Content uploaded by Naveen Donthu
Author content
All content in this area was uploaded by Naveen Donthu on Jul 25, 2023
Content may be subject to copyright.
Closing the gap between perceived and
actual waiting times in a call center:
results from a field study
Anita Whiting
Department of Marketing, School of Business, Clayton State University, Georgia, USA, and
Naveen Donthu
Department of Marketing, Georgia State University, Atlanta, Georgia, USA
Abstract
Purpose – The purpose of this paper is to investigate what factors influence the gap between caller’s perception of how long they think they waited
and how long they actually waited on hold and to determine what call managers can do to reduce this gap called estimation error.
Design/methodology/approach – A field experiment was conducted with a corporation’s call center.
Findings – The findings were: the higher the estimation error of callers, the less satisfied they are; music increases estimation error, unless callers can
choose the music; waiting information reduces estimation error; callers with urgent issues have more estimation error and they overestimate more; and
females have higher estimation error and they overestimate more than males.
Research limitations/implications – Limitations are one call center in one context. Implications are identification of antecedents of overestimation.
Practical implications – The paper provides guidelines for call center managers for reducing estimation error and increasing caller satisfaction. It
discusses the need for understanding callers and measuring items that are important to them.
Originality/value – The study investigates an under researched variable called estimation error. Study also provides information about some of the
causes for why consumers overestimate or underestimate their waiting time. Study provides guidelines from an actual call center and discusses
variables that managers can easily use to decrease estimation error and overestimation.
Keywords Call centres, Operating times, Estimation, Individual perception
Paper type Research paper
An executive summary for managers and executive
readers can be found at the end of this issue.
Introduction
Call centers have become the dominant form of contact with
customers (Micak and Desmarais, 2001). Over 70 percent of
customer contact occurs through call centers (Feinberg et al.,
2002). Because call centers handle a diverse array of issues
ranging from complaint resolution to order taking, call
centers have become a critical touch point for managing and
increasing customer satisfaction (Anton, 1997; Dawson,
1998). According to Bennington et al. (2000, p. 162), call
centers have the potential to become the “hub of successful
customer relationship management (CRM) strategies and the
fulcrum of organizations”. Call centers will only continue to
grow in importance as more and more companies focus on
CRM (Burgers et al., 2000).
With call centers becoming a critical touch point for most
organizations, it is important to investigate and understand
factors that influence caller satisfaction. Despite the
contributions of research on service quality and call centers,
there is still a strong need for research in this area (Jack et al.,
2006). Organizations with call centers have been criticized for
focusing on what is easy to measure (e.g., number of callers
served per hour) instead of what is important to measure
(e.g., perceived wait time) and for focusing on quantity of
calls instead of quality of calls (Robinson and Morley, 2006).
Academic literature also lacks knowledge about what is
important to caller satisfaction (Feinberg et al., 2002). Most
academic studies on call centers have focused on employee
issues such as staff dissatisfaction and emotional labor rather
than on caller satisfaction (Bennington et al., 2000). Feinberg
et al. (2002, p. 179) claim that uncovering the significant
variables that influence caller satisfaction is “crucial if we are
to provide guidance for call center managers”. Thus, both
managers and academics are very concerned about the lack of
knowledge about what influences and drives caller
satisfaction.
Within the few studies that have been conducted on caller
satisfaction, there is one important variable that has been
shown to influence callers and that variable is waiting time.
Millions of customers wait on hold in telephone queues to
speak to a call center representative (Knott et al., 2004). This
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/0887-6045.htm
Journal of Services Marketing
23/5 (2009) 279– 288
qEmerald Group Publishing Limited [ISSN 0887-6045]
[DOI 10.1108/08876040910973396]
Received: October 2007
Revised: May 2008, September 2008
Accepted: September 2008
279
waiting on hold experience has been shown to directly impact
satisfaction (Whiting and Donthu, 2006; Antonides et al.,
2002; Unzicker, 1999). There are two important variables in
a waiting on hold experience. The first variable is the actual
(objective) waiting time which is defined as how long the
customer actually waited on hold (Hornik, 1984). The second
variable is perceived (subjective) waiting time which is defined
as how long the customer thinks they waited on hold (Hornik,
1984).
For most consumers, there is usually a gap or discrepancy
between actual and perceived waiting time with most
consumers overestimating how long they have waited
(Hornik, 1984; Katz et al., 1991; Chebat et al., 1991; Knott
et al., 2003). This discrepancy between perceived and actual
wait times is defined as an estimation error (Knott et al.,
2003). Some researchers refer to estimation error as
overestimation but some consumers may underestimate
their waiting time too. Estimation error is a very important
variable because it has been show to influence customer
satisfaction (Jones and Peppiatt, 1996).
Because estimation error has been shown to have a
significant impact on customer satisfaction, it is important
to investigate variables that influence the discrepancy or gap
between perceived and actual waiting times within a call
center context. Call center managers need to know what
variables are causing estimation error and what factors are
causing it to increase or decrease. In particular, are there
some variables that are causing callers in a call center to
overestimate their waiting time while other variables are
helping callers to be more accurate in their perceptions of
their on hold waiting time? Answering these questions and
helping call center managers to decrease estimation error
(especially overestimation) is the goal of this research project.
In particular, this paper will develop and empirically test a
conceptual model that examines determinants of estimation
error and its impact on caller satisfaction in a call center. The
model contends that real time, expectations, individual
differences during the wait, and situational factors during
the wait will influence estimation error and satisfaction within
a call center context.
This article seeks to make many contributions to the
marketing and call center literature. First, this article focuses
on perceived wait times, actual wait times, and estimation
error within a call center. Most of the research on wait times
has focused on either perceived wait times or actual wait times
but rarely the discrepancy between the two. Second, this
research extends the waiting time literature by investigating
the neglected variable called estimation error. Third, this
study seeks to explain what factors cause estimation error and
why some consumers overestimate their waiting time while
others underestimate their waiting times. This research also
seeks to add to the literature by investigating waiting times in
a new context that is a call center. Most services literature has
focused on waiting times in physical settings such as banks,
hospitals, and fast food restaurants. However, according to
Maister (1985), people will perceive waits differently under
different circumstances and therefore, waiting on the
telephone may be very different than waiting in an actual
service environment. Thus, the findings in a call center may
be very different from previous studies in physical service
environments. Last, this article provides managerial
implications and guidelines to help call center managers
decrease estimation error and overestimation.
