J. Service Science & Management, 2009, 2: 96-106
Published Online June 2009 in SciRes (www.SciRP.org/journal/jssm)
Applying the Service-Profit Chain to Internet Service
Businesses
Jin-Woo Kim1,*, Michael Richarme1,2,#
1Department of Marketing, College of Business, The University of Texas at Arlington, Arlington, Texas, USA; 2Decision Analyst,
Arlington, Texas, USA.
Received March 2nd, 2009; revised April 7th, 2009; accepted April 28th, 2009.
ABSTRACT
Service-profit chain (SPC) is a powerful tool for evaluating the relationship between the degree of service effort and
profit. This framework has been empirically tested in the banking, hospital, retail, and other brick and mortar service
verticals. This paper extends the applicability of SPC to Internet-based businesses. The paper also investigates the
moderating roles of customer satisfaction and advertising spending on the service operation efficiency (SOE) and profit
relationship. Analysis of U.S. Internet service providers during a seven year period from 2000 to 2006 indicates that
service operations efficiency is positively associated with a firm’s profit. However, the customer satisfaction and adver-
tising spending constructs are negative moderators of the relationship between service operations efficiency and profit.
Keywords: service-profit chain, service operations efficiency, customer satisfaction, data envelopment analysis
1. Introduction
Service-profit chain (SPC) is a useful framework in ex-
plaining the relationship between service operation and a
firm’s profit. From the SPC literature, a service firm’s
profit can be improved by effective service operational
input through customer perception and actual behavior
[1]. SPC has been empirically tested with bank service
[2,3,4], retail chain [5], and business market [6] verticals.
SPC can provide a useful integrated foundation for
measuring service improvement effort as it relates to
ultimate management goals.
Given that prior empirical testing of this framework has
focused on brick and mortar services, there is an opportu-
nity to extend the framework to Internet-based businesses.
Moreover, the moderating roles of customer satisfaction
and advertising spending have not been fully addressed in
the SPC model. Prior research in the field of SPC omits
the roles of strategic levers such as customer satisfaction
and advertising expenditures in transferring service im-
provement effort to profitability. Little attention has been
paid to how customer satisfaction and advertising expen-
ditures moderate the relationship between the degree of
service effort and the firm’s profit. Motivated by this re-
search gap, the study investigates these moderating vari-
ables using U.S. Internet service business data.
For this paper, a research framework is developed
based on extant theory. Next, a methodology description
section addresses the research setting and data available
for the analysis. U.S. Internet service business data re-
garding service operations efficiency, customer satisfac-
tion, advertising spending and real profit are utilized. For
this analysis, Data Envelopment Analysis (DEA) is util-
ized to determine the service operations efficiency by
integrating the operations input and outcome. Then, re-
gression analysis is conducted to shows that Internet ser-
vice operations efficiency is related to real profit. In the
analysis and discussion section, the results are inter-
preted and some theoretical and managerial insights are
presented.
2. Research Framework
As mentioned earlier, prior research tends to focus on the
service quality assessment of individual consumers using
traditional brick and mortar retail and banking services.
In this study, Internet service operations efficiency is
considered as an independent variable and profit as the
dependent variable. Two possible moderators, customer
satisfaction and advertising expenditures, are added to
the SPC framework as shown in Figure 1.

