OPS - A Case Study
Using OPS to Decrease Call Center Labor
OPS was used to model callers of a call center during the various processes accomplished on the call. After a review of the current process, a future process was identified that decreased the callers service time and decreased the number of operators that were needed to keep up with the call volume.
The call center was processing support calls for the company. The caller would call during normal business hours. The time between calls used in the model was obtained through 12 observations of time between calls that occurred during the busiest hour of the day. Upon calling they would be placed in a queue to wait for an operator. 10 Operators staffed the call center during normal business hours. When an operator was available the caller would be asked for their account number to verify the customer status. 12 observations of how long the account verification took were used as input into the model. Once the account number was verified the operator would then answer the questions of the caller. 12 observations of how long it took to answer the question were used as input into the model. Once the answer was determined and communicated the call would end. For the purpose of the study the rare case that an operator was unable to answer a question was not considered. The data used in the model can be seen in the table below. Also a screenshot of one model run is shown in the current state OPS screenshot below.
|Call #||Time Between Call (sec)||Account Verification Time (sec)||Question Time (sec)||Total Operator Time (sec)|
After running 20 replications of the current state scenario it was found that the average time a caller spent on the phone during the busiest hour was 5.75 minutes (345 seconds). Of that time 3.25 minutes (195 seconds) was spent talking to an operator and the other 2.5 minutes (150 seconds) was spent waiting. This was the determined to be the baseline for improvement.
After reviewing the current state the team tested what would happen if the account verification step was done by the answering system instead of the operator. An OPS future state simulation was created for that purpose. It used the same information as the current state, but moved the account verification into another process with a higher capacity. A screenshot of a single run of this scenario can be seen below.
After running 20 replications of the proposed future state it was found that the average time a caller spent on the phone has decreased by 40% to 3.42 minutes (205 seconds). This was caused by a decrease in wait time from 2.5 minutes to .22 minutes on average. As the current state wait time was deemed to be acceptable during peak hours, it was determined that the company would like to determine how many operators would be necessary, with the new automated account verification system, to maintain the original total call time. This was done by setting up an experiment. The experiment tested the total call time on average when there were 10 (current state), 9, 8, 7, and 6 operators. Each scenario was run for 20 replications, and the averaged results are shown in the table below.
|Operators||Hold Time (sec)||Operator Time (sec)||Total Time (sec)|
From the future state analysis, it was determined that if the company implemented the automatic account verification system for the call center, then it would be possible to use only 8 operators. Using 8 operators the average wait time of the caller would decrease by 30 seconds. This would save the company approximately $60,000 a year in labor related costs for this call center. Further analysis will be done at other company call centers to identify if this improvement can be used in those locations.