The article first begins by summarizing the literature on
actual waiting times, perceived waiting times, and estimation
error. Next, the model is presented and discussed. Third, the
article describes the methodology and data collection. Fourth,
the article describes the findings from the study and, finally,
the article discusses the implications and conclusions from the
study and future research opportunities.
Literature review
More and more businesses are adding call centers to their
organization. According to the Center for Customer Driven
Quality (CCDQ) at Purdue University, the number of call
centers has grown from 75,000 in 2001 to an estimated
115,000 in 2005. Approximately 98 percent of Fortune 500
companies have call centers (Feinberg et al., 2002). Many
organizations are adding call centers because their customers
are demanding and expecting telephone access to companies
(Cowles and Crosby, 1990).
As the number of call centers continues to grow, businesses
must begin to investigate and focus more on managing the on
hold telephone wait experience. Waiting on hold to speak to
an employee may not be a pleasant experience for some
consumers. Many consumers are very conscious of their time
costs when waiting (Berry, 1979) and most consumers resent
having to wait (Unzicker, 1999). Consumers who have a
negative wait experience may even retaliate against businesses
by switching to competitors and spreading negative word of
mouth (Tom et al., 1997). In order to keep customers happy
and satisfied, businesses must be concerned about their
customer’s waiting on hold telephone experiences.
As discussed previously there are two important variables
within an on hold telephone experience. These two variables
are actual waiting time and perceived waiting time. Studies on
these variables have shown that both influence customer
satisfaction (for a review of waiting time literature see
Durrande-Moreau (1999)). Estimation error is the difference
in perceived and actual wait times.
Estimation error is a very important variable because it
occurs very frequently among many consumers. Most
estimation error studies have focused on overestimation but
consumers may also underestimate their actual waiting time.
Research on overestimation error has shown that many
consumers greatly overestimate how long they have waited.
According to Jones and Peppiatt (1996, p. 47), it is commonly
assumed that “the average customer’s perception of waiting
time is different from reality” with most customers thinking
that they have waiting longer than they actually have. Other
studies have also found that most consumers overestimate
their waiting time. Hornik (1984) conducted a field study on
waiting times within the retail industry and found that
consumers in a shopping context overestimated their waiting
time by 36 percent. Katz et al. (1991) found that bank
customers overestimated their waiting times by twenty five
percent. Jones and Peppiatt (1996) found that their
respondents overestimated their wait times by 40 percent.
Feinberg and Smith (1989) found that 77 percent of its
respondents overestimated their waiting times. Thus,
estimation error is occurring in many consumers and the
error or discrepancy between actual and perceived waiting
time is rather large with most consumer overestimating how
long they have waited.
Closing the gap between perceived and actual waiting times
Anita Whiting and Naveen Donthu
Journal of Services Marketing
Volume 23 · Number 5 · 2009 · 279 – 288
280
With so many consumers experiencing estimation error and
by such a large percentage, it is important to investigate what
drives the discrepancy between perceived and actual waiting
times especially within a call center context. As previously
discussed there are only a few studies that have investigated
estimation error. However, these studies did not investigate
the causes of estimation error and they did not investigate
estimation error within a call center context. Most waiting
time studies have focused on the customer’s perception of the
wait (and not actual wait time) and most studies have focused
on waiting within a service setting (e.g., bank or store) and
not on the telephone. Most wait studies collected perceived
wait times but they did not measure actual wait times; and
thus did not investigate estimation error (Jones and Peppiatt,
1996). The lack of literature on estimation error may be due
to the challenges of collecting actual wait times from
consumers. This article seeks to address this gap in the
literature by developing a model of determinants that
influence estimation error and caller satisfaction within a
call center context.
Model development
In order to investigate the determinants of estimation error,
we chose to rely on Durrande-Moreau’s (1999) review of the
waiting literature. She reviewed over 30 papers on wait
management between the years of 1984 through 1997 and she
concluded that there are six factors that influence consumers
while waiting. These six factors are:
1 real time;
2 personal expectations;
3 individual factors before the wait;
4 situational factors before the wait;
5 individual factors during the wait; and
6 situational factors during the wait.
Because factors before the wait cannot be easily controlled by
call center managers, we chose to focus on four variables that
can be managed and their impact on estimation error and
caller satisfaction. The four variables are:
1 real time;
2 personal expectations;
3 individual factors during the wait; and
4 situational factors during the wait.
The effects of these four variables will be explained by
applying Zakay’s (1989) Resource Allocation model (see
Figure 1).
It is important to note that Durrande-Moreau’s review did
not find estimation error to be a frequently investigated
variable. Most of the studies reviewed by Durrande-Moreau
were focused on perceived wait times and satisfaction. This
study builds upon and extends Durrande-Moreau’s review by
investigating four of her six variables and their impact on
estimation error.
Real time
Real time has been shown to have a negative impact on the
waiting experience. According to Durrande-Moreau’s (1999)
review, real time was the central stimulus for reactions to the
wait and that the longer the duration, the more negative the
reaction to the wait. In addition to satisfaction, real time may
also have an impact on estimation error. Studies on estimation
error have found that shorter the wait, the greater the
estimation error (Davis and Volman, 1990; Jones and
Peppiatt, 1996). Evangelist et al. (2002) found that
customers with waits of less than three minutes were more
likely to overestimate their waiting time while customers with
waits greater than five minutes were more likely to
underestimate their waiting time.
The inverse effect of real time on estimation error can be
explained by Zakay’s (1989) Resource Allocation model.
