Professional Documents
Culture Documents
Simulation of Call Center Presentation
Simulation of Call Center Presentation
Center
By-
Pradyumna Raghavendra
Methodology
● The project focuses on a call center ● The main focus of this project ● Improved cycle time and
for an electronic company that sells
analyze the data while improving employee productivity.
five different products, namely
television,refrigerator, air
the call center efficiency and
conditioning, toaster and washing lowering the call abandoned rate.
machine.
●In this project, we have used the dataset for the call center model for the electronics
company. The data collected for January 2015 was collected.
●The data for the simulation is collected from Kaggle and is then processed using Excel.
●It was found that the calls arrive according to the exponential distribution. All the incoming
calls either goes to through the system or get abandoned depending on the availability of
the trunk lines.
Collected data introduction
●There are a total of 8 trunk lines – meaning there are 8 operators. All the operators can work for any
●All the completed calls, attempted calls, and abandoned calls are recorded using the record module.
●The company works from 9 AM-7 PM. After 7 PM all the incoming calls are rejected and any work-
in-progress calls are completed using the terminating condition in the simulation run setup.
Parts sold by the electronics company
1. There were missing and inconsistent data values for the dates January 1, 2015, to January 3,
2015. Hence, the data was cleaned by eliminating the values from January 1, 2015- to January 3,
2015. The analysis of the call center data was therefore performed in 4 complete weeks of
2. As per the data, the call center remained closed on Sundays, since there is no data for the same.
Distribution of calls by weeks
●It is observed that the Week 4 (1/25/2015- 1/31/2015) receives the most amount of
calls.
●The abandoned rate is the lowest in the Week 4 ( Jan 25, 2015 to Jan 31, 2015).
●Week 3 has the highest overall satisfaction score with 3.52 and 370 total calls;
but also has the highest amount of the abandoned call rate.
●Week 3 had the highest rate of Satisfaction score.
Percentage distribution of calls by department
● The graph indicates the Monthly Call Data analysis for the
total calls, calls answered, the number of calls not answered
and the % of Abandoned calls sorted with respect to the
agents.
● It is observed that Stewart gets the highest amount of calls
(211) followed by Jim and Dan respectively. Martha has the
highest rate of abandoned calls at 21.83% followed by
Becky with 19.07%.
● We can conclude that Martha, Becky, Stewart, Greg, and
Jim need to improve on answering the total number of calls
by lowering the % of abandoned call rate.
Call resolution rate per agent
● It is observed that Becky has the lowest Avg Speed of Answers of 62.99 seconds followed by
Diane with 63.68 seconds.
● Dan has the highest call resolution rate of 76.85% followed by Jim with a Call resolution rate
of 76.79%.
● In conclusion, all the agents should work on improving the Call Resolution rate and Abandoned
call rate.
Agent overall satisfaction rate
● The MATLAB code simulates 28 days at the call centre for 100 cycles. Hence, the loop
● The number of days can be increased if needed, but this increases the run time.
Algorithm for MATLAB Code
1. Generate a Poisson random variable for the number of calls the call centre receives daily.
2. Divide the number of calls as per department. The percentage values of the calls received by each department are
obtained from the Excel analysis.
Algorithm for MATLAB Code
3. Randomly assign an agent to the calls received by each department. The below snippet shows for
one department. The same is done for other departments as well .
Results obtained from MATLAB analysis
● The first section of the call center simulation model is the call arrival logic.
● It was found that the data was exponentially distributed with a mean of 9.4 minutes. It
was also observed that there were usually two calls for a given minute which leads to 2
calls per arrival.
● When the call arrives, it first goes to the record module, where it is recorded using the
count feature.
● Next, the call goes to the decide module. In this module, the call can either go further
through the call center process or get abandoned depending on the availability of trunk
lines.
Call arrival logic
● There are total 8 trunk lines. If all the trunk lines are busy, the call is abandoned, and it is
● If there is the availability of trunk lines, then the call goes to the seize module and then moves
● After the trunk line is seized by the caller, the call is recorded as a work in progress call by the
● During the initial recording delay, the customer gets a chance to choose one of the five products
which they want to talk about to the call center worker.
● ,From the data collected , it was found that 20% of customers want to complain about the
washing machine, 21% want to talk to the toaster department, 18% want to talk to the
fridge department followed by 19% wanting to talk to air conditioning department and
remaining 22% wants to talk to television department.
● The call then goes to the decide module and depending on the above percentages, the call goes
to the respective departments.
Washing machine department logic
Washing machine department logic
● The assign module in the washing machine department gives the call a specific color and assign
● The call then goes to the processing of the call which takes TRIA(3,10,15) minutes.
