3.6 Service level-a core value

3.6 Service level—a core value

At the heart of effective incoming call center management is the principle of service level. A service level objective can be used to determine the resources required and the effectiveness of the center in its impact on the corporate business goals. Here are some of the questions that can be answered by establishing and monitoring a specified level of service:

• How accessible is the call center?

• How much staff is required?

• How does the center compare to the competition?

• Can the center handle the response to marketing campaigns?

• How busy will the CSRs be?

• What will the costs be?

Defining a service level

Service level is often referred to by various terms. In some call centers, it is the telephone service factor, or TSF. Others refer to it as grade of service (GOS), although this may be confused with the term for the degree of blocking on a group of trunks. Service level is also referred to as accessibility or service standard. Typically, the term service level is used to refer specifically to transactions that must be handled on arrival at the call center. Response time, often called speed of reply, may even be called service level as well. To avoid confusion, response time will be used in a specific sense in this book, to describe the level of service assigned to transactions that can be handled at a later time and do not need to be handled "on arrival. "

The most widely-accepted definition of service level is based on the percentage of calls answered in a given time frame, for example, 90 percent of calls answered in 20 seconds. Some managers define service level as a percentage only or as an abandonment rate. Others refer to the percentage of the time the service level objective is met, whatever that objective may be. And there are those who define service level as "average speed of answer" or longest delayed call.

The various interpretations and other definitions of service level often lead to misunderstandings and mismanagement. By its nature, service level should be defined as a specific percentage of all calls answered in a specific time frame, as previously noted. Planning should be based on achieving this target. Choosing an appropriate service level objective is one of the first steps a call center manager should take to ensure effective planning and management of the operation and to establish budgets.

Establishing a service level helps to link resources to results and measures the degree to which customers are being transferred and handled by a CSR. Service level is a tested and proven criterion in call centers worldwide for transactions that must be handled when they arrive—most commonly inbound phone calls. However, as customer contact methods change, new multimedia services—video calls and calls integrated with the World Wide Web—may also become part of the service level criterion. Because of its universal acceptance as a primary call center criterion, service level will remain an important objective to the next generation of call centers.

Other response categories

In addition to the "immediate response" category, most incoming call centers are required to handle transactions that belong in a second category, those that don't have to be handled at the time they arrive. Some examples of these transactions are

• Postal correspondence (snail mail)

• E-mail

• Faxes

• Voice mail

• Video mail

These transactions allow a larger window of time for the call center to respond. It is as important, however, to establish specific response time objectives for these interactions as it is for the first category of transactions. All categories of transactions can contribute to meeting the service objectives of the call center if appropriate priorities are established.

Other response criteria

Average speed of answer (ASA), another often-used response criterion, is related to service level because it is derived from the same set of data. However, ASA is often misinterpreted. In any set of data, it is generally assumed that the average lies somewhere in the middle or that "average" represents typical experience. This is not true for call center purposes. Although mathematically correct, the average does not represent the experience of individual callers. In a call center, most callers get connected to a CSR much quicker than the average, but some wait far beyond the average. For example, with an average speed of answer of 15 seconds, about 70 percent of callers get answered immediately, but a small percentage of callers will wait three or four minutes in the calling queue. Although ASA is useful in calculating some call center requirements—for example, in calculating trunk load—service level is a more reliable and more telling measure of a caller's experience.

Abandoned calls

Considering call abandonment rates alone as a measure of whether staffing levels are appropriate can be quite misleading. A high abandonment rate is probably a symptom of staff problems. But a low abandonment rate doesn't necessarily mean the center is optimally staffed. If abandonment rates are unacceptable, call center managers need to evaluate the situation to determine what is wrong. It is most likely that the evaluation will reveal a too low service level. When service level is being achieved, abandonment rates tend to take care of themselves.