Zakay’s model proposes that time perception is a function of
the number of time units recorded by a cognitive timer. This
cognitive timer is activated when people pay attention to the
passage of time. At the beginning of the waiting experience,
consumers are occupied with the passage of time and they
actively engage in time estimations. However, as the wait
continues, consumers become distracted by stimuli and they
begin to make fewer time estimations. These fewer wait
estimations lead to more accurate perceptions of the wait time
or even under evaluations of the wait time. Based on Zakay’s
model and on the findings in physical service settings, the
following hypothesis is proposed:
H1. The longer the actual waiting time in a call center, the
lower the estimation error.
Estimation error has also been shown to influence consumer
satisfaction. Jones and Peppiatt (1996) investigated the gap
between actual and perceived waiting times and found that
estimation error had an impact on satisfaction. In particular,
they found that higher estimation error leads to less
satisfaction. This inverse relationship between estimation
error and satisfaction can be explained by Parasuraman et al.’s
(1985) widely accepted service quality model. According to
the model, there is a gap between actual delivery of service
(actual wait time) and customer’s perception of the service
(perceived wait time). This gap along with the other gaps has
a negative influence on customer satisfaction and service
quality. Therefore, the following hypothesis is proposed:
H2. The higher the estimation error of a caller, the lower
the caller’s satisfaction.
Personal expectations
According to Durrande-Moreau’s (1999) review, personal
expectations strongly influence outcome variables. In her
review of 18 articles on expectations and waiting time, she
observed a “classical comparative mechanism between
expectation and reality that exemplifies the confirmation-
disconfirmation paradigm” (Durrande-Moreau, 1999, p. 175).
She also reported that customers who expect a short wait will
react more negatively than others. We predict that
expectations of a short wait will have a negative impact on
estimation error and satisfaction. This prediction is based on
the discrepancy theory (Michalos, 1985) and the expectancy
disconfirmation paradigm. These theories suggest that
consumers establish expectations, observe the performance,
compare the performance to expectations, and then form
disconfirmation perceptions (Yan and Lotz, 2006). When the
disconfirmation between expectations of the wait and the
actual wait time is large, the estimation error will be large and
the consumer will be less satisfied. The following hypotheses
are therefore proposed:
H3. Individuals with expectations of a short wait will have
higher estimation error than those with expectations of
a longer wait.
Closing the gap between perceived and actual waiting times
Anita Whiting and Naveen Donthu
Journal of Services Marketing
Volume 23 · Number 5 · 2009 · 279 – 288
281
H4. Individuals with expectations of a short wait will have
lower customer satisfaction scores than those with
expectations of a longer wait.
Individual factors during the wait
Maister (1985) proposed that people will perceive waits
differently. There have been many individual factors reported
to influence waiting times such as type of customers
(experienced vs novice), value of purchase, and time
pressure. Based on Durrande-Moreau’s (1999) review of
individual factors, we chose to look at music preference,
gender, and experience and their impact on estimation error
and satisfaction.
Music played during the wait has been shown to influence
waiting times. In particular, music has been shown to
influence perceived waiting duration and behavior (Hui et al.,
1997). Music adds to the service environment and helps
create a more positive evaluation (Baker et al., 1992). Both
Kellaris and Kent (1992) and Katz et al. (1991) found that
playing music reduces the negative effects of waiting. North
et al. (1999) investigated the effects of liking and fit of music
on the amount of time callers would stay on hold. He found
that callers would wait on hold longer when music they liked
was played. The beneficial effects of liked music can be
explained by Zakay’s (1989) Resource Allocation model.
According to the model, consumers are occupied with the
passage of time and they actively engage in time estimations.
However, when liked music is played, consumers become
distracted by the music and they begin to make fewer time
estimations. These fewer wait estimations lead to more
accurate perceptions of the wait time or even under
evaluations of the wait time. We therefore predict that liked
music will have a positive influence on estimation error. The
following hypothesis is proposed:
H5. Callers who like the music played will have lower
estimation error than callers who don’t like the music
played.
Gender is also another individual factor that may influence
estimation error. Gender has been shown to influence many
outcome variables. According to Karatepe et al. (2006,
p. 1088) there is a distinction between how each gender
“observes the environment, processes, evaluates and retrieves
information, and makes judgments”. Women look at the
details and process lots of information when making decisions
while men use heuristics and process less information
(Sunden and Surette, 1998). It has also been shown that
females experience higher levels of stress (Nelson and Quick,
1985). Based on these findings, we predict that gender will
influence estimation error. The following hypothesis is
proposed:
H6. Females will have higher estimation error than males.
Despite the previous findings on gender, we acknowledge that
other studies have found contradictory results showing that
there are no differences between men and women and their
waiting experiences. Both Davis and Volman (1990) and
Jones and Peppiatt (1996) did not find any gender differences
in their waiting studies. However, there were many other
variables investigated in their studies which may have caused
noise in the data and thus caused the gender differences not to
come through.
Experience may also play a role in estimation error.
Customer’s prior experience has been shown to influence
both perceived wait and satisfaction (Davis and Volman,
1990). Jones and Peppiatt (1996) found that new or
infrequent users had significantly higher perceived wait
times than frequent users. Customers’ prior experience can
also by explained by Zakay’s Resource Allocation model. New
and inexperienced users may focus on the passage of time and
actively engage in time estimations. Experienced users may
not engage in as many time estimations because they are
familiar with the waiting situation. We therefore predict that
lack of experience will have a negative impact on estimation
error. The following hypothesis is proposed:
H7. Novice callers will have higher estimation error than
experienced callers.
Situational factors during the wait
Durrande-Moreau’s (1999) found that situational factors
were the most examined factor and that many of them have an
influence on consumers. Some of the situational variables
investigated have been type of queue, television, and
information displays. For this study, we chose to focus on
Figure 1 Model of estimation error in a call center
Closing the gap between perceived and actual waiting times
Anita Whiting and Naveen Donthu
Journal of Services Marketing
Volume 23 · Number 5 · 2009 · 279 – 288
282
presence of music, waiting information given, and urgency of
the call.
The presence of music has been shown to impact many
important dependent variables such as length of time in store,
amount purchased, and likelihood of returning (Oakes,
2000). Presence of music differs from the previously
mentioned variable about liking the music that is played.