JIN-WOO KIM, MICHAEL RICHARME97
Figure 1. Research framework
2.1 Impact of Service Operations Efficiency on
Firm’s Profit
Service operations efficiency is expected to lead to an
increase in a firm’s profit. There is often a time lag for
this relationship to become evident [7]. Linking the con-
structs of service operations, employee assessments and
customer assessments to profit was proposed from the
empirical evidence derived from 20 large service or-
ganizations [1]. A modified version of the service profit
chain was employed to take into account the employee
and customer assessments relationship to profit [8].
Loveman and Heskett (2008) updated the SPC model
by adopting employee capability and external service
quality. Basically, there are four principal components of
the service profit chain: internal service quality, external
service quality and service value, target market of cus-
tomers, and financial performance [9]. In the same man-
ner, the SPC framework can be applied to Internet busi-
nesses.
2.2 The Moderating Role of Customer Satisfaction
SPC is a valuable framework that offers insights to make
it possible observe service related issues in an integrative
perspective [4]. The comprehensive SPC framework
models processes and constructs between service quality
and profit, indicating that operating strategy and service
delivery system could lead to employee satisfaction, cus-
tomer satisfaction and customer loyalty [9]. In prior re-
search, customer satisfaction is expected to moderate the
relationship between corporate social responsibility and
market value. This result can be attributed to customer
satisfaction, which makes it possible to convert corporate
social responsibility to financial value [10].
In this sense, service improvement effort will have an
impact on profit after it interacts with customer satisfac-
tion. When a firm with high customer satisfaction em-
ploys a strategy to improve service quality, net income
will increase. If consumer does not notice the service
effort, changes in profit will be stable or only marginally
improved. However, if the customer recognizes the in-
creased SOE, the firm can realize the more profit. There-
fore, customer satisfaction will play a moderating role
between service quality effort and firm’s profit. Service
improvement effort without customer recognition of the
improvement will not be effective.
2.3 The Moderating Role of Advertising Expen-
ditures
Advertising has often been treated as an input variable of
marketing credibility and advertising efficiency [11,12,
13,14]. In this expanded SPC model, advertising expen-
ditures will be examined as a moderator of the relation-
ship between service efforts and profit. Information on
service improvement can be delivered to customers by
advertising and other promotion activities. Therefore,
advertising expenditures are expected to positively mod-
erate the relationship between service effort and profit.
This proposition is based on advertising effects over
times. Even though advertising effects will increase to an
initial peak but decrease due to time passage, advertising
amount and cost will be a moderating effect regardless of
this factor. All the advertising does not have a positive
impact on customer satisfaction and profit. Advertising
will increase customer awareness of the level of service
effort generated by the firm, and ultimately, the level of
the firm’s profit [15].
3. Research Setting and Data Description
The analysis for this application consists of two steps as
seen in the Figure 2. Multiple input and outputs are se-

JIN-WOO KIM, MICHAEL RICHARME 98
lected to represent the service operations efficiency con-
struct. After defining these inputs and outputs, SOE is
estimated using Data Envelopment Analysis (DEA).
This technique is one of the most popular management
science tools used to determine a firm’s efficiency level
from multiple inputs and multiple outputs. A quick back-
ground of the DEA technique is provided in the Appen-
dix. From this first step, the SOE estimates will be the
regressor and the firm’s profit will be the response in the
regression equations.
3.1 Measuring Service Operations Efficiency
The concept of service operations efficiency can be a very
useful indicator that focuses on “Do things in the right
manner” rather than “Do the right things.” Since the early
1990’s, many researchers have started to study service
quality and assessment from a macro-economic point of
view. The DEA technique has become a useful manage-
ment science tool, in which multiple inputs and multiple
outputs can be involved simultaneously, whereas regres-
sion cannot include multiple responses [16]. In particular,
DEA is a tool for evaluating marketing’s credibility [13],
retail productivity [17,18], sales force performance
[19,20], and advertising efficiency [11,12,14]. Also, DEA
is used to compute efficiency of supply chain processes
[21,22] and operations efficiency [23].
The efficiency of one DMU1 can be obtained as a solu-
tion to maximize its efficiency subject to the efficiency of
all DMUs being less than or equal to 1. The solution pro-
duces the weights which are most favorable to the DMU1
and provides a measure of efficiency for that DMU1. The
solution produces the weights which are most favorable to
the DMU1 and provides a measure of efficiency for the
DMU1. The algebraic model is as follows:



i
ii
r
rr
DMUInputweight
DMUOutputweight
hMax
1
1
0
subject to 1


i
jii
r
jrr
DMUInputweight
DMUOutputweight for each DMUj,


ir weight,weight
Several studies have examined efficiency measures of
on-line businesses by using the DEA methodology. The
efficiency of service delivery processes was computed
based on a survey of 135 large U.S. retail banks hold-
ing more than 75% of the total assets in the industry in
1994. Activity time (minutes), technology, and function-
ality were regarded as multiple inputs and check cycle
time (day), ATM cycle time (day), and customer time
(minutes) were considered as multiple outputs [24].
The efficiency of several search engines was investi-
gated by using a query search as a production process
and compared to each other by employing a simple data
envelopment analysis model. The seven primary search
engines utilized were Alta Vista, Excite, Lycos, Infoseek,
Open Text, Hotbed, and Web Crawler. Intermediate in-
puts included the number of pages, update freshness,
time for data retrieval, and feature on fourteen informa-
tion retrieval items. Output variables included precision
and circulation. A survey conducted by Internet media
companies provided fifteen queries that are submitted to
each engine and then the first twenty returned results
were considered in calculating the values [25].
Define SOE
DEA phase
Choose
ts
Choose
tsmultiple inpumultiple outpu
Estimate SOE level
usin
g
DEA
Figure 2. Research process
Result of SOE estimation
Regression p
rofitPredict P
hase

