engineering, in general, is about applying scientific knowledge and the scientific method to practical problems. Medicine can be thought of as engineering for the human body. Mechanical engineering is engineering for machines. And Quality Management, including Six Sigma, is engineering for business operations and processes, seeking to understand and eliminate the causes of error to improve quality. In fact, the earliest name for Quality Management, back in 1911, was Scientific Management.
In 1924, a physicist at AT&T Bell Labs, Walter Shewhart, developed two tools that are at the core of Six Sigma and DMAIC. One is the application of statistics to quality management problems. The other is PDCA: Plan, Do, Check, Act, an adaptation of the Scientific Method to improving business operations. A younger physicist at Bell Labs, W. Edwards Deming, developed PDCA and quality management into Total Quality Management (TQM) and took it worldwide. By the 1980s, TQM was transforming industry and helping companies achieve Three Sigma quality, that is, fewer than 3 defects per thousand items produced. That was a huge improvement from a level of about 50 defects per thousand items. In fact, it was such a big difference that Japanese companies using TQM were a serious threat to American automotive, steel, and photocopier manufacturers.
America caught up and wanted to go further. How much further? Motorola and GE set goals of 4 Sigma, 5 Sigma, and 6 Sigma, which, in practical quality terms, meant fewer than 7 defects per million events. When Jack Welch, the famous and controversial CEO of GE announced his Six Sigma initiative, Six Sigma went from being a statistical concept to being a breakthrough goal for American corporations. And DMAIC, a more precise variation of PDCA, was the core method.
The Two Levels of DMAIC: Strategic and Project
In well-designed Six Sigma initiatives, DMAIC is used at two levels. First, it is used at the executive level. The corporation defines its goals and goes through the DMAIC process, with Six Sigma Black Belts assisting senior management in deciding which goals and projects will be most beneficial to the company fastest. Selecting the right projects - the ones that provide the biggest bang for the buck, as soon as possible - is a key component in creating a breakthrough level of improvement.
Here is how DMAIC works at the corporate level:
- We define which problems, when solved, will most improve the bottom line, fastest.
- We measure project goals, comparing one to another. For example, if we only have enough money for one project, do we prefer one that saves a million dollars just once, or one that saves $150,000 per year for 10 years.
- We analyze the benefits of each project for the company. For example, if cash is tight, saving the million dollars comes first. If we're okay for cash, saving $1.5 million over 10 years might be preferable.
- We improve our overall Six Sigma plan by prioritizing each project for maximum benefit.
- As we launch projects and go forward, we control the whole Six Sigma program. For example, if one project is taking too long, we can delay the launch of a later project and give the first project extra help to get it done.
Once each goal has been set, and the goals have been prioritized, DMAIC is used again to plan each individual project, then to execute that project and bring it to success, and then to ensure that procedures are in place so that the benefits of the changes are not lost due to backsliding into old ways of doing things. This work is done by the operations manager in charge of the process to be improved. If he is not a Six Sigma Black Belt himself, then he oversees the project, and the black belt manages it. Six Sigma Green Belts assist on the project, ensuring success by using DMAIC and any of 50 other Six Sigma tools, as appropriate. (Knowing the tools and when to use each one is a key part of Six Sigma Black Belt and Green Belt training.)
For our example, we'll say that our company's biggest problem is that we are getting too many defects in the production of LCD panels. A large defect-free LCD panel can be be sold for a very high price to become a part of a top-end flat-screen TV. But if there are too many defects, the sheets of LCD material must be cut into smaller pieces and sold for computer monitors at a much lower cost. The goal of our project is to reduce (or eliminate) defects in the LCD screens so that we can sell many more large screens at a higher price.
DMAIC Step 1: Define the Problem
The project to improve LCD screen quality is launched, and our engineers begin to define the problem. Each pixel (tiny dot of one color), if it doesn't work, is defective. Looking at current and recent production data, we investigate the problem.
Some investigation was already done before the project was launched. It was necessary to determine the value of the project, that is, the increased revenue that will be gained by reducing the number of failed pixels per million (or billion). Our research will confirm this number, and also lay the groundwork for analyzing the problem.
This is a crucial distinction. Initial definition focuses on the results of the defects and the cost of the problem. That justifies the project. But now, inside the project, we are taking measurements that might help us define the causes (not the results) of the problem. And defining the causes is the first step in analyzing the problem.