Unanswered calls

One important consideration about service level is what happens to calls that don't get answered in the specified service-level time frame? Most Erlang C and computer simulation software programs can calculate the answers to this question and others. For example, for a service level of 80 percent answered in 20 seconds, experience indicates that about 30 percent of callers end up in the queue, that the longest wait will be around three minutes, and that the average speed of answer will range from 10 to 15 seconds. This example points up the obvious fact that different callers have different experiences with call centers, even if they are part of the same set of data measured by service level, ASA, and other measurements. The reason for this is "random call arrival," a reality of call center operation and a factor that needs to be considered when deciding how to measure quality of service. Service level is the single best measure of quality, largely because it enables the center to determine what happens to different callers.

Inbound transactions—priority levels

There are two major categories of inbound transactions, with two priority levels, that a call center needs to handle:

• Those that must be handled when they arrive (e.g., inbound calls)—Performance objective: Service Level

• Those than can be handled at a later time (e.g., correspondence)—Performance objective: Response Time

The rationale for a service level

Establishing a service level based on calls answered in a specified time as opposed to percentage of calls answered or percentage of calls abandoned or even average speed of answer provides a clear-cut indication of a caller's experience when contacting the call center. Service level is the most stable measurement of the inbound call-in queue. The importance of a defined service level can be summed up by examining the effect on customers and call center operations as it relates to the following factors:

• Agent burnout and errors

• Levels of lost calls

• Customer goodwill

• Links between resources and results

• Focus on planning activities

Applying service-level metrics

It is important that service level be interpreted in the context of call blockage, that is, calls not getting through. Any time some portion of callers is getting busy signals, no matter whether generated by the system resulting from a limited number of staff and lines during a busy period, service level reports only report on the calls that are getting through. Reports based on service level and average speed of answer can be configured to look very impressive simply by limiting the number of calls allowed to get through.

Service level is obviously a time-dependent parameter, and daily service level reports may often conceal important information. Service level may be down in the morning; however, if staff levels improve and every call in the afternoon is handled immediately, the daily report will look very good against service-level objectives. On the other hand, the level of service from a callers' perspective is a different story. It is not difficult for managers accountable for daily reports and meeting service-level objectives to "fudge" these reports or call center activity to make the situation look better than it really is. If the morning service level was low, they may keep CSRs on the phones through the afternoon when the call load drops, just to make reports look better. This is a waste of valuable time and resources and provides inconsistent service to customers.

Consider this: If daily reports are potentially misleading, the longer the time frame between reports, the more misleading they can be. Therefore, monthly averages for service level are virtually meaningless, because they don't reflect the day-by-day, half-hour-by-half-hour realities. Even so, monthly reports are a popular way to summarize activity for senior management, although there are more meaningful methods of reporting call center activity.

ACDs and service level

There are a number of alternative methods to calculate service level using ACDs. Following are some of the most common calculations used, although some ACDs allow users to specify other definitions of service level using a variety of other call center parameters:

1. Calls answered in Y seconds divided by calls answered:

   

   This is a very simple but incomplete measure of service level. It is not recommended for a definitive analysis because it considers only answered calls. It is an incomplete recognition of call activity and, therefore, not a good measure of service level. For example, call abandonment is entirely ignored in this calculation.

2. Calls answered + calls abandoned in Y seconds divided by (total calls answered + total calls abandoned):

   

   For most situations, this alternative is preferable because the calculations include all traffic received by the ACD; therefore, it provides a complete picture of call center activity. The combination of total calls answered (TCA) plus total calls abandoned (TCB) is often referred to as total calls offered.

3. Calls answered in Y seconds divided by the sum of (calls answered + calls abandoned):

   

   This alternative tends to be the least popular among call center managers because calls that enter the queue but then fall into the abandoned category drive service level down. It is appropriate in situations where calls enter a queue after callers receive a delay announcement. It is not recommended in situations where callers enter a queue before they receive the delay announcement.

