Memphis Light, Gas and Water (MLGW) engineers took Bill Gates to heart when he stated, “Your most unhappy customers are your greatest source of learning.” Pursuing that line of thinking, MLGW decided to investigate disgruntled customers to better understand their tolerance to power interruptions. This insight changed the way MLGW viewed customer reliability. Appreciating this new outlook requires examining the previous methods of measuring reliability.
MLGW (Memphis, Tennessee, U.S.) is a municipally owned distributor serving more than 400,000 customers in Memphis and Shelby County, Tennessee, U.S. Like most utilities, reliability has always been a high priority. Normally in the past, MLGW charted its system reliability indices and used them to measure and prioritize its distribution projects. These methods have been helpful, and MLGW has charted improvements in SAIFI, SAIDI and CAIDI over the years. However, it became apparent that these standard indices did not represent the individual customer's point of view. Customer experiences don't necessarily correlate to system indices, and averages can hide a multitude of sins. For example, a 20% improvement in SAIFI may go unnoticed by the customer base. Conversely, a minute increase in SAIDI may indicate a large number of dissatisfied customers.
The standard indices are invaluable in measuring system performance. They work well in benchmarking with other utilities and are also useful for internal comparisons, especially in measuring restoration efficiency and system improvements. They are vital in tracking outage trends and system reliability goals. However, these numbers do not reflect individual customer perspectives. Knowing the average number of interruptions per customer does not tell us anything about the customers who experienced the most outages.
Beyond the Averages
Recently, the Institute of Electrical and Electronic Engineers (IEEE) introduced a new indice that specifically looks at customer experiences through a relative frequency approach. This new indice, Customers Experiencing Multiple Interruptions (CEMI), measures the percent of overall customers that have experienced more than a specific number of interruptions. For the first time, an indice looks at the full spectrum of customer interruptions rather than system averages. CEMIs for various numbers of events are shown in Fig. 1. The results were surprising when first published at MLGW. It was difficult to accept that with a system SAIFI of less than 1.5, greater than 20,000 customers had experienced more than six interruptions. The difference between service levels to various customers was eye opening, but it also identified how to improve reliability for specific groups and how to be more customer-oriented.
Consequently, MLGW's executive staff charged engineering with exploring reliability expectations of individual customers. MLGW wanted to know how customers responded to various levels of service with the hope that the information could be used to set reliability goals and to tie improvements in reliability to a real improvement in customer satisfaction.
Like many utilities, MLGW surveys its customers regarding reliability and other customer service issues. Unfortunately, this process can be less than informative when exploring the customer psyche on acceptable performance levels. In many cases, customers are “pushed” by the wording of the question and do not have an expressible conviction of their own. Unless reliability is an immediate concern, customers tend not to dwell on the issue. Their memory of specific events tends to be short and answers may not be an accurate reflection of their expectations. Typically, a high percentage of customers are satisfied with their overall reliability, but it is also clear that external issues such as the customer's latest utility bill, personal experiences with utility personnel, and other non-related matters drive opinions and skew results. It is amazing how reliability satisfaction nose-dives immediately after a rate increase. In any event, efforts to determine customers' tolerances by conducting surveys and correlating them to actual customer reliabilities have proven fruitless.
The Root Cause
After being enlightened about the value of surveys, MLGW decided to take the lead from Bill Gates and learn from its most unhappy customers. It began by mining about 3000 complaints received by the reliability department during the previous three years. The object was to determine why the customers made the effort to call and complain. This feedback provided an unbiased gage of customer attitudes at various service levels. Registering a complaint with the reliability department is indicative of a motivated customer who was not satisfied with the normal customer service channels. Some are simply seeking a fix to an ongoing reliability problem but many are quite irate.
Normally, complaints are investigated by researching the customer's reliability history. This is the first step in addressing the root cause of the complaint, and in most cases, this history paints a clear picture of why the customer complained. MLGW decided to employ the outage history to quantify the catalysts that engendered the customer complaints. MLGW determined 600 complaints were specifically related to the frequency of customer interruptions. This information was broken down by the number of interruptions and the interval in which they occurred. Next, the ratio of customers' complaints was compared to the total number of customers experiencing a particular interruption frequency. The numbers were then normalized by the number of complaints per 10,000 customers (Fig. 4).
