Cleaning up

Editor’s note: Steve Lewis is co-founder of Development II, inc., a Woodbury, Conn., research firm.

When Kazuhiro Matoba, chairman of Miyago Co., Ltd., decided to survey his customers in Japan, he wasn’t quite prepared for the wild ride that would ensue. Miyago, located in Yokohama, is one of Japan’s largest janitorial firms, providing custodial services to many of Japan’s prime public, office and retail spaces. Founded in 1972 by the entrepreneurial Matoba, Miyago now employs 2,000 people in a country that expects exceedingly high service levels.

When our company first approached Matoba about this project, he said, “Our main management strategy is not about revenue itself. Our most important policy is to improve our customers’ satisfaction. We need to measure and benchmark their satisfaction and then improve it.”

In 1997, during the first year of this study, we contacted 180 of Miyago’s largest customers, including bus terminals, grocery chains, retail outlets and office buildings. An average of five people were contacted from each of these customer firms. These same customers have remained relatively constant over the course of the four years we have been conducting this customer satisfaction study.

Delve into the detail

Our first task was to explain to Matoba and his management team how to look at their customers’ responses to the forthcoming survey. While most people are used to looking at the “satisfied versus dissatisfied” division, few are prepared to delve into finer detail. We divide responses into five basic groupings: totally satisfied, somewhat satisfied, somewhat dissatisfied, totally dissatisfied, and no opinion (or no response).

When a customer answers a question with a “totally satisfied” response, their relationship with the company is clear. “Totally satisfied” is effectively saying that there is little or no room for improvement. A “somewhat satisfied” answer indicates that something is not quite right with the relationship and improvements are required. A “somewhat dissatisfied” customer is not quite lost, but almost, and a “totally dissatisfied” response indicates that the customer is about to make, or has already made, another vendor choice. We also found that the word “total” translates easily and that its meaning is definitive across all languages.

By merging the satisfied or dissatisfied responses into one measurement, the message becomes cloudy. Our long-term analysis indicates that the purchase decision profile of a “totally satisfied” customer is very different from that of a “somewhat satisfied” one. One strategy is necessary to keep the “totally satisfied” customer in that position. Another is required to move a “somewhat satisfied” customer upward. If the survey results are combined, how can one differentiate, implement, and measure the success of corporate initiatives at two distinct levels?

Adjust thinking

As the survey progressed, the Miyago team had to adjust their thinking about satisfaction, for the percentage of “totally satisfied” customers is significantly lower than the combination of all satisfied customers. Also, the “somewhat satisfied” category of customers is usually the largest segment.

When we look at the range of responses from “totally satisfied” to “totally dissatisfied,” we see that responses on either end of the spectrum are definitive, easily measurable points. When we move into the “somewhat” categories, the customers’ responses become less conclusive. There is no guesswork needed as to where a “totally satisfied” (or “dissatisfied”) customer stands. A “somewhat satisfied” customer, on the other hand, can easily slip over the line to “somewhat dissatisfied” (or vice versa), dependent upon any number of factors, some seeming quite minor but nonetheless critical. In other words, while the end point of a “totally” response is clear, a “somewhat” response has a wide latitude.

With Miyago, as well as in other studies, we have found that the purchasing decision profiles of customers are fully dependent upon their level of overall satisfaction with the company. For example, a customer who states that they are “totally satisfied” with Miyago overall is twice as likely to repurchase from Miyago as is a customer who says they are “somewhat satisfied” with Miyago overall. Likewise, that same “totally satisfied” customer is 3.5 times more likely to recommend Miyago to a business associate than is one of its “somewhat satisfied” customers.

In a separate analysis, we discovered that those customers claiming they are “totally satisfied” with the company overall are indeed a firm’s most loyal customers. Since loyalty can really only be assessed after the fact, once the repurchase intent has been actualized, measuring total overall satisfaction is the most effective method for determining loyalty.

Once we realized that “totally satisfied” customers have different purchasing decision habits than “somewhat satisfied” customers (and, obviously, “dissatisfied” customers), we needed a way to focus in on the key drivers of overall satisfaction. We needed to identify what attributes drive satisfaction and the level of impact each attribute has upon increasing or decreasing overall satisfaction.