JIN-WOO KIM, MICHAEL RICHARME99
The concept of ‘customer efficiency’ was established
and linked to e-business management practices. For this
purpose, purchase activity time and non-purchase activ-
ity time were used as multiple inputs and number of in-
formational transactions, number of service transactions,
and number of purchases were considered as multiple
outputs for computation of the efficiency construct [26].
Given this, service operations efficiency is measured
with two inputs and two outputs in this study. From Re-
source Based View (RBV) theory, total assets and num-
ber of employees are considered as inputs. Multiple out-
puts include total number of visit and total sales. Sup-
pose that the Amazon’s total asset and number of em-
ployees are 5 and 6, and total number of site visits and
total sales are 7 and 8 while eBay’s input values are 5
and 8, and output values are 3 and 4. The efficiency of
Amazon (h0) can be obtained by solving the following
model:
0
78
56
asset employee
visit sales
weight weight
Max hweight weight


subject to
78
1
56
asset employee
visit sales
weightweight
weight weight


 (Amazon),
34
1
58
asset employee
visit sales
weight weight
weight weight


 (eBay),
……………. for remaining DMUs
and
 
salesvisitemployeeasset weightweightweight,weight
If h0 is equal to 1 then Amazon is efficient relative to
other DMUs but if h0 is less than 1, some other DMU is
more efficient than Amazon.
3.2 Predicting Firm’s Profit
After SOE is computed, several multivariate regression
analyses were conducted to grasp the effects of SOE and
other factors on firm’s profit as the following:
Model 1 Profitt = a + b1 SOEt + b2 ADt + b3 CSt
Model 2 Profitt = a + b1 SOEt + b2 ADt + b3 CSt
+ b4 ASSETt + b5 SALESt
+ b6 EMPLt + b7 SOEt*ADt
+ b8 SOEt*CSt
Model 3 Profitt = a + b1 SOEt + b2 ASSETt
+ b3 SALESt + b4 EMPLt
+ b5 SOEt*ADt + b6 SOEt*CSt
+ b7 YEAR_1+ b8 YEAR_2
+ b9 YEAR_3 + b10 YEAR_4
+ b11 YEAR_5 + b12 YEAR_6
+ b13 INDUSTRY_1
+ b14 INDUSTRY_2
+ b15 INDUSTRY_3
Model 4 Profitt = a + b1 SOEt + b2 ADt + b3 CSt
+ b4 ASSETt + b5 SALESt
+ b6 EMPLt + b7 SOEt*ADt
+ b8 SOEt*CSt+ b9 YEAR_1
+ b10 YEAR_2 + b11 YEAR_3
+ b12 YEAR_4 + b13 YEAR_5
+ b14 YEAR_6
+ b15 INDUSTRY_1
+ b16 INDUSTRY_2
+ b17 INDUSTRY_3
One year-lagged impact of SOE on profit was ana-
lyzed using multivariate regression as following:
Model 5 Profitt+1 = a + b1 SOEt + b2 ADt + b3 CSt
Model 6 Profitt+1 = a + b1 SOEt + b2 ADt + b3 CSt
+ b4 ASSETt + b5 SALESt
+ b6 EMPLt + b7 SOEt*ADt
+ b8 SOEt*CSt
Model 7 Profitt+1 = a + b1SOEt + b2 ADt
+ b3CSt + b4 ASSETt
+ b5 SALESt+ b6 EMPLt
+ b7 SOEt*ADt + b8 SOEt*CSt
+ b9 YEAR_1 + b10 YEAR_2
+ b11 YEAR_3+ b12 YEAR_4
+ b13 YEAR_5 + b14 YEAR_6
+ b15 INDUSTRY_1
+ b16 INDUSTRY_2
+ b17 INDUSTRY_3
3.3 Data Description
This study obtains data on total assets, total number of
employees, total number of visits to web sites, total sales,
customer satisfaction levels, advertising expenditure
sand profitability for 70 observations of U.S. Inter-
net-based business firms. For example, Amazon, Yahoo,
Travelocity and NewYorkTimes.com are included to
represent retail, portal, travel and news service firms,
respectively. The data are obtained from various secon-
dary sources. The multiple inputs and outputs chosen in
this study are shown in Table 1.
For computation of SOE, total annual assets and total