Liking the music played is an individual factor while presence
of music is a situational factor that deals only with the
presence or absence of music. Research on the presence of
background music has been shown that it has positive effects
on consumers by decreasing stress and increasing relaxation
(Tansik and Routhieaux, 1999). Research on music has also
shown that there is a significant relationship between waiting
and music (Chebat et al., 1993). Music has been shown to
reduce perceived wait times in restaurants and supermarkets
(Milliman, 1982, 1986). These positive effects can be
explained by Zakay’s Resource Allocation model. The
presence of music may distract individuals from the passage
of time and it may cause them not to engage in as many time
estimations. These fewer time estimations may lead to more
accurate perceptions of the wait time or even under
evaluations of the wait time. We therefore predict that the
presence of music will decrease estimation error. The
following hypothesis is proposed:
H8. Callerswithbackgroundmusicwillhavelower
estimation error than callers without background
music.
In addition to music, waiting information has also been shown
to influence consumer’s perception of the wait (Ahmadi,
1984; Katz et al., 1991). There are two types of waiting
information: estimated wait time and number in the queue.
Estimated wait time is information about the expected length
of the wait and queuing information is the consumer’s
position in the queue (Hui and Tse, 1996). According to
Maister (1985) uncertain waits are longer than known waits.
Zakay and Hornik (1994) suggest that information about the
wait reduces consumers from thinking about how long they
have been waiting and thus reduces their perception of the
waiting time. Based on these findings, we predict that waiting
information will have a positive impact on estimation error.
The following hypothesis is proposed:
H9. Callers with waiting information will have lower
estimation error than callers without waiting
information.
Urgency of the call may influence estimation error. Criticality
of time to the customer has been shown to influence
perception of the wait and satisfaction (Davis and Volman,
1990). Davis and Heineke (1998) found that satisfaction with
the wait depends on the differences in the needs of the
consumer. According to Maister (1985), people perceive
waits differently under different situations such as an urgent
situation. Maister also proposes that uncomfortable waits
(such as an urgent call) seem longer than comfortable waits.
The relationship between urgency and estimation error can be
explained by Zakay’s Resource Allocation model. Individuals
with urgent issues are very focused on the passage of time and
they are constantly making time estimations. These frequent
time estimations may lead to very inaccurate accurate
perceptions of the wait time with most urgent callers greatly
overestimating their wait time. We therefore predict that
urgency of the call will have a negative impact on estimation
error. The following hypothesis is proposed:
H10. Callers with urgent issues will have higher estimation
error than callers with nonurgent issues.
Methodology
Overview
A national corporation agreed to let us use their call center to
collect data. Their call center supports franchisees with their
point of sales systems and their back office systems. The
respondents in this study were independent franchise owners
who paid monthly fees for the services provided by the call
center. The call center was currently using background music
and two information cues and they wanted to see how these
variables were affecting their callers. Therefore, we conducted
an experiment to investigate these situational factors while
also gathering data on other variables.
For the experiment, we manipulated music, estimated wait
time given, and number in the queue given. The experiment
consisted of eight different treatments. The treatments were:
1 no music, no information;
2 with music, no information;
3 no music, estimated wait time given;
4 with music, estimated wait time given;
5 no music, number in queue given;
6 with music, number in the queue given;
7 no music, both estimated wait time and number in queue
given; and
8 with music, both estimated wait time and number in
queue given.
The experiment was conducted over an eight-week period
with each week being a different treatment.
Procedure
Franchisees would call into the call center for assistance with
issues about their point of sales systems or their back office
systems. While they were on hold, the callers were exposed to
the treatment for that week. The experiment was conducted
over eight weeks and measures were taken to ensure that
survey respondents were only questioned once about their
waiting hold experience. Call center employees were
instructed to write down the actual waiting time of each
caller and the store’s number (both of which were on the
computer screen) as they answered each call. The employees
also wrote down the caller’s name. Later that evening, an
e-mail survey was sent to the franchisee at the store’s e-mail
address. The e-mails were sent out from a university
e-mail address so that callers could be more candid with their
responses. Reminder e-mails were also sent out a day after the
initial e-mail. A total of 211 completed e-mail surveys were
returned. The response rate was approximately 18 percent.
Measures
The survey consisted of 14 questions. Participants were asked
to state the reason for their call and they were asked how long
they think they waited on the phone before an agent answered
the call. Participants were also surveyed about the presence of
music and their feelings about the music. Additional
questions on the survey were about expectations about the
wait and satisfaction with the wait. Actual wait time data and
perceived wait time data were matched up for each caller and
Closing the gap between perceived and actual waiting times
Anita Whiting and Naveen Donthu
Journal of Services Marketing
Volume 23 · Number 5 · 2009 · 279 – 288
283
thus estimation error was calculated. The company also
provided additional information to help with the data analysis.
Gender and years of experience as franchisee (based on
company records) were provided by the organization. A
supervisor in the call center categorized the data from the
reason for the call question into two categories: urgent and
nonurgent calls.
Data analysis and results
We first began our data analysis by analyzing the average wait
times for each treatment group. We did not find actual wait
times for each treatment group to be significantly different; so,
we were able to move forward with analyzing the data. For
each treatment, we calculated estimation error by subtracting
the actual wait times from the perceived wait times. We
conducted an ANOVA on estimation error and found that
estimation error approached significance at the 0.08 level
among all eight treatments. There were also significant
differences among some of the treatments but this will be
discussed later under situational factors. We also conducted
an ANOVA on perceived waiting times and found that there
were no significant differences in perceived wait times among
the treatments. Thus estimation error was impacted more
than perceived wait times by the experimental variables of
music and information cues.
The average wait time for the entire sample was 1 minute
and 38 seconds with actual wait times ranging from 12
seconds to 8 minutes. The average perceived wait time for the
entire sample was 3 minutes and 23 seconds with perceived
wait times ranging from 20 seconds to 15 minutes. The
average estimation error for the entire sample was 1 minute
and 39 seconds with estimation error ranging from less than 4
minutes to 10 minutes. Approximately 79 percent of the
sample overestimated their waiting time. The average
estimation error for each of the eight treatment groups is
shown in Table I.
