Six Sigma Lean DMAIC Process Plan

The Six Sigma formula of DMAIC - Define, Measure, Analyse, Improve and Control - is one of the most effective process improvement methodologies in business today. The essence of Six Sigma Yellow Belt certification and Green Belt training is to deliver a working knowledge of this powerful and structured approach to tackling quality problems.

Six Sigma is of course a leading quality improvement technique that was initiated by Motorola in the early 1980s and achieved prominence under Jack Welch at GE in the 1990s. It has since been adopted by industrial and consumer corporations worldwide.

In simple terms, each business or industrial process either performs to customer specification, or it does not. Failure to do so is a defect, and each process will therefore have a DPMO or defects per million opportunities score. This simply states how many defects will occur per million repetitions of the process.

The DPMO number drives Sigma scores, which measure variation in a process and are graphically defined by distance from the mean under the bell curve of a normal distribution.

Sigma Score --- Defects per Million Opportunities (DPMO)

2 Sigma ---> 308,357

3 Sigma -----> 66,807

4 Sigma -------> 6,210

5 Sigma ----------> 233

6 Sigma ------------> 3.4 defects per million opportunities!

The aim of the DMAIC process is to accelerate and turbo-charge the process towards the world class levels of quality that Six Sigma demands. Two Sigma quality is similar to a spelling mistake on every page of every book in a library. Six Sigma quality is equivalent to one typo in the entire library. The journey to achieve these demanding levels of quality is never easy, but DMAIC provides an ideal framework for both conceptualising the problem and implementing solutions.

DEFINE: Every Six Sigma Yellow Belt or Green Belt project must start with an accurately defined problem. What does the internal or external customer want (Voice of the Customer) and what is not being met? The problem statement must be succinct, quantified and preferably a single sentence. For example - "31% of retail customer orders for Type 4 refrigerators received by the Miami-Dade facility were not delivered within the five day target in March 2010".

The DPMO is clearly 31 per 100 or 310,000 per 1 million, which equates to a two sigma score that is indicative of a poorly performing process. The focus at this stage must be on working out what is truly "Critical to Quality" and will result in genuine customer satisfaction as per the Kano model. Don't try and "boil the ocean" - focus on a discrete and achievable goal. Attack large challenges in stages via multiple projects if necessary.

An important part of the Define stage is developing an effective "current state" process map. This will be ideal later on for identifying the key bottlenecks and constraints in the process.

MEASURE: Effectively and accurately collecting data is pivotal to the success of a Yellow Belt project.You might wish to start with a "Gage R&R", which assesses the measurement system for reproducibility and repeatability. This effectively assess whether your measurement system is working properly. Otherwise, this phase of the process might be as useful as measuring your weight by standing on a pair of broken scales.

Start this phase with constructing some useful operational definitions to ensure consistency in measurement, then set about collecting the data. For example, in measuring refrigerator delivery time you may wish to consult factory product release logs, time-cards and vehicle log books. Alternatively you might have an electronic tag on the product that feeds into a relational database, so keeping track of a delivery and identifying any bottlenecks or delays will be relatively simple.

ANALYZE: This is the meat and bones of the project. Here you can deploy a full array of statistical techniques, ranging from hypothesis testing with p-values to a full-on FMEA,or Failure Modes and Effects Analysis. For simpler projects where the full array of tests is unnecessary, try drawing a simple Ishikawa or fish-bone diagram to identify the root causes of the failure mode. After all, one fundamental concept in Six Sigma is that y=f(x) i.e. that inputs drive outputs.

Some of the biggest breakthroughs in Six Sigma can be made from simply using the "Five Whys" - i.e. asking the question "Why?" five times in a row.

IMPROVE: This is one of the most rewarding phases of the DMAIC process, since it is now that intelligent solutions which address the real root causes of the problem can be implemented. Pilots, brain-storming sessions and trials are all good ways to road-test a solution. More sophisticated projects may need designed experiments which are outside the scope of this hub. Note that solutions need not be rocket science. As a Six Sigma Black Belt, one of my most lasting and significant contributions to improving the performance of a key business process was simply to set up a decent filing system!

CONTROL: Power is nothing without control, and even the most spectacular DMAIC improvements are wasted if they are not entrenched for the long haul. In the industry jargon, you need to "sustain the gain". In the Control phase, the emphasis is on "mistake proofing" the newly efficient process through poka-yoke control solutions. (A good example of a poka-yoke control is a childproof safety cap on a medicine bottle). Use control charts to track process performance on an ongoing basis, for data drives decisions and is essential to maintaining success. Consider a long term MSA plan to gauge measurement system effectiveness and a reaction plan so people know what to do if there is a problem.

Above all, enshrine the newly improved lean process in standard operating procedures and draw up a new process map that shows the new "way to do business". Finally, don't stand still. The natural state of every process is to degenerate towards entropy. Customer quality standards and expectations will continue to improve. Competitors will raise their game on a continual basis. Quality is truly a never-ending journey.

(c) J Rozel 2010

The hidden factory is the rework, scrap and wasted effort that is expended in making any process meet customer specifications. This is typically concealed from the customer but is a cost of poor quality that is borne by the business as an internal failure cost. The classic approach to calculating yield -- final test yield -- does not account for the hidden factory.

A better approach is to calculate rolled throughput yield, which measures the yield at each step of internal testing, to form a reducing balance percentage at each step. The final result may show a dramatic lesson in just how much waste and inefficiency exists in the manufacturing or business process.

Those particularly interested in this topic should watch out for future hubs on process costing, where scrap and rework are an intrinsic part of the costing calculations.

A "hidden factory" of wasted time, or time spent on unproductive pursuits, also exists in service sector environments. One of the best tools for identifying and eliminating this wastage is a Yamazumi board, which graphically represents the efficiency of each step of any business process through highlighting in red, yellow or green. Red is a delay or "hidden factory", yellow is unavoidable but non-value-added set-up or admin time, and green is the cutting edge of value-added activities.