Optimum use of resources

The most important conclusion that can be derived from Miyago’s customer satisfaction survey, then, is to understand the optimum use of resources to increase overall customer satisfaction and drive the “somewhat satisfied” customers into the “totally satisfied” corral. To do so has a direct impact upon revenue. In Miyago’s case we identified that for each $1 a “totally satisfied” customer spent, a “somewhat satisfied” customer spent 38 cents, a “somewhat dissatisfied” customer spent 8 cents, and, a “totally dissatisfied” customer cost Miyago $3.10. Thus, moving his customers up the scale became Matoba’s top priority.

Several years ago we used to ask survey respondents to rank “importance” with various attributes. Invariably, it was noted that every attribute was “extremely important.” This method is ineffective because context and recent experiences influence the responses to importance ratings. To illustrate that point, ask a recent automobile buyer to rate the importance of service in relation to the product’s various performance and style attributes. In all probability, service will rank fairly low. Ask the same question after the car breaks down, and the answer will be very different. In specific situations, people also do not always act in accordance with what they claim is important to them in the abstract. We gave up asking.

Instead, we began to play around with a technology called neural network analysis. Neural networks are a type of system that simulates many of the abilities of the human brain. By using the same architecture as your brain, but on a smaller scale, artificial neural systems work much the same way your brain does. Human abilities, such as noticing patterns and trends, finding hidden relationships in data, and learning by studying the past, can now be performed by a computer. Neural networks combine our ability to analyze and learn with the computer’s ability to process a great deal of data quickly, easily, and dispassionately. Just like people, neural networks learn from experience, not from programming. They are fast, tolerant of imperfect data, and do not need formulas or rules.

We developed a technique for employing this technology in a way that enables our clients to identify, prioritize and quantify the relative impact that improving various attributes (e.g., sales, service, billing, pricing, delivery, etc.) will have on overall satisfaction. This technology can give a fundamental but otherwise unavailable level of understanding of the customers’ thinking and priorities. We moved from the “what” customers think to the “why.”

The neural network analysis identified for Miyago the key drivers for its “totally satisfied,” “somewhat satisfied,” and “dissatisfied” customers. Thus, management and the remediation teams could evaluate the drivers for the customer segments, note similarities and differences, and base future policies and improvement spending upon only those attributes that have a direct impact upon improving overall satisfaction.

Considerable impact

In 1997 the neural network identified three attributes that had considerable impact upon the satisfaction levels of Miyago’s customers. One of the categories, “commitment from Miyago management,” however, had an impact level that was more powerful than the others. The other two key attributes were “quality of services” and “the ease of doing business with Miyago.”

When Matoba saw that his customers did not feel that Miyago’s management was committed to their success, he took immediate action. He spent the next three months visiting each of the 180 customers to discuss the results of the survey and to personally commit his team’s dedication to Miyago’s customers. By the way, in 1997, the percentage of Miyago’s “totally satisfied” customers was 11 percent.

We measured this same customer group one year later and identified that after Matoba’s tour, the “totally satisfied” group increased to 48 percent. Using Miyago’s 1997 revenue as a baseline, the reported income for 1998 increased by 9 percent. There was only one key driver identified in 1998: the ease of doing business with Miyago. However, it was about half as significant as in the prior year. Matoba persisted with his customer visits.

The survey results in 1999 showed substantial increases. The “totally satisfied” segment rose to 67 percent and revenue increased by 18 percent. The key drivers of satisfaction reverted to quality of services and a new attribute, billing. The fact that quality became an issue again after two years is not surprising. After the first year, Miyago paid close attention to its quality program, its customers became used to a new level of service and eventually expected more. Billing became an easily correctable issue.

In 2000 total satisfaction hit 78 percent, a number rarely ever achieved and one that places Miyago in a world-class category. Not unexpectedly, revenues increased by 5 percent.

By paying attention to his customers’ complaints, visiting each one to discuss the survey results, committing Miyago’s employees to his customers’ success, and following up with the results, Matoba increased his company’s revenues by 35 percent in four years. Stay tuned for 2001!