Next we analyzed the data from our hypotheses on real
time, personal expectations, individual factors during the
wait, and situational factors during the wait. Our empirical
findings of these four variables on estimation error are
discussed below (see Table II).
Real time
We first investigated the actual length of the wait and its
impact on estimation error and found a significant negative
correlation between the two (20.243 correlation which is
significant at 0.05 level). Thus, H1 was supported and the
finding was that as the actual wait increases, estimation error
declines. Next, we investigated the relationship between
estimation error and caller satisfaction.
As predicted by H2, estimation error had a significant
negative correlation with satisfaction (20.280 which is
significant at 0.01 level). Thus, the more estimation error,
the less satisfied they are.
Expectations
Next, we investigated how expectations about the wait would
impact estimation error and satisfaction. Findings from an
ANOVA tests showed that there we no significant differences
in estimation error among those whose waits were more than
expected, about what expected, and less than expected. Thus,
H3 was not supported. We also investigated how expectations
impact satisfaction using ANOVA. We found there to be a
significant difference in satisfaction scores among consumers
with difference expectations (0.001 sign). Thus, H4 was
supported. Overall, we found expectations about the wait did
notimpactestimationerrorbutitdidgreatlyimpact
satisfaction.
Individual differences during the wait
The three individual differences that we investigated were
feelings about the music, gender, and experience (novice vs
experienced user). An ANOVA test on feelings about music
and estimation error was significant at 0.027. The estimation
error scores were 1.88 for no music, 0.275 for liked music,
and 2.18 for does not like music. Pair wise comparison tests
showed a 0.021 significant difference between no music and
liked music and a 0.011 significant difference between liked
music and does not like music. Thus, H5 was supported and
the major finding was that estimation error is significantly less
for consumers who like the music that is played. For gender,
we conducted another ANOVA on estimation error. From this
analysis, we found gender to be significant at 0.016. The
average estimation error score was 0.559 for males was and
2.27 for females. Thus H6 was supported and the major
finding was that females have higher estimation error scores
than males and they significantly overestimate wait times
more than males. For experience, we categorized our
respondents into two categories: novice (0-2 years of
experience) and experienced (three to 30 years). Despite
correlation and ANOVA tests, we did not find experience to
influence estimation error or satisfaction. Thus, H7 was not
supported.
Situational factors during the wait
The three situational variables that we investigated were:
1 presence of music;
2 waiting information given; and
3 urgency of call.
Presence of music was manipulated in the experiment with
callers with four of the eight treatments having music played
in the background while waiting on hold. An ANOVA was
conducted on estimation error with treatments as the
independent variable. Treatments approached significance at
the 0.08 level. Pair wise comparisons of the different
treatment groups yielded significant findings. Treatment 3
(no music and estimated wait time given) and treatment 4
(with music and estimated wait time given) were significantly
Table I Average estimation error in experimental treatments
Treatment
Average
estimation error
(1) No music, no information 2.25
(2) With music, no information 2.34
(3) No music, estimated wait time given 20.34
(4) With music, estimated wait time given 2.07
(5) No music, number in queue given 0.36
(6) With music, number in the queue given 2.59
(7) No music, both estimated wait time and
number in queue given 0.37
(8) With music, both estimated wait time and
number in queue given 1.70
Closing the gap between perceived and actual waiting times
Anita Whiting and Naveen Donthu
Journal of Services Marketing
Volume 23 · Number 5 · 2009 · 279 – 288
284
different at the 0.05 level. The estimation error values for
treatments 3 and 4 are 20.34 and 2.07 respectively. When
treatment 5 (no music and number in queue information
given) was compared to treatment 6 (with music and number
in queue information given), a 0.26 significant difference was
found. The estimation error values for treatment 5 and 6 are
0.36 and 2.59 respectively. These findings demonstrate that
the presence of music actually increases estimation error.
Thus, H8 was not supported. The overall finding from this
analysis is that the presence of music greatly increases
estimation error among callers.
Waiting information
It was also manipulated in the experiment. The four
categories were:
1 no information given;
2 estimated wait time given;
3 number in queue information given; and
4 both estimated wait time and number in queue
information given.
AnANOVAwasconductedonoverestimationwith
treatments being the independent variable. Treatments
approached significance at the 0.08 level. Pair wise
comparisons of the different treatment groups against the
control group yielded three significant findings. Treatment 1
(no music and no information) and treatment 3 (no music
and wait information given) were significantly different at the
0.003 level. The estimation error values were 2.25 and 20.34
respectively. Thus, estimated wait time information
significantly reduced estimation error among callers. The
second pair wise analysis showed significant differences at the
0.014 level between treatment 1 (no music and no
information) and treatment 5 (no music and number in
queue information given). The estimation error values for
treatments 1 and 5 are 2.25 and 0.36 respectively. Thus, the
second finding is that number in queue information greatly
reduces estimation error among callers. The third pair wise
analysis showed significant differences at the 0.046 level
among treatment 1(no music and no information) and
treatment 7 (no music and both estimated wait time and
number in queue given). The estimation error values for
treatments 1 and 7 are 2.25 and 0.37. Thus, both estimated
wait time and number in queue information greatly reduced
estimation error among callers. Thus, H9 was supported and
the overall finding from these three analyses was that waiting
information does in fact greatly reduce estimation error
among callers.
Urgency of call
An ANOVA was conducted on estimation error with urgency
of call as the independent variable. Urgency of the call was
found to be significant at the 0.013 level. The average
estimation error for nonurgent calls was 1.11 while average
estimation error for urgent calls was 2.60. Thus, H10 was
supported and the overall finding was that callers with more
urgent issues are more likely to have more estimation error
and are also more likely to inflate how long they have waited
on hold.
Discussion and implications
An experiment was conducted at a call center to understand
what drives the gap between perceived and actual waiting
times in a call center. This was done in order to help call
center managers understand the huge impact that estimation
error has on caller satisfaction and to provide guidelines for
reducing estimation error and increasing satisfaction. Call
centers are a critical customer touch point that must be
managed. Managers of call centers have to not only focus on
reducing the actual waiting time for callers put on hold, but
they must also manage and align the perception of wait time
more closely with the actual waiting time, and thus decrease
estimation error. Managers must focus on closing the gap
between perception and reality.
