DSC Tech Library
This section of our technical library presents information and documentation relating to Call Center technology and Best Practices plus software and products.
DSC is a leading provider of contact center technology and software solutions as well as predictive dialer phone systems for the modern call center. Customer contact center software includes CRM software and computer telephony integration solutions. These modern products help call center phone agents communicate effectively with your customers and prospects.
The following article presents product or service information relating to call centers and customer service help desks.
Call Center Simulation Modeling:
By Vijay Mehrotra, Department of Decision Sciences
Methods, Challenges, And Opportunities
College of Business - San Francisco State University
Jason Fama, Engineering Group, Blue Pumpkin Software Inc.
From a simulation perspective, each agent is viewed as
a resource to perform certain types of work. Note that in
the call center context, agents are actually productive only
during the interval in which the agents are scheduled to be
actually handling phone calls.
In addition, it is conventional to model agents as completing
the task that they are engaged, even if it extends
past the time at which they are to switch activities. That is,
an agent within our simulation will be modeled as completing
the phone call that he is working on before leaving for
a break or a lunch.
A common step in call center simulation is to translate a
set of individual agent schedules into a matrix of resources,
where the dimensions of the matrix are defined by the number
of Agent Groups and the number of Time Intervals.
In our example, we have leveraged the fact that our
schedules are at a 15-minute level of granularity, and therefore
prior to running the simulation we have converted
these schedules into a number of on-phone agents for each
group for each 15-minute interval.
4.5 Key Inputs: Abandonment
Model and Parameters
Abandonment is one of the most hotly debated topics in
call center management and research. There are two basic
questions that must be answered in order to effectively
model customer abandonment behavior:
1. What is the customer’s tolerance for waiting, and
at what point will this customer hang up and
thereby leave the queue?
Many researchers (e.g. Hoffman and Harris 1986, Andrews
and Parsons 1993) have examined the challenge of
modeling these problems from both an empirical point of
view and from an analytic perspective.
2. How likely is the customer to call back, and after
From our experience, these questions are difficult to
answer not only because of the mathematical complexity of
the queue dynamics but also because of a lack of observable
data about customer abandonment and retrial. While
many surveys have been done, we have observed great differences
in customer behavior across different industries
and different companies’ operations. In addition, information
provided to callers about expected waiting time and/or
position in queue can have a marked impact on abandonment
In our example model, simulated customers arrive at the
call center and are served by an agent if one is available. If
not, they join the queue, at which point they are also assigned
a “life span” which is drawn from an exponential distribution.
If a customer’s life span expires while they are
still waiting in queue, they then abandon the queue.
That is, we represent customers’ tolerance for waiting
in queue as an exponential random variable (as suggested
by Garnett et al. 2002). We refer to the mean of this distribution
as “the patience factor.”
Given this modeling choice, we must still with the
challenge of selecting the patience factor, which we estimate
from historical data about callers’ time in queue.
We do not include caller retrial in the example model.
4.6 Key Inputs: Agent Skills
Our definition of “Agent Skills” is comprised of three major
types of inputs for each agent or group of agents:
1. What calls is the agent capable of handling?
When combined with routing logic and call forecasts,
these attributes fully specify the queueing model to be
2. Given a choice of multiple calls waiting, which
will the agent handle (“call priority”)
3. How fast will the agent be able to handle each
type of call, and how often will the agent resolve
the issue successfully (“call proficiency”)
In our example, we have three distinct groups of
agents, each with different skills:
4.7 Other Modeling Considerations
- Agent Group #1 (Inbound Only) handle only Inbound
calls on a First-Come-First-Served basis.
These agents have a call proficiency of 1.0 for Inbound
Calls, so that their AHT is equal to the
forecasted AHT for Inbound Calls.
- Agent Group #3 (Outbound Specialists) handle
only Outbound calls. These agents have a call
proficiency of 1.0 for Outbound Calls, meaning
that their AHT is equal to the forecasted AHT for
- Agent Group #2 (Cross-Trained Outbound) handle
both Inbound and Outbound calls. These agents
have a call proficiency of 1.0 for Outbound Calls,
meaning that their AHT is equal to the forecasted
AHT for Outbound Calls. However, these agents
will give priority to Inbound Callers if there are any
waiting in queue, and have a call proficiency of 2.0
for Inbound calls, reflecting the relative inefficiency
of cross-training (see Pinker and Shumsky
2000 for more discussion of this phenomenon both
in and out of the contact center).
It is well known that a certain amount of agent time will be
lost, either in large blocks (unanticipated shift cancellations,
partial day absences for personal reasons) or in small
blocks (late arrivals to the call center, extra-long breaks,
trips to the bathroom).
There is an important distinction between two different
kinds of lost agent time. On one hand, agent time that
is known to be lost prior to the creation and publication of
a schedule has essentially no additional impact on the
simulation model beyond the fact that this particular agent
is not included in the schedule.