4. Calls answered before Y seconds divided by (calls answered + calls abandoned) after Y seconds:

   

   With this calculation, abandoned calls only impact service level if they happen after the specified Y seconds. This measurement provides a way to avoid "penalizing" the service level due to callers who abandon quickly, without ignoring abandoned calls altogether.

Turning service level into quality of service

As many call center managers have discovered, it is important not to confuse service level with quality of service. It is possible to regularly and continuously meet service-level objectives and at the same time create extra work, have low productivity, and provide a poor quality of service to customers. A narrow focus on service level will not necessarily provide quality. CSRs can have an excellent service level but still make some or all of the following mistakes that may not be reflected in service level because they are content related and not traffic related:

• Relay the wrong information to callers

• Make callers upset

• Fail to accomplish call center objectives

• Record incorrect information

• Miss opportunities to capture valuable feedback

Service level—a limited measure

Service level is a limited measure of overall call center performance because it indicates only that "not too many callers had to wait longer than a certain number of seconds before reaching a CSR." Unfortunately, service level measurement devices such as those provided in an ACD cannot measure whether callers and the organization achieved their mutual goals. It is important not to play the "numbers" game and to keep the primary objective in mind.

Optimizing service level with quality is an ongoing consideration in every call center. If service level is the only characteristic that is being measured and managed there can be too much emphasis on it. A good service level is an enabler for other important objectives—calls are coming in and being answered so that the organization and callers can achieve their mutual goals: getting information on product or services, selling products, or providing other customer-oriented information.

On the other hand, a poor service level reduces call center productivity. As service deteriorates, more and more callers are likely to complain when calls are finally answered. CSRs will spend valuable time apologizing to callers and will not be able to answer as many calls as the service level deteriorates. Costs will increase and revenues will likely be affected negatively. Other negative situations will also develop. Calls will get longer because CSRs will eventually pace themselves differently. And they will take breaks when they are on calls if they are so busy they cannot take breaks between calls, because the "in-between" time no longer exists. In the longer term, as service level starts to slip and continues to decline, CSRs often try to clear up the queue. If they are not able to do this, they eventually adopt work habits that are detrimental to the call center. Call handling time goes up and employee moral is affected and turnover and burnout increase, along with recruitment and training costs. This is obviously a disastrous spiral for a call center environment.

The impact of a poor service level will ultimately be felt in the quality of service offered. When CSRs are overworked due to constant congestion in the queue, they often become lazy and can also become less "customer-friendly." Callers are telling them about the difficulties they had getting through to the center, and CSRs make more mistakes under these conditions. These mistakes contribute to repeat calls, unnecessary service calls, escalation of calls, and complaints to higher management, callbacks, and so on—all of which drive service level down further, again illustrating that a poor service level is the beginning of a vicious cycle.

Based on this discussion, it is apparent that quality should never be considered as an attribute that is opposite to service level—the two must go together.

Choosing a service-level objective

The number of staff needed to handle transactions and the schedules should flow from the service-level objective. (see Figure 3.6) Imagine that the call center receives 50 calls that last an average of three minutes in a half-hour period. If there are only two CSRs answering calls, the delay time for most callers will be long, and abandonment rates will be high. Adding CSRs will reduce delay times. An acceptable rule of thumb is reduce the queue to an acceptable level for both the call center and the callers. The number of CSRs required to provide this degree of service then becomes the service-level target and defines the correct level of resources to meet that target.

Figure 3.6: Customer inputs to a multimedia call/ contact center.

There are no generally accepted industry standards for service level, but there are several factors, mostly subjective, that affect service level:

• Value of the call

• Fully loaded labor costs

• Trunk costs

• Caller tolerances

• An organization's desire to differentiate products or services by level of service provided in the call center

An industry standard would have to be based on all call centers placing the same values on these factors, which would be difficult, if not impossible, to achieve. However, some regulated industries have defined service levels. For example, service levels are defined by regulation for cable TV companies in the United States and for telecom call centers in some countries. These levels of service may be regulated through a service-level agreement (SLA). In Canada, Bell Canada service levels are regulated by the CRTC (Canadian Radio and Telecommunications Commission).