From our study, we found eight major findings. These
findings are very actionable and they have major implications
for call center managers. The major findings are:
.The larger the gap between perceived and actual waiting
times, the less satisfied consumers are.
Table II Summary of hypotheses and findings
Hypotheses Findings
Real time
1 The longer the actual waiting time in a call center, the less the estimation error Supported at .05 level
2 The higher the estimation error of a caller, the lower the caller’s satisfaction
Expectations
3 Individuals with expectations of a short wait will have higher estimation error than those with expectations of a
longer wait
Not supported
4 Individuals with expectations of a short wait will have lower customer satisfaction than those with expectations of a
longer wait
Supported at 0.01 level
Individual factors
5 Callers who like the music played will have lower estimation error than callers who do not like the music played Supported at 0.05 level
6 Females will have higher estimation error than male Supported at 0.05 level
7 Novice callers will have higher estimation error than experienced callers Not supported
Situational factors
8 Callers with background music will have lower estimation error than callers without background music Supported at 0.05 level
9 Callers with waiting information will have lower estimation error than callers without waiting information Supported at 0.05 level
10 Callers with urgent issues will have higher estimation error than callers with non urgent issues Supported at 0.05 level
Closing the gap between perceived and actual waiting times
Anita Whiting and Naveen Donthu
Journal of Services Marketing
Volume 23 · Number 5 · 2009 · 279 – 288
285
.As actual wait increases, estimation error declines.
.Expectations about the wait do not impact estimation
error, but it does greatly impact customer satisfaction.
.Presence of music greatly increases estimation error
among callers.
.However, estimation error of waiting times is significantly
less for consumers who like the music that is played.
.Compared to males, females have significantly more
estimation error and they overestimate their wait time
more.
.Waiting information does greatly reduce estimation error
and overestimation among callers.
.Callers with more urgent issues have more estimation
error are more likely to inflate how long they have waited
on hold.
Implications
First, call centers are a major customer touch point that must
be managed and it must be managed well. A majority of
customer contact occurs through them and the image of the
company from the customer’s eyes can either be demolished
or enhanced from interactions with a call center. There are
major consequences for companies that do not focus on caller
satisfaction. Dissatisfied customers are more likely to spend
less, switch to competitors, and spread negative word of
mouth communication. Organizations that do not focus
heavily on caller satisfaction may want to evaluate how much
and how important the business activities are that occur
through their call centers. For organizations that conduct
critical and/or numerous activities through their call center,
the stakes are much higher for increasing caller satisfaction.
Second, organizations must focus on estimation error.
Estimation error occurs frequently with most consumers
thinking that they have waited significantly longer than they
actually have. Decreasing estimation error and overestimation
should be an initiative for organizations because of its direct
impact on customer satisfaction. Organizations must also
collect data on perception of the wait and estimation error
(and not just collect actual wait time data because of its
convenience). Information on perceptions of the wait and
estimation error is important because consumers may feel like
a one-minute wait lasted ten minutes. Organizations may also
want to rethink the data (e.g., number of calls handled per
employee) that they are collecting to see if that data are really
important to customers. From the study we found that actual
wait times did not influence customer satisfaction, but
perception of the wait and most significantly estimation error
of the wait did. However, not many organizations are
currently collecting this data.
Third, just playing music while callers are put on hold is not
adequate. Music should be liked by the callers for it to reduce
the estimation error of waiting time. Organizations should
research their customers and match the music to the
customer’s preferences. For organizations with many
different types of consumers, they may want to use a system
that would give callers a choice of music. Organizations may
also want to consider the option of silence or the news.
Silence may be soothing for stressed customers and it may be
better than listening to something that is not liked. News may
also be a good option for consumers who want to be informed
and educated while waiting on hold instead of just wasting
time.
Fourth, organizations should consider providing
information cues such as estimated wait times. Information
cues reduce estimation error and it helps callers be more
realistic about how long they actually waited. Providing
waiting information also helps reduce uncertainty and it helps
consumers decide whether or not to wait on hold. Consumers
who hear an estimated wait time of 25 minutes may quickly
decide to call back later. Estimated wait times also set
consumers’ expectations for how long they will have to wait.
With telephone holds, consumers can not see the virtual line
of callers in front of them and they have no way of knowing
how long the wait may be. Managing customer’s expectations
is critical for providing good customer service.
Fifth, urgent callers should be given special attention and if
possible not be put on hold. One way to address this is by
letting callers specify why they are calling via a menu
selection. An example of this would be press 1 for computer
problems, 2 for payroll problems, and 3 for e-mail problems.
More urgent calls should be answered first. Organizations
may also use different telephone numbers for different types
of calls.
Sixth, organizations need to understand their callers. This
study and a study by Jones and Peppiatt (1996) found that
females had more estimation error and they overestimated
more than males. Thus, organizations need to know the
gender of their target audience. Retailers whose customer
base is mostly women need to focus heavily on perception
techniques. The waiting strategies of organizations must line
up with the target market that it serves.
Overall, the theme that emerges from these implications is
that organizations must understand who the caller is.
Organizations need to know their gender, music preference,
and urgency of the call. By using customer profiling, the call
center can do every thing possible to reduce estimation error
and overestimation of waiting times. Reducing estimation
error and overestimation of waiting times is key to increasing
caller satisfaction. From CRM initiatives, organizations are
moving toward understanding and creating relationships with
its customers. Call centers are a crucial touch point where
relationships can be created and grown.
Limitations and future research
Even though this study was conducted in a corporate call
center, it still has some limitations. First, the current study
findings are based on one experiment that was conducted in
one call center. Different types of call centers may produce
different results. Another limitation is that waits of different
lengths may produce different responses. The average wait
time in this study was approximately two minutes. Future
work should increase the generalizability of the findings by
testing in other contexts with different samples and with
different waiting times. Despite these limitations, this field
study demonstrated that organizations can close the gap
between perception and reality and decrease overestimation.