JIN-WOO KIM, MICHAEL RICHARME 100
Table 1. Measures and operationalization
Variable Measure Operationalization Data Source
Independent variable SOE Service operations efficiency Calculated by DEA
ASSET Annual total assets (10 Thousand dollars) COMPUSTAT North America
(http://wrds.wharton.upenn.edu/ds/comp/inda/)
DEA Inputs
EMPL Total number of employee (persons) COMPUSTAT North America
VISIT Total number of visit to web site MediaMetrix:ComScore
(http://www.comscore.com)
DEA Outputs
SALES Annual total sales (10 Thousand dollars) COMPUSTAT North America
AD Advertising expenditure (10 Thousand dollars)COMPUSTAT North America
Moderator
CS Customer satisfaction score ACSI (http://www.theacsi.org/index)
Dependent Variable PROFIT Annual Profit (10 Thousand dollars) COMPUSTAT North America
YEAR Six annual dummies 2000 ~ 2006
Control Variables
INDUSTRY Three industry dummies News, Portal, Travel, Retail
number of employees are considered as DEA inputs and
total sales and total number of web visit are taken into
account as DEA outputs. Total number of web visits, one
of the DEA outputs, utilizes MediaMetrix ComScore
results for the Top 50 Web and Digital Media Properties.
A customer satisfaction score is provided by the Ameri-
can Customer Satisfaction Index. To capture the effect of
year and industry types, six and three dummy variables
are utilized by the regression equation, respectively. The
data were examined for completeness and outliers. After
data cleansing, data from 2000 to 2006 were merged to
provide the final dataset with 70 observations.
4. Analysis & Discussion
The service operations efficiency variable is computed
by DEA, and then a regression procedure is conducted to
test the main effect of service operations efficiency on
profit and the moderating effects of customer satisfaction
and advertising expenditures on the service-profit chain.
Table 2 displays the correlation matrix between variables
and shows that some variables are positively or nega-
tively related to each other.
4.1 Results of SOE
Unlike regression, DEA does not impose any particular
functional form on the data, creating a more flexible
piecewise linear function. Also, unlike total factor pro-
ductivity indexes, DEA gives each of the observations its
own set of weights. This efficient frontier line provides a
more realistic benchmark because the decision-making
units (DMUs) are compared to best practices rather than
to average performance [27,28].
Table 2. Correlation matrix
SOE CS ASSET SALES PROFIT EMPL AD VISIT
SOE 1
CS .375 1
ASSET -.351 -.362 1
SALES -.352 -.311 .953 1
PROFIT -.275 -.331 .901 .915 1
EMPL -.485 -.401 .749 .829 .600 1
AD -.413 -.408 .879 .824 .644 .833 1
VISIT .123 .046 .526 .490 .594 .112 .312 1

JIN-WOO KIM, MICHAEL RICHARME101
Table 3. Input oriented SOE
DMU Name Input-Oriented EfficiencyDMU Name Input-Oriented Efficiency
2000 Amazon.com 0.21098 2003 NYTimes.com 0.13394
2000 barnesandnoble.com 0.19039 2003 Orbitz Worldwide 1.00000
2000 eBay 0.21376 2003 Yahoo 0.50792
2000 Ask.com 0.06937 2004 Amazon.com 0.33998
2000 Microsoft 0.17061 2004 ABCNEWS.com 0.09009
2000 priceline.com 1.00000 2004 eBay 0.23426
2000 CNN.com 0.13340 2004 Expedia 0.16024
2000 Travelocity.com 0.32821 2004 USATODAY.com 0.07569
2000 Yahoo! Inc. 0.42136 2004 Google 0.66749
2001 Amazon.com 0.34563 2004 Ask.com 0.06921
2001 ABCNEWS.com 0.09131 2004 Microsoft News 0.18779
2001 eBay 0.26161 2004 CNN.com 0.14406
2001 USATODAY.com 0.07650 2004 Yahoo 0.41334
2001 Ask.com 0.08376 2005 Amazon.com 0.36273
2001 Microsoft News 0.15443 2005 ABCNEWS.com 0.09489
2001 NYTimes.com 0.13850 2005 eBay 0.18532
2001 Travelocity.com 0.43571 2005 Expedia 0.15167
2001 Yahoo! Inc. 0.56876 2005 USATODAY.com 0.07622
2002 Amazon.com 0.35086 2005 Google 0.52688
2002 barnesandnoble.com 0.75451 2005 Ask.com 0.07423
2002 ABCNEWS.com 0.07992 2005 Microsoft News 0.18954
2002 eBay 0.23832 2005 NYTimes.com 0.15993
2002 Expedia 0.26294 2005 CNN.com 0.14439
2002 USATODAY.com 0.07385 2005 Yahoo 0.37270
2002 Ask.com 0.05788 2006 Amazon.com 0.38766
2002 Microsoft News 0.16322 2006 ABCNEWS.com 0.09023
2002 NYTimes.com 0.13380 2006 eBay 0.21310
2002 Yahoo 0.59576 2006 Expedia 0.13320
2003 Amazon.com 0.43322 2006 USATODAY.com 0.07819
2003 ABCNEWS.com 0.08548 2006 Google 0.39674
2003 eBay. 0.30988 2006 Ask.com 0.11476
2003 USATODAY.com 0.07206 2006 Microsoft News 0.18124
2003 Google 1.00000 2006 NYTimes.com 0.21718
2003 Ask.com 0.07155 2006 CNN.com 0.13863
2003 Microsoft News 0.17006 2006 Yahoo 0.34974
The DEA results were computed by DEA Excel
Solver to obtain overall efficiency scores for U.S. Inter-
net service providers from 2000 to 2006. DEA analysis
provides the service operations efficiency for all compa-
nies based on each company’s combination of inputs and
outputs compared to those of others in the sample. For
calculation of the overall SOE of each company, seven
years data are combined as one dataset. Given this, the