It is reasonable to conclude from the discussion here that the correct service level for a call center, apart from legal regulations, is the one that meets the following conditions:

• Minimizes expenses

• Keeps abandonment to an acceptable level

• Maximizes revenue

• Meets caller needs and expectations

• Minimizes agent burnout and errors

• Is agreed upon and supported by senior management

Guidelines for determining service-level objectives

There are a number of methods for determining service-level objectives, but the following four approaches have been distilled from the collective experience of call center managers:

• Minimize abandonment

• Take the middle of the road—follow the crowd

• Relate to competition

• Conduct a customer survey

Each approach requires some subjectivity and judgment on the part of management personnel.

Minimize abandonment

No single service level would satisfy all situations affecting how long callers will wait for a CSR to respond. A number of factors influence caller tolerance, including

• How motivated callers are to reach the call center

• What substitutes for a telephone call are available

• The competition's service level

• The caller's expectations based on past experiences

• How much time the caller has

• The conditions at the locations callers are calling from

• Who is paying for the call

The first approach to choosing a service-level objective essentially involves asking the question, How low can response times go without losing callers? This assumes that a higher level of service means lower abandonment and vice versa; that is, as long as callers don't abandon, service is acceptable. But that is not always the case—abandonment is not static and will fluctuate as the seven factors of caller tolerance change. Abandonment is difficult to forecast, and choosing a service level around abandonment is one of the least desirable ways to establish a service level.

Take the middle of the road—Follow the crowd

The "middle-of-the-road" method defines service level as percentage of calls answered in so many seconds, for example, 80 percent answered in 20 seconds. The 80/20 objective has been cited in some ACD manuals as an "industry standard." However, it has never been recognized as such, even though many early call centers used it. The 80/20 objective is still fairly common because for many call centers it is a reasonable balance between callers' expectations and the practicality of having enough staff to meet the objective.

Benchmarking the competition

Another popular method for choosing a service level is to benchmark competitors or other similar organizations and then use this information as a starting point. This can be done informally by simply asking for the information or by conducting a formal benchmarking study. Whatever the approach, keep in mind that the results reported and those actually achieved may not reflect the actual situation. Human nature tends to "color" the truth on the positive side, especially when the competition may have access to the responses! Cases have been documented where companies with the same service level objectives—80 percent of calls answered in 30 seconds—achieved very different results.

A more formal way to determine the potential impact of abandonment on overall costs is incremental revenue analysis, a variation of the benchmarking approach. Traditionally, this approach has been used in revenue-generating environments, for example, airline or railway reservation centers and catalog companies, where calls have a measurable value. It is more difficult to use in customer service centers and help desk environments, where the value of calls can only be estimated. In incremental revenue analysis, a cost is attached to abandoned calls and assumptions made as to how many calls would be lost at various service levels. CSRs and trunks are added as long as they produce positive incrementals, either marginal/additional revenue or value, after paying for the initial costs. As long as the assumptions are clearly understood and communicated to management, this approach can be very useful when combined with other approaches.

Customer survey

A fourth method for choosing service level is to conduct a customer survey. This involves analyzing caller tolerance.

It is always a good idea to know what callers expect, but random call arrival means that different callers have different experiences with a call center. Even for a modest service level such as 80 percent answered in 60 seconds, over half the callers will get an immediate answer. Some may still be in the queue for three to five minutes (assuming no overflow or other contingency). This significant range of response times means that many callers in a set would claim that the service level was great, while others would describe it as totally unsatisfactory!

There are variations in customer survey methodology. Some managers take samples of individual callers and then compare the responses to the actual wait times for their calls. Others conduct general customer surveys. These samples indicate that waits of up to 60 to 90 seconds are acceptable to a fair percentage of the callers surveyed.