With the interest in Lean Six Sigma, most companies, during their early years, are looking for Lean or DMAIC type projects.
Lean projects are projects which focus on the elimination of waste from the process. They are good projects for early improvement efforts because once people understand the categories of waste, it can be identified by conducting Process (Gemba) Walks and by gathering input from those working in the process. Working to eliminate waste can produce significant gains.
When opting to pursue a DMAIC project the problem solver needs to select a specific process. Based on the boundaries (the first and last step of the process), an output metric from the process should be identified. Most output metrics are either time, cost, or some performance metric, like accuracy or some dimensional characteristic of the product created by the process.
Creating a Goal Statement for a DMAIC project follows this format – a verb (usually “Increase” or “Decrease” or some similar words) followed by the Output metric and then the baseline value for that metric.
In the Bahama Bistro example, a good Goal Statement would be, “Decrease the time to deliver take-out lunches from XX minutes to YY minutes by 4/1/2019.” The purpose of writing a goal in this fashion is that it provides a very clear focus on the process and it defines the expected improvement.
Here lies the problem of using Customer Satisfaction as a project goal. Certainly, knowing the perception of our customers is important to any business. But, Customer Satisfaction is typically determined by conducting surveys of customers or evaluating the comments (both complaints and compliments) from the customers. Hence it is not the output of any given process, but the aggregate perception from customers or a variety of processes within the business.
Customer Satisfaction is typically determined by conducting surveys of customers or evaluating the comments (both complaints and compliments) from the customers.
So, the first problem with using Customer Satisfaction as the improvement metric of a DMAIC project is that it does not provide a single process to improve.
In 1984, Dr. Noriaki Kano published a paper on process performance in relation to customer satisfaction. Now referenced as the Kano model, it divides customer expectations into three general categories: Must Be (Dissatisfiers), Performance (Satisfiers), and Delighters.
The Must Be category are those needs that most customers expect to be performed without problems. In the Bahama Bistro take-out lunch example, these expected needs might include delivering exactly what was ordered by the customer and having fresh ingredients.
The Y axis on the Kano model is customer satisfaction. The X-axis is the ability of the process to meet these needs. In our Bahama Bistro, it would represent the percent of time that these expected needs are met.
In observing the Kano model, it is interesting to note that the middle of the Y axis is neutral (or apathy). This means that if you meet the “Must Be” expected needs 100% of the time, Customers do not feel any great degree of satisfaction – they are neutral. Why not? You are simply doing what you are expected to do. But if the performance begins to slip, the customers become increasingly dissatisfied.
The Performance or “Satisfiers” in Kano’s model are for those requirements where the satisfaction changes as the presence of that requirements changes. A good example from the Bistro is the length of time to deliver the order. If we measure the time, it is a performance type of requirement. This means if we can deliver the order faster, we expect customer satisfaction to increase.
In contrast, “Delighters” are doing things for the customer that are unexpected. Examine the Kano model and notice that Delighters can never cause dissatisfaction. Customer don’t have a negative experience if you do something that they don’t expect or have as a need. Delighters are those things that cause customers to go, “Wow”!
Delighters are those things that cause customers to go, “Wow”!
The second issue of using customer satisfaction as a DMAIC project improvement metric is that it isn’t a single metric at all. It is the sum of two components – The amount of dissatisfaction that the customer feels on some of their requirements added to the amount of satisfaction they experience due to receiving unexpected benefits from the process.
Suppose in our example a meal is delivered to a customer that isn’t what the customer ordered. As the customer discovers this mistake, they feel some amount of dissatisfaction. Now, suppose that when confronted with the error, the restaurant offers to refund the cost of the meal and offers a further discount on future business. As you might expect, the customer feels some “delight” from getting free food, so if asked for their satisfaction, they might report little dissatisfaction, perhaps even a slightly positive level of satisfaction.
The Delighters have offset the dissatisfaction resulting from the lack of an expected service need. Because this “Satisfaction” result is the total of two or more perceptions, it isn’t a good measure of process performance. Imagine, you could have terrible process performance, but if you gave away enough “Delighters” you could completely mask this poor performance – at least as measured by customer satisfaction.
Another issue is how frequently this measure can be collected. Certainly, it is feasible to ask every customer their satisfaction and they would provide you with a result. But, because the result would be based on their individual perception, the results would vary wildly.
Part of the DMAIC process involves checking the amount of measurement error. When the variation in the measurement process exceeds more than 30% of the total observed variation (30% if the reference point suggested by AIAG standards), it is recommended that you fix the measurement system before proceeding with process improvement. I can almost guarantee the measurement error from collecting individual satisfaction will exceed this percentage.
Customers comprise three general distributions of people: Those who are extremely disappointed, those who are “just fine,” and those who are elated with your performance. Which category do you feel most of the customers of Bahama Bistro would fall into? If you suggest those that are “just fine,” I would concur.
Most satisfaction scores use a Likert Scale (ratings of 1-5) for measurement. It is a scale typically used in psychometric research to measure people’s perception. When used in measuring satisfaction, the results are typically calculated as an average of all responses received. This summary statistic makes this measurement even more problematic as an improvement metric, because “averages” tend to hide the variation in the process. It is difficult to show that an average value for this type of Likert scale metric has improved.
Finally, if your process involves the same customers time after time, you must consider how frequently they can be surveyed before the result becomes a rote answer. Most parents would recognize this phenomenon when daily asking their child, “How was school today?” After some time, the answers are no longer a viable assessment of the actual performance of the system.
Does this mean that a Customer Satisfaction score isn’t a valuable metric? No, it is valuable, but as a scoreboard metric, not a process improvement goal. We need to use the Satisfaction score in order to better understand why the customer gave us that response. Where did we fail to meet their needs? We need to stratify those failures and determine which processes produced them. Then, using that data, create a Project Charter that focuses on eliminating those failures.
DMAIC projects cannot be used to “create Delighters,” but rather are meant to “eliminate Dissatisfiers” or “improve Satisfiers.” Creating Delighters simply buys time with customers so that you can improve the process.