By Pete Babich Total Quality Engineering Inc., 1992 Introduction As a result of Total Quality Management activities, more and more companies are conducting surveys to measure their customer satisfaction. Companies are even advertising survey results. Companies are very proud of their product's 90% customer satisfaction levels. Still others boast of 95% satisfaction. Intuitive logic says the higher the satisfaction level, the better product. The business question, however, is how much satisfaction is good enough? Should your company invest $100,000 to improve satisfaction from 95% to 98%? As the Quality Manager of the San Diego Division of Hewlett-Packard (HP), I had these questions put to me by our management team, and I was hard-pressed for answers. After an extensive literature search, I learned how studies showed that it's five times more costly to recruit a new customer than it is to keep an old customer, and that dissatisfied customers tell eight to twenty people about their bad experiences while satisfied customers tell only three to five. These were all interesting facts, but they didn't satisfy my analytical peers. My engineering experience has shown that mathematical modeling provides significant insight into complex problems, so I decided to develop a customer satisfaction model. Defining customer satisfaction The first step in the modeling process was to define customer satisfaction. Companies do this in many different ways. Most surveys ask the customer to grade his or her satisfaction on a one-to-five or one-to-ten scale, where the highest number typically has a description like "highly satisfied." On the one-to-five scale, companies define customer satisfaction as the percent of customers rating their satisfaction a four or five. On the wider scale, satisfaction is generally defined as the percent rating their satisfaction higher than six. HP's surveys used a simpler definition. A satisfied customer is one who would recommend our product to a friend. This seemed to make sense. For some of our mature products, as much as 70% of purchases were made because of prior positive experiences or referrals from associates. Customers who said they would recommend 2 of 5 Dedicated to improving business competitiveness
the product to a friend typically add
comments like "And have done so many times" and "I will only buy HP products." The small percent of respondants who said they would not recommend the product included negative comments such as "I will never buy HP again!" In almost all cases, the negative response correlated to a hardware or software problem or to a misconception of product expectations. Our follow-up efforts allowed us to retain many of these customers, but some didn't even want to talk to us. Since our surveys only sampled customers, we had to assume that there was a finite group of customers who switched brands due to dissatisfaction with our product. Customer dissatisfaction The concept that satisfied customers continue to purchase products from the same company and dissatisfied customers will buy from another is the fundamental premise of my model. Even if your company has no direct competitors, customers purchase substitute products if they are dissatisfied with your product. For example, assume that only one airline company existed: Terrible Airlines. Although Terrible Airlines has a monopoly on air travel if its customers were not satisfied they would avoid Terrible Airlines and travel by train or car. The main point is that customer satisfaction is measured on a relative scale. It is based on the alternatives customers perceive they have. Figure 1 shows my customer satisfaction model. It assumes a closed system of three suppliers providing products of comparable performance and price to a growing customer base. Customers purchase one product during each time period. If they are satisfied with the product, they will always purchase their next product from the same supplier. If they are dissatisfied, they will always purchase their next product from one of the other two suppliers. The algorithm to determine how dissatisfied customers choose their next supplier is a key part of the model. A simple algorithm would be to split them equally between the 3 of 5 Dedicated to improving business competitiveness
other two suppliers. This makes for simple
math, but it doesn't follow intuitive logic. Once burned, customers are more discriminating with their next purchase. They will conduct more research. They put more value on their associates' experiences. Complex models can be developed to describe this process, but in general, companies with bigger market shares will have the largest absolute number of satisfied customers and therefore have the most referrals. This model assumes that dissatisfied customers from one supplier will be distributed to the other two suppliers based on their current market share. The model's equations are shown in figure 2. The equations can easily be put into a spreadsheet program to allow what-if analysis. I created my model using Microsoft Excel. The unit numbers can be adjusted to show market share for each supplier. Now that the model is defined, can it provide insight into tough business decisions? Figure 3 shows how a product with 95% satisfaction would compete against products with 90% and 91% satisfaction. This example assumes equal initial market share. Notice that after 12 time periods, the 95% satisfaction product would basically own the market. Figure 4 shows how that same 95% satisfaction product would stack up against products with 98% and 99% satisfaction levels. In this case, the product would have less than 10% share in 24 time periods. Figure 5 shows the effect of a market operating at lower satisfaction levels. It takes the same satisfaction relationships as shown in figure 3, but increases the dissatisfaction levels by a factor of two. As can be seen in both figures, supplier A achieves market dominance in both cases, but does so much faster in the higher dissatisfaction case (figure 5). This observation would imply that higher satisfaction levels make a company more resistant to competitive moves. Playing with the model, one observes that market growth (the G parameter) does not have much effect unless it is greater than 20% per time period. Rarely does a market grow that fast, so its effects can generally be ignored. In high-growth markets it is 4 of 5 Dedicated to improving business competitiveness
reasonable to expect that initial purchases
are based more on advertising, availability and price than customer satisfaction, but you should not overlook the powerful position a company can have because of high satisfaction ratings of its other products. The model predicts a final value of market share based on fixed satisfaction levels. The final value is achieved when dissatisfied customers leaving one supplier are exactly offset by dissatisfied customers arriving from other suppliers. At equilibrium (A t+1 = A t) a solution can be derived. I will not prove the solution, but instead leave it to the more adventurous reader. For the rest of us trusting souls, the solutions are shown here: A (fv) =(y+z-x)/(x+y+z) B (fv) =(x+z-y)/(x+y+z) C (fv) =(x+y-z)/(x+y+z)
Where (fv) = final value
These equations quantify the value of
customer satisfaction improvement. Just about every person in Marketing can estimate the value of a 1% change in market share, and if you know how much investment is required to obtain that 1% share, you can make business decisions. For example, assume a product has a dissatisfaction level of 4% and 25% of those people complain about noise. Fixing the problem would require a $100,000 investment. Marketing concludes that every 1% share increase is worth $25,000 profit per time period. Not many managers would turn down an investment with that much payback. Summary This customer satisfaction model is not perfect, and it does not account for technological advances. Customer satisfaction is never static and, in most cases, it is always getting better. The time period needs to be defined for each product category. Clearly people don't purchase a new automobile every month. The model does, however, provide a first-order method of quantifying the long-term effect of customer satisfaction. It points out the need to measure not only your own customer satisfaction, but also the satisfaction level of your competitors. It is also one of the best tools I have used to share the concepts of customer satisfaction with new employees. The model has reinforced my perception that constantly improving customer satisfaction is the best way to achieve business success. 5 of 5 Dedicated to improving business competitiveness
ABOUT THE AUTHOR
Pete Babich, MSEE, is the President of Total Quality Engineering, Inc., a company that specializes in applying quality engineering principles to improve business competitiveness. Formed in 1991, TQE provides software, training, and consultation in strategic planning (with emphasis on Hoshin Kanri), product/process quality and reliability improvement, and customer satisfaction measurement and improvement. Previously employed by Hewlett-Packard for seventeen years, he held positions of R&D engineer, Reliability Engineering Manager, and Quality Manager for the San Diego Division. He is an ASQ Certified Quality Engineer, ASQ Certified Quality Manager, and a past Malcolm Baldrige National Quality Award examiner. The author appreciates receiving your comments regarding this paper. Pete Babich President ph: +1 (858) 748-2916 Total Quality Engineering, Inc. fax: +1 (858) 748-0427 15997 Grey Stone Road e-mail: pbabich@tqe.com Poway CA 92064-2363 Web: http://www.tqe.com This paper may be copied and distributed to others provided it is copied intact and with no modifications. It may not be sold or included in another publication without the author's written permission.