● The call then moves forward in the simulation model towards the end of the call and call
disposal.
Toaster department logic
Toaster department logic
● The call then goes to the processing of the call which takes TRIA(3,10,15) minutes.
● The call then moves forward in the simulation model towards the end of the call and call
disposal.
Refrigerator department logic
Refrigerator department logic
● The call then goes to the processing of the call which takes TRIA(3,10,15) minutes.
● The call then moves forward in the simulation model towards the end of the call and call
disposal.
Air conditioner department logic
Air conditioner department logic
● The call then goes to the processing of the call which takes TRIA(3,10,15) minutes.
● The call then moves forward in the simulation model towards the end of the call and call
disposal.
Television department logic
Television department logic
● The call then goes to the processing of the call which takes TRIA(3,10,15) minutes.
● The call then moves forward in the simulation model towards the end of the call and call
disposal.
Call departure logic
Call departure logic
● Once the talk with the customer service representative is completed, the call then goes to
the release trunk line module where the trunk line is released by the call which is then
seized by another incoming call.
● The assigned module of decrement total WIP decreases the variable total WIP. The
record module of total completed calls records the total completed call using the count
feature and is reported in the crystal reports at the end of the simulation.
● The call finally leaves the system using the dispose module.
Improvements suggested
Improvements suggested to improve the
baseline model
● From the analysis of the raw data collected and simulating the Baseline model, it was
observed that there is a need for additional agents at the call center, to make the whole
process more productive..
● The two improvements suggested are :
1. Hiring and training of additional agents
2. Splitting the call center into two departments namely- Tech Department and Sales
Department
Modeling of the
improved scenario
( Proposed model)
Proposed model
Proposed model
● As per the improvement suggested,in this model, the departments have been divided
into two categories – the tech department and the sales department.
● The tech department is further divided into 5 product departments as in the baseline
simulation model.
● It was observed the need for more operators in the sales department than in the tech
department..
Proposed model
● Hence, 6 operators have been assigned to the sales department and 4 operators to
the tech departments. Thus, there is a total of 10 trunk lines for the new proposed
call center simulation.
● The processing of the call in the proposed model takes the calls in TRIA (3,6,10) and
TRIA(3,5,8) minutes in the two replications.
● This proposed model is designed to achieve the maximum customer satisfaction and at
the same time to keep the company’s costs under control.
Results from the simulation
models
Results of the Baseline
model
Entities
Entities
● The simulation model is run for 10 hours (600 minutes) from 9AM to 7PM .
● From the table, It is observed that there is a total of 128 incoming calls out of which the
highest calls are for the television department ( 34 calls) followed by AC, toaster, washing
● The table indicates the number of attempted, completed and rejected calls for the
proposed model.
● The proposed model has two major departments - tech department and sales
department.
● Two replications for the proposed model were performed to find the actual range of
the rejected calls.
● The number of rejected calls in the new proposed model ranges from 0 to 2.
Type of calls and their efficiencies
Replications and their efficiencies
Replications and their efficiencies
● The efficiency for the total number of completed calls for the baseline model is
93.75% and the efficiency of the proposed model has been increased by 5-7%.
● The new proposed model has an efficiency of 98.63 to 100% for the total
answered calls. Hence, the new model is a better fit and more efficient than the
baseline model.
Comparison of results of
the Baseline and the
Proposed models
Baseline vs Proposed model
● The process to find the optimal solution for the call center involved a lot of trial and error.
● The baseline model built based on the raw data gathered from the electronic company’s website
showed that the company is not able to attain customer satisfaction as the number of rejected
calls were approximately 8 calls per hour..
● The proposed model with 98.63-100% efficiency for completing the incoming calls will help
the call center to improve its customer satisfaction ratings.
Future scope of the project
● The call center model ultimately accomplished the most vital requirements of the project
while still needing a few minor adjustments, much like any real system, there will
always be areas for improvement.
● The future scope of this project includes working on a model to accomplish a higher rate
of resolving calls and problems by adding new skilled and trained agents in the
respective departments.
Challenges faced
1. Pandey, A. (2020, December 4). Call center dataset. Kaggle. Retrieved May 11, 2022,
from https://www.kaggle.com/datasets/ashishpandey5210/call-center-dataset/code
2. Design and Simulation of Call Center Using Arena" by S. Karthikeyan, R. Ravi, and V.
Manikandan.
3. ."Call Center Performance Analysis: A Simulation Approach" by Y. F. Akgun, A. E. Aras,
and A. Kocak.
4. "Simulation Modeling and Analysis of Call Center Operations" by P. M. Pinto, D. D. D.
Ruiz, and G. B. Braga.
THANK YOU!