We define what defects are occurring and where they happen. (Perhaps all of our problems are coming from one assembly line. Wouldn't that be nice!) But let's say it's not that simple. First of all, some pixels fail and never light up. Others fail because they never turn off. Do those two different problems have one cause, or two different causes. Or are there multiple causes of each problem?
At this point, we have no idea. We're just defining the problem - bad pixels - and gathering all the information we can about the bad pixels on our LCD flat screens.
DMAIC Step 2: Measure What's Happening Now
All that data gets cranked through some heavy statistics, maybe on a supercomputer, and we begin to measure the problem. Sooner or later, with charts and graphs, we can tell you everything there is to know about bad pixels in LCD panels coming out of our factory. When billions of pixels are involved, that's a big job. We create a process definition - a computer model of our manufacturing process. We define Critical To Quality (CTQ) measurements - the things to measure that may be making bad pixels. We can correlate the yield (number of large flat screens produced) to the rate of failed pixels.
Did I mention that this work is done by engineers and statisticians?
In this example, I discussed measurement methods related to yield and quality. On a different problem, say, one where we want to speed up production, we would measure different factors. Cycle time, for instance, the time it takes to create one product, end-to-end, is a critical measure for improving speed, but irrelevant to improving quality.
Now, we've got tons of data, and hundreds of charts, graphs, and diagrams. It's time to analyze our data and find the causes of our problem.
DMAIC Step 3: Analyze the Problem and Its Causes
So, our problem is bad pixels. But why is the manufacturing process failing to produce perfect pixels. We find out when we analyze the problem. At an engineering level, we look for variability in the materials and in the production process that might cause defects. We might find:
- Microscopic analysis of the plastic sheets shows impurities in the plastic. We then trace those to one of the chemicals that goes in to making the plastic, and find that that chemical is not pure enough when it comes from the manufacturer.
- We find that there were more bad pixels in April than in March or May. We find we're totally stumped until an engineer finds himself repeating an old nursery rhyme: April showers bring May flowers. Aha! He correlates the dates and times of bad pixels with weather reports and finds that there are more problems on rainy days. An issue with humidity, perhaps.
- We find that the panels made in the assembly line closest to the east wall fail more around 9:30 to 11am. Perhaps there is excess heat around the machinery.
The engineers who figure all of this out turn it over to the statisticians. The two most important tools are:
- Histograms that display the significance of the sources of error according to Pareto's Law. Pareto's law (also called the 80/20 rule) says that 80% of our bad pixels will come from 20% of the causes. If we find 10 causes, we only have to fix two of them to make 80% of the bad pixels shine just right. The histogram shows us which two causes to fix first.
- Control charts track defects in relation to various causes such as humidity and temperature. Reviewing these will allow engineers to propose solutions that will bring processes under closer control and eliminate defects.
All of this information is brought back to management so that solutions can be designed. Management can look at issues such as whether the problem can be solved in-house, or whether we need to talk to a current vendor or find a new vendor.
DMAIC Step 4: Improve Operations to Create Better Results
Now, we take one problem at a time, and we solve it. There are many possible solutions:
- To get a pure chemical from our vendor, we contact them and let them know the cost of the problem. We might also research other vendors. But this work could involve more than purchasing: If the contract included a guarantee of purity on the chemical, we might have legal recourse to make a claim for damages. We could try to recoup money, or, perhaps, just use the claim as leverage to get attention, then get a quick solution and a steep discount.
- There are two ways to address the humidity issue: Inside the manufacturing plant in general, or inside the machinery. We might assign two teams to come up with solutions, then compare how well they work and which one is less expensive.
- The problem on the east wall might be fixed by an awning or by better temperature control inside the machinery. But what if we get innovative? What if we install solar panels in front of the east wall? That both creates shade, solving the problem, and reduces our energy costs by generating electricity.
Note that solutions can be straightforward, or very innovative. Also note that each cause is solved separately, and the solution might be management (purchasing and legal), facilities (plant operations), or engineering (equipment or procedure changes governing process control.
Another key question is: Is the solution self-maintaining. Or does it require control to ensure it doesn't change?
DMAIC Step 5: Control Production to Ensure Improvements Don't Fall Off
So, we've made our changes. We're getting fewer bad pixels and selling larger LCD screens. That means we're making a lot more money.
But how long will it last?
- Our chemical supplier runs into production problems, and we are getting impure chemicals for our plastic again. If we put in a new procedure to check chemical purity before we make the plastic, we'll catch the problem before manufacturing fails again. But if we just assumed that we had solved the problem forever, the first sign of slippage will be bad pixels - and lost revenue.