JIN-WOO KIM, MICHAEL RICHARME 102
Table 4. Regression result (DV=Profitt)
Model 1 Model 2 Model 3 Model 4
Beta P-value Beta P-value Beta P-value Beta P-value
(Constant) 19.703 .041 1.685 0.763 0.418 0.414 2.316 0.71
SOE -.143 .965 37.149 0.092 29.509 0.05* 24.707 0.25
SOE*CS -4.216 0.101 -3.44 0.046 -2.867 0.256
SOE*AD -2.591 0.101 -2.333 0.051 -2.149 0.152
AD 2.309 .000** -0.707 0.098 0.18 0.69
CS -.231 .067 -0.018 0.814 -0.025 0.76
Asset 0.013 0.656 -0.01 0.501 -0.02 0.455
Sales 0.556 0** 0.41 0** 0.423 0**
Employee -0.857 0** -1.218 0** -1.275 0**
Year1 0.662 0.297 0.646 0.327
Year2 0.701 0.254 0.713 0.269
Year3 1.145 0.105 1.168 0.122
Year4 0.513 0.464 0.536 0.457
Year5 1.403 0.07 1.404 0.077
Year6 1.339 0.099 1.343 0.106
Industry1 7.372 0** 7.628 0**
Industry2 0.23 0.615 0.231 0.716
Industry3 -0.395 0.552 -0.403 0.565
Adjusted R2 0.415 0.956 0.973 0.97
* significant at 0.05
** significant at 0.01
same company from 2000 to 2006 is regarded as differ-
ent DMUs. An input-oriented constant return to scale
(CRS) DEA model was used to compute service opera-
tion efficiency.
Table 3 shows that the least efficient company is-
Ask.com in the year 2002 with a 0.05788 SOE score.
Three DMUs, priceline.com in the year 2000, Orbitz
Worldwide in the year 2003, and Google in the year
2003 are identified as most efficient with an SOE score
of 1 based on input-oriented CRS DEA method.
4.2 Results of Regression
Regression considering only three factors including SOE,
customer satisfaction, and advertising expenditure is
conducted to provide a preliminary result. The regression
result shows that only ADt is positively associated with
profitt (p<0.001). Then, SOE impacts on profit are ana-
lyzed by the same regression method using control vari-
ables including each year and the type of Internet service.
Table 4 shows the results of regression taking parallel
data into account. As stated, the service operations effi-
ciency appears to have a positive impact on the firm’s
profit. In particular, Model 3, one of four regressions
using SOE as an independent variable and profit as the
dependent variable, results in a main effect of SOE on
profit (b=29.509, p<0.05), suggesting that SOEt has a
significantly positive influence on profitt.
As tested in the previous studies, customer satisfac-
tion can be an indicator to predict the changes in a firm’s
profit. Many empirical studies have examined this rela-
tionship [29,30,31]. In the long term, customer satis-
faction is an ultimate goal of the firm and profit can be
viewed as a satisfaction input. However, the significant
interaction between SOE and customer satisfaction is
observed in Model 3 (b= -3.44, p<0.05) but the sign of
effect is not positive. That means there might be a
negative moderating effect of customer satisfaction in
the service profit chain.
In addition, the research framework posits that adver-
tising expenditure moderates the relationship between
SOE and profit. Result shows that the relationship be-
tween SOE and profit is moderated by advertising ex-

