A Non-Mathematical Approach To Basic Forecasting

The non linearity of queuing is known, from this knowledge the first leap of faith is to believe that improvement in queuing performance at the call center can have a positive impact on customer satisfaction. While this does not seem like a very big leap, there is little experimental evidence to prove that it is true. There is a great deal of statistical evidence that will show that improvements in queuing performance correlate at very high levels with increased values of customer satisfaction. The purpose of making this leap is to prepare for the next leap which is to believe that the improvement in queuing performance at the call center with the implied positive impact on customer satisfaction also leads to an improvement in customer retention which then will be translated into to improved financial performance.

Call Center Queuing Performance Unpacked

Before I continue, let me describe queuing performance. The five parameters that are generally used to describe queuing systems are:

  1. The average time in the queue
  2. Average time in the system (waiting time plus talk time)
  3. Average number in the queue
  4. Average number in the system (those waiting plus those being served
  5. The percent utilization of the servers (percent of time they are busy

I believe many call center managers use the average time in the queue, the average handle time and the percent utilization of the servers as the three measurements to manage their operation. In this case, I am using the word server to provide a generic description of the person at the phone/terminal who is "serving" the customer (other terms such as agent, customer rep would be equivalent). While I may utilize any one of the five performance parameters of queuing performance to make the case for the "leap of faith," the focus will be on the three most often used by call center managers.

I stated above that there is very little experimental evidence that I am aware of that shows that a customer chooses to stay or leave a specific vendor as a direct result of the time spent waiting on hold (or lack of time waiting on hold) at the call center.

Call Center Wait Time and Customer Attrition

There is some evidence that the wait time at the call center was the final action that caused a customer to leave, but no evidence that it was the only reason. The leap of faith is based on statistical evidence that indicates a statistical relationship between customer satisfaction and queue (waiting) time. The key point to remember is that statistical relationships DO NOT imply cause and effect relationships. The example I give to my California students of the difference between statistical relationships and cause and effect relationships is the very high correlation between the number of ice cream cones bought at the beach and the number of drownings.

As the number of ice cream cones purchased at the beach increases, the number of drownings also increases. The question I raise to my students is, "Would drownings decrease if they stopped selling ice cream cones?"

Florida Power and Light Company Case Study

The fact is that sometimes a statistical relationship is an indicator of a cause¬-and effect relationship. A case where there is statistical evidence of the relationship is the Florida Power and Light Company that actually set out to examine the relationship between waiting time and customer satisfaction. They performed a study to examine how customer wait time affects the level of customer satisfaction. As a secondary objective of their study, they wanted to know what could be done to increase satisfaction when long waiting times do occur. The first finding of their survey was that customers' perceptions of wait times are not consistent with actual wait times. The comparison of the perceived versus actual is shown in the table below.

Actual Wait Time (seconds)

Perceived Time Less Than 30 30-60 Greater Than 60
Less than 30 0.31 0.14 0.12
30-60 0.41 0.38 0.5
Greater Than 60 0.28 0.49 0.38

Thus, only 31 percent of the callers perceived the wait time to be less than 30 seconds when it was less than 30 seconds. In fact, 28 percent of the callers perceived the time to be greater than 60 seconds when the actual time was less than 30 seconds. Equally interesting to note is that 17 percent of the callers perceived the wait time to be less than 30 seconds when the actual wait time was greater than 60 seconds. The conclusion is that our built in time clocks are not very accurate.

Linking Customer Satisfaction to Hold Time

The next step was to link customer satisfaction rating to the time on hold. While other factors may be involved, the results provided the statistical relationship we are looking for in this article. They used a one to five scale to measure customer satisfaction where one signified very satisfied and five signified very dissatisfied. The average satisfaction with the phone contact experience seems to decrease as the time on hold increases as shown in the following table.

Mean Rating Time On Hold


>30 seconds

1.57 30-60 seconds
1.67 1-2 minutes
2.14* 2-3 minutes

* indicates that 2.14 is statistically significantly different than the others at the alpha = 0.05 level.

In fact, there appears to be a negative trend in customer satisfaction with phone contact experience as the time on hold increases. While this may seem obvious, it is always of great value to have your "gut" belief verified with actual data. Based on this test that has been documented, there is some evidence to make the first leap of faith that performance of the call center does have a direct impact on the satisfaction of the phone contact experience.

However, the extension of the evidence to overall customer satisfaction has not been documented to my knowledge. In fact, Florida Light and Power Company may have such evidence. For now, since the first step of evidence has been given, the leap of faith seems a little shorter.


As a sidebar to the main thrust of this article, Florida Light and Power Company found that by telling the customers the expected length of the wait, the customers were voluntarily willing to wait longer without a decline in customer satisfaction.

Their research indicated the following statistics:

  • 10 percent were willing to wait 1/2 minute longer
  • 17 percent were willing to wait 1/2 to I minute longer
  • 23 percent were willing to wait I to 2 minutes longer
  • 22 percent were willing to wait 2 to 3 minutes longer
  • 28 percent were willing to wait 3+ minutes longer

The tabulated research indicated that customers would be willing to wait an average of 105 seconds longer when they were made aware of the actual expected waiting time at the beginning of the call. The technology for providing this information to customers is available and should be considered by call center managers.

First published on Call Center IQ