DMAIC's Analyze

Analyze is the third step of the Six Sigma five step process DMAIC. The objective of Analyze is to analyze the current state data and determine the root causes, the opportunities, to improve. Here we take a more in-depth look at the data collected in measure and try to determine the root causes of the issues that the data shows us. Many times, we have to go back out and take more data from the process in what I call a “Deep Dive” to determine the real root cause of what we see in measure. Remember that as you get opinions of what is causing thing to happen go collect the data to back up (validate) that opinion. Here are a series of questions that you will want to answer before moving on to improve:

1. What are the perceived causes of the process variability and which can we control?
2. What is of value to the customer?
3. What are the detail steps of the process?
4. Have you validated the “As-Is” causes?

It is very important that you get all of these answered. Some that stand out are what is of value to the customer and have you validated the “As-Is” causes. Make sure that the root causes that you find really do impact something that the customer really cares about; something that is of value to them. If you are not working causes that impact customer value then they will never return value back to the company and you will be spending time fixing something that the customer really does not care about. If you find this is true the solution maybe to totally eliminate this step since the customer finds it not of value. If you think that you can not eliminate it, then ask yourself what customer does find it important to do. We always are creating value for someone even if it is you. If you are not then you are just wasting your time.

The second thing I mentioned was to validate the “As-Is” causes. You may have had a group meeting and come up with some causes of effects you saw in measure that your customer really cares about, but until you go collect data on those causes and validate the opinions that they cause what you saw in measure you should not move forward to improve. Remember we are all about facts and data that support what we are doing.

To answer the question above it may take several tools and techniques to collect the “facts and data.” So here in Analyze, there are several good tool and techniques that can help get you that data.

• Value Flow Analysis – This analysis take the process observation log[1] that one usually creates in measure and reviews each step listed and the time it takes to do that step, to determine if it is of value or not. Usually you will sum up all the times of the value added and non value added steps. In this way, you can see how much time you spend creating value for your customers and where there are steps that need to be reduced or eliminated because they do not create any value for your customer.

• Cause & Effect Diagrams – This diagram organizes group knowledge about causes of a problem and displays the information graphically. It was invented by Dr. Kaoru Ishikawa and is sometimes called a called Ishikawa Diagrams. Some see this diagram and think it resembles a fish skeleton and that is why it is sometimes called a Fishbone Diagram. In the Cause and Effect Diagram, you start by drawing a box on the right or left hand side and in the box you put the effect that you saw in measure. Next, you brainstorm[2] causes of that effect and add them to the “branches.” Sometimes you also combine it with the five whys[3] tool to get the detail branches and the true “root” cause or causes of the effect.

• Scatter Plots – This plot is used to visualize the relationship between two variables. In the plot below, we are looking to see if there is a relationship between weight and days. If there was no relationship, the points on the chart would be scatter randomly all over the chart. In this chart, there is a relationship that shows as the number of day’s increase the weight also increases.

• Confidence Intervals – These are statistically created to give you an area that you will feel confident that the real value will be found in. Say you make 12 inch rules and you like to be very confident (99% sure) that all your rules are 12 inches. Well you first have to understand that everything varies and so do your rulers but how accurate are they? You would take a sample of them and calculate the confidence interval for their length. You might find that you can be 99% sure that any rule you make is within 11 7/8 inch to 12 1/8 inch or 12 inches plus or minus 1/8 of an inch.
• T Tests – This is also called the “Student’s t test”. This is statistically created when we want to compare two group averages and determine if they are the same or different. Many times, we want to know if a process has changed or shifted from what it was doing before. This test would tell you if it did.
• F Test – This is statistically created when we want to compare two groups variations or variability. Let me try to explain this thing called variation. Everywhere you look, you see variation. In the people, the traffic at an intersection, or even pencils in a box. Let’s say you are buying pencils and there are two boxes of them on the shelf made from different companies. Both make pencils 10 inches long but you want to know if one company’s are more consistently 10” long. You would use this test to see if the variation in pencil length in one box versus the other is different or the same.
• Chi Square Test – Not every thing is “measurable” sometime all we have is count data. The Chi square Test is statistically recreated to see if two group proportions (percentage of a count) are the same or different. Say, you run a grocery store and the current shipment of eggs seem to have a lot that are cracked, and you want to know if the next shipment that just arrived is any better. Here you could take a sample of what is on the shelf and a sample from the new shipment and find out if the two shipments have the same number of cracked eggs or not.

n      The above are tools that I have not talked about before, other useful Analyze tools that I have talked about that are; Brainstorming, LCS, Affinity Diagramming and 5 Whys from my article The First step of DMAIC – Define. Plus  DE & UDE, Gauge R&R, Basic Summary Stats and Pareto Charts from my article The Second step of DMAIC – Measure

Well there you have it, a little more understanding of the Analyze step of the Six Sigma 5 step DMAIC process. I hope that this gives you a better understanding of what to questions to answer and what tools to uses to get those answers in Analyze.

Peter Bersbach

Six Sigma Master Black Belt

Bersbach Consulting

From Process to Profits

1.520.829.0090

[1] Process observation Log was talked about in the article “The Second step of DMAIC – Measure”

[2] Brainstorming was talked about in the article “The First step of DMAIC – Define”

[3] The Five Whys was talked about in the article “The Second step of DMAIC – Measure”