JIN-WOO KIM, MICHAEL RICHARME103
penditure but that the moderation effect is unexpectedly
negative (b= -2.333, p<0.1). Total sales as a control vari-
able have a significantly positive impact on profit in all
models. In contrast, the number of employee is nega-
tively associated with profit. Industry 1 is positively re-
lated to profit but other control variables are not.
Next, this paper addresses the relationship between
SOE and subsequent period profit. If the firms maintain a
high level of service efficiency, the firms are expected to
realize more profits. However, the previous service op-
erations efficiency level does not influence subsequent
period profit as seen in Table 5, indicating that there is
no lagged effect of service operations efficiency. No
interaction effects are found in Model 6 and Model 7.
Only similar effects of sales and employees are observed
as in Model 3.
5. Conclusions & Future Research
This paper attempts to examine whether or not SPC is
applicable to Internet business models. It also investi-
gates the moderating effect of customer satisfaction and
advertising expenditures in the service profit chain. The
research results are summarized in Table 6. In summary,
a firm’s service improvement efforts are expected to fa-
cilitate positive evaluation of the firm by its customers.
Service operations efficiency will create more customer
satisfaction, demonstrating if a firm employs a service
operations strategy to improve the service quality, more
profit can be realized.
Given these results, SOE is one of the best ways to
improve a firm’s short-term profit, but the lagged impact
of SOE on subsequent period profits is not shown by this
research. As an example, the news and information
category has a relatively high contribution to profit. This
can be attributed to unique service delivery characteris-
tics of Internet news services.
The research verifies that service operations efficiency
has a relationship with a firm’s profit but fails to show a
positive moderating effect of customer satisfaction and
advertising expenditures in SPC. Unexpectedly, negative
moderating effects of those two variables are observed in
the regression results. In this sense, if the Internet service
providers want to realize more profits, they should place
emphasis on service operations efficiency rather than
customer satisfaction and advertising spending. As this
Table 5. Regression result (DV=Profitt+1)
Model 5 Model 6 Model 7
DV=Profitt+1 Beta P-value Beta P-value Beta P-value
(Constant) 20.037 .061 3.945 0.542 6.882 0.311
SOE -.328 .925 21.859 0.377 -5.464 0.821
SOE * CS -2.432 0.399 0.659 0.816
SOE * AD -1.423 0.425 0.141 0.934
AD 2.339 .001** -0.908 0.081 -0.179 0.752
CS -.231 .097 -0.049 0.566 -0.086 0.325
Asset 0.007 0.871 0.006 0.879
Sales 0.619 0** 0.316 0.01**
Employee -0.883 0** -1.217 0**
Year1 0.231 0.736
Year2 0.536 0.448
Year3 0.334 0.694
Year4 1.129 0.17
Year5 1.652 0.068
Industry1 9.809 0
Industry2 0.223 0.758
Industry3 -0.659 0.418
Adjusted R2 0.389 0.952 0.971
* significant at 0.05
** significant at 0.01

