In this article, we will see what variation is and how to reduce it. Reducing variation is the working area of six sigma. Six sigma states that when we reduce variation we will be able to improve our processes. This can be a quality improvement, a cost reduction, or a lead time reduction.
What is variation
Variation is an important aspect of a process. It occurs in all processes. This means that not only products are affected but services as well
One of the most know forms of variation is when we play a game that has a dice. Whenever you roll a die, you know that the outcome will be a number between one and six, but you cannot predict which exact number the outcome will be. This means there is a variation when rolling a dice that you cannot avoid.
The outcome is considered as an independent random result. We can blow on the dices, let somebody else throw the dices, the changes that you throw a certain number will remain the same.
central limit theorem
In six sigma, the central limit theorem is one of the keystones. The central limit theorem states that, in many situations, when independent random variables are summed up, their properly normalized sum tends toward a normal distribution
Okay, let’s break this down with an example!
One of the most classic representations of a random variable is a dice that we use in many games. A classic dice has 6 sides and you have a change of 1 to 6 to throw each number. If we use multiple dices we get several independent random variables since the outcome from one dice has no effect at all on the outcome of the other dices. If you throw a “6” with your first dice, it does not affect the chance that you throw again a six with your second dice. It is still a chance of 1 to 6.
Let’s say we have four dices we will throw. This means that the sum of the dices will be between 4 and 24. The central limit theorem says that if we create a graph with on the X-axis possible outcome of a throw and on the Y-axis the amount of times the number has been thrown and we throw enough times the shape of the graph will convert towards a bell curve. Also called a normal curve. This can be seen in our simulator.
The bell curve shape is symmetrical. The center is at the number we throw the most often. This is also called the mean. There are two Inflection Points. These are where the curve changes concavity. The distance between the mean and the inflection points is called the standard deviation. We will see later on in this lecture what a standard deviation exactly is.
A real-life example of variation .... coffee
Let’s translate this example into a real-life example. The strength of your coffee depends on several variables:
The type of coffee, the amount of coffee, the amount of water, the temperature of the water, the filter system of your coffee machine, And each of these variables depends again on other variables. So can the amount of water depend on the way you are measuring the amount, the water crane which is used, the person that is filling it, the water pressure… and these variables are depending again on other variables and so on.
Adding all these variables up will result in the fact that the strength of your coffee can follow a normal distribution.
Depending on the process you can have more or less variation within your process.
So variation in a process is just as natural as it is for us to breathe air in and out. Differences in materials, machines, environment, measurements, manpower and applied methods cause variation in the output of our process. These are also the categories you can find back on a fishbone diagram.
how to reduce variation
Before we look at how we can reduce variation, let us quickly answer why we should reduce variation!
Customers are, just like all humans, creatures of habit. Which means they like consistency. In other words, they do not like variation. They expect a certain level from your product and or service each time over and over again. If you deliver they will be satisfied, if not it will impact your revenue one way or the other.
Only working with averages could make us blind to certain problems within our process. High process variation will result in lower customer satisfaction and lower customer loyalty. It can even damage your company image. If this happens, you will have to spend a lot of resources in order to regain the trust of your customers. Therefore, preventing customer complaints is better than solving customer complaints.
We need to use our resources to reduce the variation instead of firefighting customer complaints. Reducing variation will help us to make our process more efficient and therefore increase revue and lower quality complaints.
The first step we have to take is to quantify the variation.
There are typically three ways to do this. The range, the variance, and the standard deviation.
The Range is the difference between the highest and lowest values.
The variance is the sum of the squared deviations of n measurements from their mean divided by the number of samples.
The standard deviation is the root of the variance.
Remember the standard deviation is the distance between the mean and the inflection point on a normal curve.
comparing performance, an example
Let’s illustrate this again with an example of two drive-throughs of two fast-food restaurants! Customers do not like to wait more than 10 minutes. They both claim that the mean waiting time in the queue is 7 minutes. So, for a hungry client, both restaurants seem equally good. However, this can be misleading. We should analyze also the variation of both restaurants.
But the standard deviation is 3 for restaurant one and 2 for restaurant two!
With these results, we can conclude that restaurant two performance overall better than restaurant one and will have generally more happy customers.
Eliminating sources of variation
Now that we know how to quantify variation, we have to find and eliminate the sources of the variation. For this, we can use the fishbone which we have seen during the PDCA lecture.
Let’s illustrate this with an example
A pizza delivery company sees a big variation in delivery times and wants to reduce the variation. they use the fishbone diagram in order to better understand the possible causes of variation.
They start creating a fishbone diagram using the 4M categories, which is a lighter variation of the 6M categories. The 4M categories are Manpower (or people), Machine, Material, and Method. They write everything they can think of which could cause this variation on the fishbone diagram.
For manpower, they write down: drivers get lost. The driver can get lost for several reasons so they create sub-causes: GPS not up to date, don’t know the town, and address not correct! Another cause they write down is not enough delivery guys.
Let’s look at the machine category. The pizza delivery company uses both bikes and cars to deliver the pizza. Thus the difference in transportation means could lead to variation. Car or bike breakdown could be another cause. another one, Not enough ovens to bake the pizzas, and another one, the credit card machine is broken!
For materials, they write down the following. run out of ingredients, and difference in cooking time between pizzas
Finally, For method, they write down the following topics: planning of the number of delivery guys. The pizza delivery company has to determine the number of delivery guys they need for delivering all the pizzas in advance. Too many delivery guys will result in overstaffing and losing money, not enough delivery guys will result in understaffing and long delivery times. It would be even possible that cold pizzas are delivered to the customer.
Poor dispatching could be another cause, poor handling of large orders is another one.
It is not worth discussing too much where a cause should be placed on the fishbone. The most important thing is that the cause is placed on the diagram.
After the team has created the fishbone they discuss together what the most likely cause is and decide that not planning the number of delivery guys correctly is the most likely cause.
5 why method
In order to reduce variation, they have to find and solve the root cause of why the planning of the number of delivery guys is not correct. The 5-why method can be used for this.
- The first question goes as follows. Why is the planning of the number of delivery guys not correct? The answer is that the number of orders is not predicted accurately enough.
- The second question goes as follows. Why is the number of orders not predicted accurately enough? The answer to the second why is: the number of orders is estimated by gut feeling.
- So now we can ask, why is the number of orders estimated by gut feeling? The answer to this question is because there is no prediction model!
- Why is there no prediction model? Because there is no historical data available.
- And finally the last question. Why is there no historical data?
The final answer is that the number of orders placed is not recorded. This is our root cause!
The team starts to note down the order per time slot and after some time they see a weekly pattern. Based on this pattern they adapt their planning and see that the variation of the delivery times is reduced.