Why Customers Complain
Several observations can be made from Fig. 4. Historically, we know that customers may complain about a single event because of needs or the poor timing of the interruption. For example, customers who work at home are more sensitive to interruptions. This is also true for customers who are elderly and infirm. Outages that occur at inopportune times also may cause complaints. If a customer misses a favorite TV show or if the outage interrupts a critical task such as cooking an important meal, they may respond to a single interruption with a complaint. However, most customers are more resilient.
Increasing the frequency of interruptions heightens customer sensitivity but the interval is also important. It can be noted that three interruptions during a six-month interval doubles the number of complaints versus three interruptions during a year. All intervals demonstrate a natural threshold when exceeding three interruptions. This drives complaints to increase more than 400%. Clearly, reducing the number of customers experiencing more than three interruptions in a year will significantly increase overall customer satisfaction.
In analyzing the customer complaints, MLGW realized that many calls were not in response to frequency of outages but rather outage durations. This is the second key parameter needed to describe customer outages. MLGW used more than 250 complaints to quantify customer reactions to different outage durations. The utility analyzed the data in similar fashion to the frequency data to explore customer thresholds for complaints (Fig. 5).
The various bars on Fig. 5 can be related to actual customer inconvenience and hardship. A short duration event may mildly disrupt the household schedule and postpone some activities. Lighting and entertainment are the first functions affected. As the outage wears on, inconvenience grows with a loss of comfort and the delay or alteration of meal plans. During the next two hours the loss of HVAC becomes particularly unbearable in inclement weather. From six to 10 hours the disruption becomes more severe as meals and chores are severely off schedule. Security may be a concern. When outages last more than 10 hours, food may spoil. Customers may be forced to seek accommodations with relatives or at a motel.
Notice from Fig. 5 that doubling outage durations results in a tenfold increase in complaints. During most years, less than 1% of MLGW's customer base experience outages of more than 10 hours in duration but they make up an appreciable number of customer reliability complaints. Since 10-hour outages are obviously unacceptable and troublesome to customers, special efforts are needed to address prevention or expedite restoration.
Proper communications often can soften customer perception of long duration events. For example, accurate estimates of restoration are important to customers along with sharing information on the nature and cause of any extended outages.
Setting Response Goals
MLGW's investigation into customer complaints has produced meaningful insights into customer tolerances for power interruptions. MLGW has moved quickly to translate these into corporate goals and explore strategies for achieving them. The long-term goal includes eliminating all interruptions more than 10 hours in duration and any situations where customers are experiencing more than three interruptions in a year. Initial investigations show these long durations are associated with several outage types, including URD cable faults and cars hitting poles during the night. Fine-tuning the cable replacement program will prevent some of these events. Others may be addressed by improving off-hour response times. Sometimes, severe storms are the culprits.
When examining outage frequency, the first line of defense will be to identify and prevent reoccurring outage events. Hastening follow-up inspections and preventative maintenance in trouble spots will reduce some recurrent interruptions. Naturally, significant improvements in these areas will translate into improvements in system indices as well. However, this approach offers the ability to target outage events that best improve customer satisfaction.
An obvious question is “How universal are these thresholds for residential customers?” MLGW's customers are not a homogeneous group, and several differences were noted when reviewing the reliability complaints. For example, rural customers tolerate outages better than urban ones. Customers who work at home or who are aged or infirm have special needs and are more sensitive to interruptions. The results discussed are valid for MLGW's service territory, whereas other customer bases may be more or less sensitive. Regardless, MLGW is confident this methodology for quantifying tolerances provides a better picture of different customer perspectives at various service levels than using customer surveys or averaged indices, and it is a better way to target reliability improvements.
Don Roberts serves in a lead engineering position in the reliability and power quality area at the Memphis Light, Gas and Water Division. He worked in the electric distribution engineering, street light engineering and energy management departments prior to joining the reliability and power quality in 1989. He holds a BSEE degree from the University of Memphis.
Xavier Davis joined Memphis Light, Gas and Water's reliability and power quality department three years ago. He works to solve customer reliability problems and also designs retrofit work on the electric distribution system. He earned a BSME degree from the University of Mississippi in 2000 and is currently pursuing an MBA degree from the University of Memphis.