JIN-WOO KIM, MICHAEL RICHARME 104
moderating effect of advertising expense and customer
satisfaction on SPC is not supported, other possible
moderators should be identified for future research [32].
Theoretically, this paper contributes to both operations
management and marketing scholars by helping focus
attention to Internet service operations efficiency,
profit, and customer satisfaction. From a practitioner
perspective, management can recognize the importance
of service operations efficiency in their business prac-
tices as a key driver of profitability. One weakness of
this paper is the small sample size, which somewhat lim-
its the generalizable nature of the findings. As more data
become available, this limitation can be mitigated.
Table 6. Result summary
IV DV Expected Result Significant
SOE(t) Profit(t) + + p<0.1
SOE(t) Profit(t+1) + +/- n.s.
SOE(t)*CS(t) Profit(t) + - p<0.05
Profit(t+1) + +/- n.s.
SOE(t)*AD(t) Profit(t) + - p<0.1
Profit(t+1) + +/- n.s.
REFERENCES
[1] L. J. Heskett, T. O. Jones, G. W. Loveman, W. E. Sasser,
Jr., and L. Schlesinger , “Putting the service-profit chain
to work,” Harvard Business Review, 72(2), pp. 164-170,
1994.
[2] G. W. Loveman, “Employee satisfaction, customer loyalty,
and financial performance: An empirical examination of
the service profit chain in retail banking,” Journal of Ser-
vice Research, 1(1), pp. 18-31, 1998.
[3] A. Soteriou and S. A. Zenios, “Operations, quality, and
profitability in the provision of banking services,” Man-
agement Science, 45(9), pp. 1221-1238, 1999.
[4] W. A. Kamakura, V. Mittal, F. de Rosa, and J. A. Mazzon,
“Assessing the service-profit chain,” Marketing Science,
21(3), pp. 294-317, 2002.
[5] M. Pritchard and R. Silvestro, “Applying the service profit
chain to analyse retail performance: The case of the man-
agerial strait-jacket?” International Journal of Service In-
dustry Management, 16(4), pp. 337-356, 2005.
[6] D. Bowman and D. Narayandas, “Linking customer man-
agement effort to customer profitability in business mar-
ket,” Journal of Marketing Research, 41(4), pp. 433-447,
2004.
[7] R. T. Rust, A. J. Zahorik, and T. L. Keiningham, “Return
on Quality (ROQ): Making service quality financially ac-
countable,” Journal of Marketing, 59(2), pp. 58-70, 1995.
[8] A. J. Rucci, S. P. Kirn, and R. T. Quinn, “The em-
ploy-customer-profit chain at Sears,” Harvard Business
Review, 76(1), pp. 82-97, 1998.
[9] J. L. Heskett, T. O. Jones, G. W. Loveman, W. E. Sasser,
Jr., and L. Schlesinger, “Putting the service-profit chain to
work,” Harvard Business Review, 86(7/8), pp. 118-129,
2008.
[10] X. M. Luo and C. B. Bhattacharya, “Corporate social
responsibility, customer satisfaction, and market value,”
Journal of Marketing, 70(4), pp. 1-18, 2006.
[11] X. M. Luo and N. Donthu, “Benchmarking advertising
efficiency,” Journal of Advertising Research, 41(6), pp.
7-18, 2001.
[12] X. M. Luo and N. Donthu, “Assessing advertising media
spending inefficiencies in generating sales,” Journal of
Business Research, 58(1), pp. 28-36, 2005.
[13] X. M. Luo and N. Donthu, “Marketing’s credibility: A
longitudinal investigation of marketing communication
productivity and shareholder value,” Journal of Marketing,
70(4), pp. 70-91, 2006.
[14] J. Mahajan, “A data envelopment analytic model for as-
sessing the relative efficiency of the selling function,”
European Journal of Operational Research, 53(2), pp.
189-205, 1991.
[15] G. E. Fruchter, “The many-player advertising game,” Ma-
nagement Science, 45(11), pp. 1609-1611, 1997.
[16] X. M. Luo, “Data envelopment analysis: A management
science tool for scientific marketing research,” Journal of
Marketing Research, 42(3), Book Review, pp. 113-116,
2004.
[17] N. Donthu and B. Yoo, “Retail productivity assessment
using data envelopment analysis,” Journal of Retailing,
74(1), pp. 89-105, 1998.
[18] W. A. Kamakura, T. Lenartowicz, and B. T. Ratchfrord,
“Productivity assessment of multiple retail outlets,” Jour-
nal of Retailing, 72(4), pp. 333-356, 1996.
[19] J. S. Boles, N. Donthu, and R. Lohtia, “Salesperson evalu-
ation using relative performance efficiency: The applica-
tion of data envelopment analysis,” Journal of Personal
Selling & Sales Management, 15(3), pp. 31-49, 1995.

