Exploring Speed of Answer Goals

By Maggie Klenke

The workforce management (WFM) process is designed to get the right number of qualified staff in place to handle an accurately forecasted workload with quality while meeting the budget and the speed of service goal. In fact, the mathematics that underpin the whole WFM process and staffing models are based on a goal to answer the contacts within a predefined time limit. Let’s explore these goals for time to answer to understand how they are similar but different.

There are typically three primary goals used in contact centers to measure the speed of answer. They are:

  • Average Speed of Answer (ASA)
  • Service Level (sometimes referred to as Telephone Service Factor)
  • Abandon Rate

Average Speed of Answer

Average Speed of Answer calculates the average wait of all callers, both those who had no wait and were answered immediately by an idle agent, and those callers who waited in the queue. For example, let’s assume that 100 customers call in during a 15-minute period and 25 of these are answered immediately while the other 75 go into the queue. The 75 callers in the queue wait varying lengths of time from 1 second to several minutes, but the overall average for them in 64 seconds. By adding the 75 callers at 64 seconds average and 25 callers at 0 seconds of wait, the ASA is calculated at ((75 X 64) + (25 X 0))/ 100 calls or 48 seconds average overall.

Now let’s assume that we set a speed of answer goal to achieve an ASA of 48 seconds. Remember this is an average that includes all the 0 second waits for callers answered immediately. Those callers that go into the queue will wait varying lengths of time and some of them will be well over the average of 64 seconds – perhaps even waiting 3 or 4 minutes while others may be just a few seconds. The calculation does not identify how many of the calls were answered in 48 seconds, only the average wait of all the callers.

The advantage of an ASA measurement is that it counts every second of wait time for all the callers. The drawback is that it provides little useful data on the length of the longest delays. As with any type of averaging, it tends to conceal the highs and lows.

Service Level

Service Level measures the percentage of calls that are answered within a set time limit. There are two numbers to be set in this formula. One is the number of seconds/minutes in the goal and the other is the percentage of calls that will be answered within that time frame. For example, a goal might be to answer 85 percent of the calls within 30 seconds. If the center met that goal on 100 calls, 85 of the callers would have been answered in anything between 0 and 30 seconds. The remaining 15 callers would have waited at least 31 seconds and could have waited several minutes as the calculation does not consider the length of the wait – only that it exceeded 30 seconds. It is essentially a “pass/fail” type of metric.

The advantage of service level is that it provides two elements to work with in establishing the goal. By setting the percentage at levels of 80 percent or more, the risk of the outliers waiting a long time are reduced. However, if there are very long delays for some callers, the length of the wait is ignored – it is simply counted as a “fail” just the same as a call that waited one second over the goal. There is also the other end of the problem when the percentage is set at 90 percent or more. At that rate, there is very little margin for error and a significant amount of idle time must be assumed for agents to be able to answer the calls within the goal.

Another challenge is understanding the impact of changing the formula. By raising the percentage but lengthening the accepted delay, the results might very well be unchanged. It is also common in centers for the percentage to be the only part of the formula reported on the ACD statistics. When asked, some personnel may not know what the goal in seconds is as it is rarely on the reports.

Abandon Rate

Abandon rate measures the number or percentage of callers who entered the agent queue but hung up before being answered. These might be people who heard the recorded announcement and realized they have dialed the wrong number or don’t have the needed information at hand. Some may realize there is a delay and choose to redial later, or have some personal interruption such as a doorbell or crying child. Others hold on as long as they are willing and hang up when their tolerance for waiting has been exceeded.

The reasons that callers hang up can vary significantly along with the tolerance that could be reasonably predicted. For example, a customer calling to rent a car who rarely does so might give up if not answered quickly by the first company called. They may think of it as a commodity and simply dial the next company. However, if that caller is a “priority” customer for that rental car company and receives certain perks, or a customer whose employer dictates which rental company must be used, the tolerance level is likely to be a bit longer. In another example, a customer calling to purchase auto insurance may have little tolerance to wait for a sales agent, but if the customer has already purchased the policy and now has a claim, the willingness to wait for a claims agent is probably higher since he is a “captive audience.”

The advantage of measuring abandon rate is it potentially looks at the impact on the business of lost revenue/customers. The disadvantage is that it is highly variable based on a variety of uncontrollable factors and human behaviors and cannot be used with any mathematical precision to determine how many agents are needed to serve a specific workload.

Other Challenges

When choosing a metric that will serve the organization’s needs best, it is important to consider a few other factors that come into the formula. For example, how will abandoned calls be factored into the calculation of ASA or Service Level? Some count all abandoned calls and all the seconds they waited in the formula. Others totally ignore abandoned calls and only calculate based on “answered calls.” Still others ignore calls that abandon in an arbitrarily set number of seconds in an effort to weed out wrong number callers or those who simply will not wait for anyone.

Another variation comes from when the measurement begins. Some start the clock when the call enters the agent queue and includes the recorded announcement time. Others start the clock when the announcement has completed, especially where the recording plays on every call for compliance reasons. Some may set an arbitrary starting point other than one of these two choices.

When comparing results between systems in a center (such as WFM and ACD), be sure the calculations match to avoid confusion. It is also important to ensure that different sites within the same company have all systems synchronized so that some sites are not being held to a higher/lower standard than the others. The systems come from the vendors with a preset formula, but they can often be changed to the method you choose. Work with your IT department and vendor to achieve the best calculations for your business.

Summary

There is no one right metric for all situations and there is no “industry standard.” Some use ASA and it works well for them. Others choose service level and find it is a good fit. It is not a good idea to use both metrics on the same queue as they are not mathematically synchronized and the center can achieve one goal but not the other as the volume and AHT vary through the hours and days.

It is also good to think about the goal and whether it should be reconsidered occasionally. Customer needs and desires change, budgets shift, automation takes over the simple tasks leaving the tougher calls for the agents, etc. Think about each call type and work type and consider whether they should have different goals based on value to the company, competitive pressure, changes in caller tolerance, or other business factors. Take a hard look at what is important to the customers. It is possible that too much emphasis is placed on speed and not enough on first call resolution. Many studies suggest that callers are willing to wait a bit longer if they are confident they will reach a knowledgeable agent who can resolve the issue in an efficient and effective manner. And it helps if the agent is not overwhelmed and can put a smile in the transaction as well. Maybe some of the money spent on faster answer could be better used to provide more training and development time as well as an occasional breath between calls.

Maggie Klenke is one of the founders of The Call Center School (now retired) and an active industry consultant, assisting contact center clients in development of strategic and tactical plans, technology applications, forecasting and scheduling optimization, service level analysis, and overall management issues. Maggie teaches seminars on a wide variety of call center topics and is a popular speaker at industry conferences in the US and abroad.

She is an honors graduate of Loretto Heights College (now Regis University). She is also one of the first to receive the CIAC industry certification as a Call Center Management Consultant. She also serves as an examiner for the Tennessee Center for Performance Excellence, a statewide program of the national Malcolm Baldrige awards. She may be contacted at maggie.klenke@mindspring.com or 615-651-3324.