- We improve ventilation and air circulation, and install triple doors instead of double doors at the main entrance to keep humidity out of the main plant. That works until, what with climate change, we get new record highs in damp weather. Our solution wasn't robust enough. If we keep tracking humidity, we'll catch the problem early.
- A senior executive gets LEAD certification in green engineering and says, "Those solar panels would produce a lot more energy on the roof." He gets them moved. No one told him that the panels also kept heat off the east side of the building, and the problem comes back. If we document the reason the panels were installed, then we could prevent this error.
Our goal is to make the improvements permanent. In the real world, that's not possible. But we come as close as we can by defining controls and updating information and procedures to ensure that we catch any variation in production or our production environment that can create bad pixels on our LCD screens. That way, we maintain the big boost this project gave to our profit margin.
Does DMAIC Do Damage?
When we do DMAIC right, it works. But any method can be over-used or misused. The very high-end engineering approach of Six Sigma has a downside - it can fail to pay enough attention to the human element. Watch out for these signs that DMAIC is being mis-used, and fix the culture before it creates big problems:
- Lip Service. When training in the value of Six Sigma is insufficient, people can end up talking the talk, but not walking the walk. They use all the right lingo, but there is no integrity, no commitment to the process. Real problems can only be solved in an environment of total honesty. Lead by example and education, including training for new hires, so that DMAIC is part of the culture, not just a buzzword.
- Rah-rah. Don't let DMAIC become a marketing ploy or a buzzword with no real teeth behind it. Make it work for your bottom line, not just your customer brochures.
- Reliance on consultants. Six Sigma initiatives are usually launched with the help of consulting firms, and, of course, training. But build a transition plan to train internal staff. Six Sigma, ultimately, will be maintained inside the company, or it won't work at all.
Does DMAIC Really Deliver Breakthroughs?
Does DMAIC really deliver breakthroughs. The precise answer is, yes, it can. It can, but it doesn't always do it. Six Sigma and DMAIC have to be done well to create worthwhile improvements, and not all of those improvements will be breakthroughs.
Here are some keys to successful DMAIC:
- Be sure to use DMAIC in project selection, not just in doing each project. Otherwise, you may focus on small improvements while skipping over really big problems.
- Note how much understanding and cooperation is necessary for DMAIC to work. Executive, line workers, and everyone in between has to be a part of the solution. Also, people with very different training, such as engineers, managers, and statisticians have to learn a common language and be able to solve problems together.
- DMAIC works best in a well-managed environment. If a recent merger or acquisition or layoff has disturbed the corporate culture, communication and commitment may not be strong enough to solve problems.
- Things always slip: Improving controls without creating a massive bureaucratic burden is always a challenge. Create simple, effective controls to maintain the benefits of your improvements.
In addition, the benefits of DMAIC change over time:
- At the beginning of a Six Sigma initiative, costs of the initiative are high because everyone is learning Six Sigma. But it will pay off, because each improvement has a lasting effect on the bottom line, quarter after quarter.
- Just after that, the biggest problems are solved, and the biggest benefits are realized. This is the breakthrough stage.
- Continuing forward, DMAIC still pays off, but the thrilling huge improvements to the bottom line drop off.
Why do the benefits of DMAIC drop off? There are several reasons:
- Solving the biggest problems provides the biggest benefits. Once those problems are solved, later problems are marginal, and the benefits of solving them are smaller.
- Smaller problems are harder to find. Usually, the first improvement effort in any one process can easily find out why there is variability producing defects. But once those variations are fixed, additional, smaller variations may be coming from a more complex, subtler mix of factors. It takes more time and more money to solve later problems.
- The competition isn't standing still. Initially, our big savings let us show strong revenue, got is bigger market share, and let us lower our prices ahead of the competition. But our competitors are improving quality and making breakthroughs, as well. So we have to keep using Six Sigma and DMAIC just to keep our place in the market.
If we keep using Six Sigma and DMAIC, what do we do when our products are nearly perfect? We apply the same tools to reduce costs and cycle time. This moves us towards the elimination of waste and lean manufacturing, which are the high-end goals of Six Sigma.
We can also extend DMAIC thinking outside of the manufacturing plant. It works in sales, services, and human resources, as well.
Test Your Knowledge of DMAIC
view quiz statistics