JIN-WOO KIM, MICHAEL RICHARME105
[20] B. K. Pilling, N. Donthu, and S. Henson, “Accounting for
the impact of territory characteristics on sales performance:
Relative efficiency as a measure of salesperson perform-
ance,” Journal of Personal Selling and Sales Management,
19(2), pp. 35-45, 1999.
[21] G. Reiner and P. Hofmann, “Efficiency analysis of supply
chain processes,” International Journal of Production Re-
search, 44(23), pp. 5065-5078, 2006.
[22] Y. Chen, L. Liang, and F. Yang, “A DEA game model
approach to supply chain efficiency,” Annals of Opera-
tions Research, 145(1), pp. 5-13, 2006.
[23] A. D. Ross and C. Droge, “An analysis of operations
efficiency in large-scale distribution systems,” Journal of
Operations Management, 21(6), pp. 673-688, 2004.
[24] F. X. Frei and P. T. Harker, “Measuring the efficiency of
service delivery processes: An application for retail bank-
ing,” Journal of Service Research, 1(4), pp. 300-312, 1999.
[25] D. Lien and Y. Peng, “Measuring the efficiency of search
engines: An application of data envelopment analysis,”
Applied Economics, 31(12), pp. 1581-1587, 1999.
[26] M. Xue and P. T. Harker, “Customer efficiency: Concept
and its impact on e-business management,” Journal of
Service Research, 4(4), pp. 253-267, 2002.
[27] J. Zhu, “Quantitative models for performance evaluations
and benchmarking: Data envelopment analysis with
spreadsheets and DEA EXCEL solver,” Norwell, MA:
Kluwer Academic Publishers Group, 2003.
[28] X. M. Luo and N. Donthu, “Benchmarking advertising
efficiency,” Journal of Advertising Research, 41(6), pp.
7-18, 2001.
[29] X. M. Luo and C. Homburg, “Neglected outcomes of
customer satisfaction,” Journal of Marketing, 71(2), pp.
133-149, 2007.
[30] V. Mittal, E. W. Anderson, P. Tadikamalla, and A. Sayrak,
“Dual emphasis and the long-term financial impact of cus-
tomer satisfaction,” Marketing Science, 24(4), pp. 544-
555, 2005.
[31] B
. Cooil, T. L. Keininham, L. Aksoy, and M. Hsu, “A
longitudinal analysis of customer satisfaction and share of
wallet: Investigating the moderating effect of customer
characteristics,” Journal of Marketing, 71(1), pp. 67-83,
2007.
[32] X. Y. Liu, “Study on service profit chain: Review and
prospect,” Services Systems and Services Management,
1(1), pp. 429-434, 2005.
[33] M. J. Farrell, “The measurement of productive effi-
ciency,” Journal of the Royal Statistical Society Series A,
120(3), pp. 253-281, 1957.
[34] A. Charnes, W. Cooper, and E. Rhodes, “Measuring the
efficiency of decision making units,” European Journal of
Operational Research, 3(6), pp. 429-444, 1978.

JIN-WOO KIM, MICHAEL RICHARME 106
Appendix. Background and Applications of DEA
Building on the ideas of Farrell (1957), the seminal work
“Measuring the Efficiency of Decision-Making Units”
by Charnes, Cooper & Rhodes (1978) applies linear pro-
gramming to estimate an empirical production technol-
ogy frontier for the first time [33, 34]. Since then, there
have been a large number of books and journal articles
written on DEA or applying DEA to various sets of
problems. In addition to comparing efficiency across
DMUs within an organization, DEA has been used to
compare efficiency across firms. There are several types
of DEA methodologies, with the most basic being CCR
based on Charnes, Cooper & Rhoades. However there
are also more sophisticated DEA methodologies which
address varying returns to scale, either CRS (constant
returns to scale) or VRS (variable returns to scale) [27].
Grounded in microeconomic theory, DEA efficiency
provides guidelines and benchmarks for both public and
private enterprises to achieve maximized desirable ends
at minimized costs. DEA measurement has been used to
evaluate and compare educational departments (schools,
colleges and universities), health care (hospitals, clinics)
prisons, agricultural production, banking, armed forces,
sports, market research, transportation (highway mainte-
nance), courts, benchmarking, index number construc-
tion and many other applications.
In addition, DEA optimizes on each individual obser-
vation and provides a ratio score to indicate the relative
efficiency performance against the set of Pareto-efficient
frontiers. An efficient observation is one for which no
other observations, or linear combination of observations
in the sample, generate as much as or more outputs given
the level of inputs (or consume as much as or less inputs
given the level of outputs). DEA is best characterized by
the following [16]:
 A focus on individual observations in contrast to
populations average
 Production of single aggregate measure for each
decision making unit (DMU) in terms of its input
factors (independent variables) to produce desired
outputs (dependent variables)
 Simultaneous use of multiple outputs and multiple
inputs, where each is stated in different units of
measurement
 Ability to adjust to exogenous variables
 Ability to incorporate categorical (dummy) vari-
ables
 No required specification or knowledge of a priori
weights or prices for the inputs or outputs and value
free
 No restrictions on the functional form of the pro-
duction relationship
 Ability to accommodate judgment when desired
 Production of specific estimates for desired changes
in inputs and/or outputs for projecting DMUs below
the efficient frontier onto the efficient frontier
 Pareto optimal
 A focus on the revealed best-practice frontier rather
than on the central tendency properties of frontier
 Satisfaction of strict equity criteria in the relative
evaluation of each DMU
 Relationship with performance evaluation and
benchmarking