About Design of Experiments (DOE)
As its name suggests, Design of Experiments (DOE) is a systematic method which requires the conduct of experiments to study how the changes in input variables or factors (commonly denoted as X) affects the output variable or response (commonly denoted as Y).
DOE is useful in differentiating between the input variables which have the most and the least impact on the output variable. This allows the identification of the best combination of levels (assignable distinct values or states for a factor) which could yield a desired output. As such, users can improve the performance of an existing process or design a new product effectively at a faster rate.
Design of Experiments : A Real-Life Experience
It was a dark day throughout, interspersed with heavy downpours by the hour. It was the end of a day’s session when my colleague and I decided to stall our departure to home given the massive anticipated traffic at 6 pm mark. We headed into a Mamak Bistro to have coffee to compensate our wait, during when we took the liberty to conceptualize how we could run an experiment to explore the critical controllable parameters required to optimize customer experience of the restaurant.
With no paper in sight to pen down our thoughts, I grabbed a napkin from the cashier. DOE is a systematic method to identify the best combination of levels from multiple factors which could yield a desired output. The method requires the conduct of several experiments or runs under varying circumstances of levels to obtain corresponding response data.
“What would be our outcome variable (In statistics, this is known as the output or ‘Y’ variable)?” my friend enquired.
Find out more : What is Y=f(X)?
I replied, “The Y or Output would be our Customer Experience (CX) Rating. It can range from the scale of 1 to 5, with 1 being Least Satisfied and five being Very Satisfied. We can deploy a touch screen near the cashier counter for the customer to rate their dining experience.”
My colleague nodded in agreement before asking, “What would be our Xs then? I believe that there are many factors which would influence one’s Customer Experience (CX).”
If Y is the Customer Experience (CX) Rating, then the different factors which would affect Y and the levels associated with each factor would first need to be identified.
I looked around the restaurant and found the place to be bustling with people. Many people had decided to take refuge in the restaurant while waiting for the downpour to abate. In fact, when we first arrived at the restaurant, we had to wait for around 5 minutes before a waiter could usher us to a vacant table. If we had waited a little longer without being approached by any of the waiter (notwithstanding the expected assurance of vacancy), we would have probably left the restaurant for another nearby place while silently cursing the waiters for not being courteous.
I broke my silence, “The engagement between the customer and waiter is an important factor. We can break the factor into two different levels: The Low Level and the High Level. Levels are different values or types of controllable factors at which experiments are run. For example, if customer-waiter engagement is is a factor which affects CX Rating, then the levels could constitute two different types of engagement; one in which indicates zero or poor interaction by the waiter (reactive approach) and another which indicates a proactive approach, where the waiter voluntarily seeks for customers who appear in need of assistance.”
Find out more: Why Do We Need Design of Experiments (DOE)?
“Interesting but that is just one factor. How about the rest of the factors?” my colleague enquired.
Few searches on Google yielded several papers which highlighted factors that affect the customer experience in a food and beverage dining industry.
With the help of my colleague, I proceeded to construct the following table, which enlists the factors which affect CX rating, including the one which we had discussed initially.
Table 1.0: List of Factors which affects Customer Experience (CX) Rating
My colleague wrote down the following equation describing the relationship between the Y (CX rating) and its relationship with the multiple factors:
“Great, so how do we come up with different levels for each of the factors, bearing in mind that we will need to run experiments at these levels?” my colleague asked.
I stopped fidgeting the pen between my fingers. The task was far from done. We would still need to identify the Low and High Levels for each of the factors. Even for the first factor, X1, we had only described the levels subjectively.
“Let us start with the first factor- the frequency of voluntary customer engagement by the waiter. I would want to run the experiments with two set values or conditions of the factor. I would set the Low Level to be 0 while the High Level to 2…” I barely finished my sentence when my colleague who has a penchant for interrupting my speech voiced up again.
“…meaning for the Low Level of the first factor, you would ask the waiters not to engage the customer unless they are being requested by the customer. For the High Level, you would ask the waiters to approach the customer twice willingly; perhaps when the customer first walks in, the waiter would assist in assigning him or her a table before taking down the order. Secondly, when the waiter would ask for any pending items or additional order after the initial ordered items have been delivered to the customer’s table,” my colleague replied.
We agreed that we would add two columns to the table which we had drafted to include both the Low and High Levels for each factor.
Table 2.0: List of Factors and Levels which affects Customer Experience (CX) Rating
“The table looks good! We can surely run the experiments using these levels,” my friend exclaimed.
His exclamation was followed by a sudden silence. “So far, we have been discussing about running experiments at different combination of levels so that we can determine the combination of levels that corresponds to the highest CX rating. In other words, if I want to achieve the highest CX rating of 5, then DOE will let me know the levels within the factors which would attribute to that rating.”
I nodded in agreement. My colleague continued, “Can DOE work in reverse? For example, if I want to achieve a CX rating of 3, would DOE be able to reveal the possible combination of levels which would contribute to that rating?”
“Yes, you can do that with Design of Experiments. Once you have conducted the experiments, you are able to identify the corresponding levels of factors for Y values within the scale of 1 to 5,” I answered.
The rain stopped and the sky started to become clear. I quickly finished my drink. Noticing that we were about to make a move, my colleague also proceeded to finish his drink and slipped a small paper into my front shirt pocket. It was a note which was scribbled by the waiter earlier, indicating the amount payable for our drinks.
“I hope you will have ample time explaining DOE to the owner when you pay our bills,” he said as he stood up.
INTERESTED IN LEARNING ABOUT DOE?
Design of Experiments (DOE) is one of the several powerful techniques which is taught by MBizM Group Sdn Bhd as part of our Lean Six Sigma Black Belt training cum certification programme. We provide various levels of Lean Six Sigma training cum certification such as Yellow Belt, Green Belt, Black Belt and Master Black Belt. Our Lean Six Sigma-certified trainers and consultants have been conducting training and providing Lean Six Sigma project consultancy services to local and global clients from diverse business verticals over many years.
MBizM Group Sdn Bhd is an Accredited Training Organization, Curriculum Provider and Examination Center by IASCC and by Council for Six Sigma Certification (CSSC). We offer the best price for Lean Six Sigma Training in Malaysia with certified trainers and in complementary of training materials. If you wish to earn Lean Six Sigma certification with experienced trainers at an affordable price, check out our website https://www.mbizm.com/ for more information or you can call us at +603 7873 8485 to speak to our person in charge to know which training suit you best.
We are looking forward to having you onboard!
The actual note taken at Mamak Bistro.