Because estimation error occurs with almost all consumers,
it is an area that must be further investigated. Businesses need
to understand the causes for people inflating their waiting
time. This study only looked at a few causes of estimation
error but there are lots more to investigate. In particular,
future studies could look at additional individual and
situational factors. For individual factors, researchers could
look at different types of callers, frequency of the caller,
Closing the gap between perceived and actual waiting times
Anita Whiting and Naveen Donthu
Journal of Services Marketing
Volume 23 · Number 5 · 2009 · 279 – 288
286
reason for call, and age. Additional individual factors to
explore are callers that balk and do not even wait on hold and
callers that renege and hang up before their call is answered.
There is much to be gained by understanding why and which
type of callers hangs up before their call is answered. There
are also many situational factors to be investigated. Music has
many different components (e.g., tempo, volume, type, etc.)
that could be researched. Researchers could also explore
reason for call (e.g., billing, making purchase, complaining),
day of week, and time of day. There are also additional audio
items to be investigated such as announcements or
advertisements to callers while holding. Caller satisfaction
and estimation error is an under researched area that has
endless opportunities. The importance of this research area
will only continue to grow as more and more organizations
focus on CRM and try to increase customer satisfaction.
References
Ahmadi, K. (1984), “Effects of social influences and waiting
on time judgment”, Perceptual and Motor Skills, Vol. 59,
pp. 771-6.
Anton, J. (1997), Call Center Management by the Numbers,
Purdue University Press, West Lafayette, IN.
Antonides, G., Verhoef, P.C. and Aalst, M. (2002),
“Consumer perception and evaluation of waiting time: a
field experiment”, Journal of Consumer Psychology, Vol. 12
No. 3, pp. 193-202.
Baker, J., Levy, M. and Grewal, D. (1992), “An experimental
approach to making retail store environment decisions”,
Journal of Retailing, Vol. 68 No. 4, pp. 445-60.
Bennington, L., Cummane, J. and Conn, P. (2000),
“Customer satisfaction and call centers: an Australian
study”, International Journal of Service Industry Management,
Vol. 11 No. 2, pp. 162-73.
Berry, L.L. (1979), “The time buying consumer”, Journal of
Retailing, Vol. 55 No. 4, pp. 58-69.
Burgers, A., Ruyter, K., Keen, C. and Streukens, S. (2000),
“Customer expectation dimensions of voice to voice service
encounters: a scale development study”, International
Journal of Ser vice Industry Management, Vol. 11 No. 2,
pp. 142-61.
Chebat, J.C., Filiatrault, P. and Zuccaro, C. (1991), “An
experiment on waiting lines: effects of interrupted service
and clients’ participation on perceived time duration,
mood, and perceived quality”, in Chebat, J.C. and
Venkatesan, V. (Eds), Proceedings of the VIIth John-Labatt
Marketing Research Seminar, Time and Consumer Behavior,
UQAM, Montreal, Canada.
Chebat, J.C., Gelinas-Chebat, C. and Filitrault, P. (1993),
“Interactive effects of music and visual cues on time
perception: an application to waiting lines in banks”,
Perceptual and Motor Skills, Vol. 77, pp. 995-1020.
Cowles, D. and Crosby, L.A. (1990), “Consumer acceptance
of interactive media”, The Services Industries Journal, Vol. 10
No. 3, pp. 521-40.
Davis, M. and Heineke, J. (1998), “How disconfirmation,
perception, and actual waiting times impact customer
satisfaction”, International Journal of Service Industry
Management, Vol. 9 No. 1, pp. 64-73.
Davis, M. and Volman, T. (1990), “A framework for relating
waiting time and customer satisfaction in a service
operation”, Journal of Service Marketing, Vol. 4 No. 1,
pp. 61-9.
Dawson, K. (1998), The Call Center Handbook: The Complete
Guide to Starting, Running, and Improving Your Call Center,
Miller-Freeman, New York, NY.
Durrande-Moreau, A. (1999), “Waiting for service: ten years
of empirical research”, International Journal of Ser vice
Industry Management, Vol. 10 No. 2, pp. 171-89.
Evangelist, S., Godwin, B., Johnson, J., Conzola, V., Kizer, R.,
Young-Helou,S.andMetters,R.(2002),“Linking
marketing and operations: an application at Blockbusters,
Inc.”, Journal of Service Research, Vol. 5 No. 2, pp. 91-100.
Feinberg, R.A. and Smith, P. (1989), “Misperception of time
in the sales transaction”, in Srull, T.K. (Ed.), Advances in
Consumer Research, Vol. 16, Association of Consumer
Research, Provo, UT, pp. 56-8.
Feinberg, R., Hokama, L., Kadam, R. and Kim, I. (2002),
“Operational determinants of caller satisfaction in the
banking/financial service call center”, International Journal
of Bank Marketing, Vol. 20 No. 4, pp. 174-80.
Hornik, J. (1984), “Subjective vs objective time measures: a
note on the perception of time in consumer behavior”,
Journal of Consumer Research, Vol. 11 No. 1, pp. 615-8.
Hui, M. and Tse, D. (1996), “What to tell consumers in waits
of different lengths: an integrative model of service
evaluation”, Journal of Marketing, Vol. 60 No. 2, pp. 81-90.
Hui, M.K., Dube, L. and Chebat, J.C. (1997), “The impact
of music on consumer’s reactions to waiting for service”,
Journal of Retailing, Vol. 73 No. 1, pp. 87-104.
Jack, E.P., Bedics, T.A. and McCary, C.E. (2006),
“Operational challenges in the call center industry: a case
study and resource based framework”, Managing Service
Quality, Vol. 16 No. 5, pp. 477-500.
Jones, P. and Peppiatt, E. (1996), “Managing perceptions of
waiting times in service queues”, International Journal of
Service Industr y Management, Vol. 7 No. 5, pp. 47-61.
Karatepe, O.M., Yavas, U., Babakus, E. and Avci, T. (2006),
“Does gender moderate the effects of role stress in
frontline service jobs?”, Journal of Business Research,
Vol. 59 Nos 10/11, pp. 1087-93.
