How to Implement Six Sigma Smarter Solutions Using Statistical Methods

Use Statistical Process Control to identify smarter solutions.

Implementing Six Sigma smarter solutions, a business management strategy, using statistical methods is easier than you might think. Start with a problem statement and conclude with a "lessons learned" review. The process steps in between are determined by you, based on data collected from the process, the objective defined in the problem statement and by applying the DMAIC critical thinking method. DMAIC is an acronym for the five phrases that define one of two Six Sigma methodologies that will be explained in the following steps. Start implementing smart Six Sigma solutions to any process and document the results.

Instructions

1. Define the problem statement, such as delivery times missed. Convert the problem statement into a wish statement with a goal or measurement, such as delivery times met 100 percent on time. This will be your objective.

   Define what you are yielding for current delivery times. Define industry standards. Define process starts and stops, for example: The process begins with taking a design, an order, a patient or a customer and it ends with the customer paying or the customer committing to contract terms. Define the processes by creating flow charts. Define what you will measure as critical to schedule key process inputs, for example: product availability, personnel availability, machine availability, communication flow or response rates impact schedules. Determine what areas are affecting your objective.

2. Measure the differences in what you are yielding and what you need to yield on key process inputs. For example, measure the number of steps in the process and measure the quality results at each stage. If upstream processes, such as approvals and inventory, are not available when you need them, no matter how hard you try, you will never meet the goal. Measure the amount of inventory. This could be product, calls or people in a service line that are sitting in queue at each stage. Measure the amount of people working on each process. Measure the number of machines that are supporting the people. Measure machine and manpower uptime compared to downtime. Measure variability from key processes that yield variation. If you don't have a reliable measurement system, create one. It does not have to be difficult; the number of paperclips in a jar could indicate how many times a data entry clerk was interrupted, how many customers left without being served or you can use poker chips to represent the number of pieces failing inspection.

3. Analyze results of data collection efforts. Compile collected data and use Attribute R&R charts for data, such as pass/fail, go/no go, pass/reject. Use continuous loop R&R charts for variable data, such as discrete numbers, addresses or customer-ship-to addresses. Use histograms to show trends and box plot charts to show repetitive patterns. Create cause-and-effect diagrams and benchmark and brainstorm for opportunities. Examine root cause of why problems occur.

4. Improve the process by setting priorities and addressing short- and long-term needs. Plan long term for additional capital needs, such as equipment or personnel. Sometimes these things take a little bit longer to implement. At the same time, address low-hanging fruit or easy fixes, such as shifting timing of hours or labor force location, realigning management approvals, consolidating or expanding job tasks, calibrating equipment, modifying standard work, providing additional training, adding quality inspection or removing value-added steps that actually provided no value. Set measurements against the inputs instead of the outputs. For example, measuring missed delivery numbers is too late in the process to do anything about. Measure how long it takes from the first step to the second step and focus on gaining control of these by setting upper and lower control limits of an acceptable range. Work to continue to reduce this upper range over time.

5. Control the process by measuring it and looking for variation. If measurements suddenly go out of the norm, find out why. Use visual indicators to identify problems. For example, in a grocery store checkout line, tools used to control the process are both at the customer and management levels. When the cashier flips a light switch, the front line manager responds to the call. Without leaving the area, the cashier has visually indicated to the manager that they need help. Document the lessons learned and share it with others to reduce the learning curve. Encourage employees to provide feedback and ask customers to do so as well.

Tips & Warnings

• Standard deviation or sigma in statistics is the square root of a data set's variance. It shows the relationship of all pieces of the data set to the average. It is often depicted with data on a bell curve with the center of the bell curve expressing the mean. The bell curve has six segments to it. The closest segments to the right and left of the center indicate a 95 percent or better confidence level. Data that falls within the 95 percent range indicates a process in control, meaning that it's consistently producing the same results. The segments on both sides of the outer edges of the bell curve normally represent between 1 percent and 5 percent of the total outcomes. These are the data points that are out of control and give you a point to start your investigation. The middle segments on both sides of the bell curve make up the difference and represent data that is under 94 percent and above 2 percent to 6 percent or around the 64 percent confidence level. This is the area to focus on next with the objective to shift the bell curve and data points closer and closer to the 95 percent range.

• There are many free online software tools to use to analyze your data. Focus on getting the right data first. Then focus on analyzing it. Using bad data can lead to costly mistakes.

References

• Six Sigma SPC