Katz, K., Larson, B. and Larson, R. (1991), “Prescription for
the waiting-in-line blues: entertain, enlighten and engage”,
Sloan Management Review, Vol. 32 No. 2, pp. 44-53.
Kellaris, J. and Kent, R. (1992), “The influence of music in
consumers’ temporal perceptions: does time fly when you’re
having fun”, Journal of Consumer Psychology, Vol. 4 No. 1,
pp. 365-76.
Knott, B.A., Kortum, P., Bushey, R.R. and Bias, R. (2004),
“The effect of music choice and announcement duration on
subjective wait time for call center hold queues”, Proceedings
of the Human Factors and Ergonomics Society 47th Annual
Meeting.
Knott,B.A.,Pasquale,T.,Miller,J.,Mills,S.and
Joseph, K.M. (2003), “Please hold for the next available
agent: the effect of hold queue content on apparent hold
duration”, Proceedings of the Human Factors and Ergonomics
Society 47th Annual Meeting, Denver, CO.
Maister, D. (1985), “The psychology of waiting lines”, in
Czepiel, J., Solomon, M. and Suprenant, C. (Eds), The
Service Encounter, Lexington Books, Lexington, MA.
Closing the gap between perceived and actual waiting times
Anita Whiting and Naveen Donthu
Journal of Services Marketing
Volume 23 · Number 5 · 2009 · 279 – 288
287
Micak, A. and Desmarais, M. (2001), “Benchmarking service
quality performance at business-to-business and business-
to-consumer call centers”, Journal of Business & Industr ial
Marketing, Vol. 16 No. 5, pp. 340-53.
Michalos, A.C. (1985), “Multiple Discrepancies Theory
(MDT)”, Social Indicators Research, Vol. 16, pp. 347-413.
Milliman, R.E. (1986), “The influence of background music
on the behavior of restaurant patrons”, Journal of Consumer
Research, Vol. 13, September, pp. 286-9.
Milliman, R.E. (1982), “Using background music to affect
the behavior of supermarket shoppers”, Journal of
Marketing, Vol. 46 No. 2, pp. 86-91.
Nelson, D.L. and Quick, J.C. (1985), “Professional women:
are distress and disease inevitable”, Academy Management
Review, Vol. 10 No. 2, pp. 206-18.
North, A., Hargreaves, D. and McKendrick, J. (1999),
“Music and on-hold waiting time”, British Journal of
Psychological, Vol. 90 No. 1, pp. 161-4.
Oakes, S. (2000), “The influence of the musicscape within
service environments”, Journal of Ser vice Marketing, Vol. 14
No. 7, pp. 539-56.
Parasuraman, A., Zeithaml, V. and Berry, L. (1985), “A
conceptual model of service quality and its implications for
future research”, Journal of Marketing, Vol. 49 No. 4,
pp. 41-50.
Robinson, G. and Morley, C. (2006), “Call centre
management: responsibilities and performance”,
International Journal of Service Industry Management,
Vol. 17 No. 3, pp. 284-300.
Sunden, A. and Surette, B. (1998), “Gender differences in
the allocation of assets in retirement savings plans”,
American Economic Review, Vol. 88 No. 2, pp. 207-12.
Tansik, D. and Routhieaux, R. (1999), “Customer stress
relaxation: the impact of music in a hospital waiting room”,
International Journal of Service Industr y Management, Vol. 10
No. 1, pp. 68-81.
Tom, G., Burns, M. and Zeng, T. (1997), “Your life on hold:
the effect of telephone waiting time on consumer
perception”, Journal of Direct Marketing, Vol. 11 No. 3,
pp. 25-31.
Unzicker, D. (1999), “The psychology of being put on hold:
an exploratory study of service quality”, Psychology and
Marketing, Vol. 16 No. 4, pp. 327-50.
Whiting, A. and Donthu, N. (2006), “Managing voice-to-
voice encounters: reducing the agony of being put on hold”,
Journal of Ser vice Research, Vol. 8 No. 3, pp. 234-44.
Yan, R. and Lotz, S. (2006), “The waiting game: the role of
predicted value, wait disconfirmation, and providers’
actions in consumers’ service evaluations”, in Kardes, F.
and Mita, S. (Eds), Advances in Consumer Research, Vol. 22,
Association for Consumer Research, Provo, UT, pp. 412-8.
Zakay, D. (1989), “Subjective time and attentional resource
allocation: an integrated model of time and estimation”, in
Levin, I. and Zakay, D. (Eds), Time and Human Cognition,
North Holland, Amsterdam, pp. 365-98.
Zakay, D. and Hornik, J. (1994), “How much time did you
wait in line? A time perception perspective”, paper
presented at the 6th John-Labatt Marketing Research
Seminar, University of Quebec in Montreal, Montreal.
About the authors
Anita Whiting is an Assistant Professor of Marketing at
Clayton State University. She received her PhD in Marketing
from Georgia State University. Dr Whiting also holds an
MBA from the Georgia Institute of Technology. She has
published in Journal of Ser vice Research,Journal of Ser vices
Marketing, and International Business: Research, Teaching, and
Practice. Services marketing, retailing, frontline service
employees, and atmospherics are her major research
interests. Anita Whiting is the corresponding author and can
be contacted at: AWhiting@clayton.edu
Naveen Donthu (PhD University of Texas at Austin) is the
Katherine S. Bernhardt Research Professor of Marketing at
Georgia State University. His expertise is in the areas of
marketing research methodology, consumer research,
advertising, electronic business and services marketing. He
has published over 65 articles in peer reviewed journals,
including Journal of Marketing,Journal of Marketing Research,
Journal of Consumer Research,Marketing Science and
Management Science.
Closing the gap between perceived and actual waiting times
Anita Whiting and Naveen Donthu
Journal of Services Marketing
Volume 23 · Number 5 · 2009 · 279 – 288
288
To purchase reprints of this article please e-mail: reprints@emeraldinsight.com
Or visit our web site for further details: www.emeraldinsight.com/reprints
Reproducedwithpermissionofthecopyrightowner.Furtherreproductionprohibitedwithoutpermission.
