=Paper=
{{Paper
|id=Vol-2220/06_CONFWS18_paper_1
|storemode=property
|title=The Effect of Default Options on Consumer Decisions in the Product Configuration Process
|pdfUrl=https://ceur-ws.org/Vol-2220/06_CONFWS18_paper_1.pdf
|volume=Vol-2220
|authors=Yue Wang,Daniel Yiu-Wing Mo
|dblpUrl=https://dblp.org/rec/conf/confws/WangM18
}}
==The Effect of Default Options on Consumer Decisions in the Product Configuration Process==
https://ceur-ws.org/Vol-2220/06_CONFWS18_paper_1.pdf
The Effect of Default Options on Consumer Decisions in
the Product Configuration Process
Yue Wang1* and Daniel Yiu-Wing Mo1
Abstract. 1 Product configurators have been accepted as an want quickly and with less burden of cognitive load. Studies have
important enabling toolkit to bridge customer needs and company proposed needs-based configuration systems facilitate consumer
offerings. In the configuration process, customers choose from a decision making, particularly for customers without much domain
set of predefined attributes and their options. The combination of knowledge [5]. The needs-based configurators show a series of
choices forms the desired product configuration. It is observed that
product descriptions to customers. Customers then just need to
some online configurators provide default options for each
attribute. Although previous studies show that the default option indicate importance or relevance of the descriptions and use
significantly affects customers’ choices during the product semantic words (e.g., ‘cheaper’ or ‘larger’) to modify an existing
configuration process, it is not clear how other factors mediate this reference product. This can greatly reduce the semantic gap
impact. In this paper, we investigate how product types, number of between customer needs and the company’s offerings, although the
choices, customers’ degree of expertise, the importance of the needs in natural language is still not supported.
attributes and the configuring sequence affect consumers’ decisions To help customers make easy decision, default options have
in the configuration process when default options are presented. been provided in many commercial configurators since mid-1990.
Based on a series of empirical experiments, we find that customers’ Studies also found that the default could potentially help predict
degree of expertise, the rating of the attribute importance, and the customers input when using an interactive online platform [13].
number of attribute choices have a significant effect on customers’
Recently, it has been observed that some online B2C configurators
choices for utilitarian products. For hedonic products, the
importance of the attributes and the configuring sequence are provide default options as well. If a customer makes no choice on
significant factors. the attribute, the default option is selected in the final product, as
can be found in the Mini Cooper’s configurator in Figure 1.
Keywords: status quo effect, configurator, default option,
customisation
1 INTRODUCTION
Due to the rapid growth of the Internet and e-commerce over the
past ten years, online choice configurators have become an
important toolkit for customisation by customers. This configure-
to-order-based mechanism has been widely used in industry.
Successful cases include Dell computers, Adidas, and Nike. By
using configuration systems, firms can increase their profit through
better sales and higher flexibility. Greater customer involvement in
the choice configurator also increases customer satisfaction [1].
Thus, companies can improve their competitive advantage and Figure 1. Screenshot of the Mini Cooper’s online product configurator
with default choices (accessed May 2017)
position by using these toolkits [2].
However, some challenges persist. One of the major challenges
In the study of economics and psychology, it has been
is to provide a more user-friendly interface to facilitate choice
acknowledged that the current situation (status quo) is often
navigation and decision making in the configuration process. Some
considered a reference point from the decision makers’ point of
effort has been devoted to this research direction. For example,
view. Deviation from the status quo is considered a loss, a
Wang et al. proposed information theory and game theory based
phenomenon called ‘status quo bias’. According to Mandl and
method to elicit customer needs adaptively [3] [4]. The
Felfernig [6], status quo bias exists in product configurators,
configuration sequence is also customised based on the active
meaning that consumers’ decisions are affected by the default
customer’s previous specifications during the configuration
options.
process. In this way, the customers’ choice navigation process is
Default options have also been studied in the marketing
more efficient and more user friendly. Customers can get what they
literature. They are considered a type of decision-making heuristic
through which cognitive load can be significantly reduced [7][11].
1 Department of Supply Chain and Information Management, School of Through empirical experiments, Johnson et al. also noticed that a
Decision Sciences, Hang Seng Management College, Hong Kong, China lack of cognitive attention leads customers to select default
*correspondence author, email: yuewang@hsmc.edu.hk
choices. Customers may be paying little or no attention when they them may be very high. In this case, customers may use the default
choose the default option [7][12]. This type of default is considered options to save effort in the configuring process.
an attention-based default. Order of the attributes – Levav showed that the order of the
Brown and Krishna argued that the default options can contain attributes also affects customers’ decisions in product
information about the product and thus affect consumer decision customisation [15]. In the present study, the order of the attributes
making, i.e., they can be considered information-based defaults [8]. in configurators is considered as a potentially significant factor in
For example, they found that low (less expensive) defaults customers’ choices when they face flexible option configurators.
sometimes have more positive effects than high (more expensive) Concern about the attribute - if a customer cares more about one
defaults in the case of information-based defaults. In addition, they particular attribute, he or she will be more motivated in the
may create negative effects when customers already know that the information processing task [16]. Often, consumers do not have
default option is the best choice. In this case, customers may be enough mental capacity to evaluate all of the attribute levels for all
less likely to choose the default choice than the non-default choice. of the attributes offered [17]. Consumers usually start with the
Compared with expert customers, novice customers more easily most important attribute and proceed based on the order of the
accept the default options [9]. Because the complexity of custom attributes’ importance [18]. In the context of product configurators,
decision-making tasks decreases the willingness of customers to concern about each product attribute is potentially a significant
participate and reduces the perceived value of the products, novice factor in customers’ choices.
customers are more affected. This means that when customers are
less familiar or have little knowledge of the product, the default
options have a greater impact [10]. 3 EXPERIMENT DESIGN
Although default options have been studied in marketing We develop configurators for a watch and a laptop, which are a
science research, it is not clear how the default options affect hedonic product and a utilitarian product, respectively. Screenshots
consumers’ decisions or which factors are significant in the of the watch and laptop configurators are shown in Figure 2. We
selection of default choices, particularly in the context of product only include the components related to aesthetics to the watch
configuration. Therefore, this paper addresses these questions configurators. Thus, all of the attributes of the watch can be
through empirical experiments. This content is organised as considered hedonic attributes, meaning that customer choices are
follows. The factors which potentially mediate consumers’ based purely on their subjective preferences. No expertise in
decision making under default option setting are introduced in watches is needed to finish the configuring task. For laptop, we
section 2. Section 3 elaborate the design of the empirical only include the functional components in the configurators. Thus,
experiment. Experimental results and discussion are in section 4. the laptop’s attributes are utilitarian. The choices are determined by
Section 5 concludes the whole paper. customers’ functional requirements. A certain amount of
background knowledge is needed to finish the configuring task.
Because the purpose of this paper is to study which factors affect
2 POTENTIALLY SIGNIFICANT
customer decisions when default choices are presented and
FACTORS customers’ satisfaction with the configured product and the
In response to the research question, we conduct empirical configuring process, the comparative study is conducted using a
experiments to identify the significant factors in customer traditional configurator. Thus, the four types of configurators used
decisions when default options are presented. The literature in this paper are developed as shown in Table 1. For each product,
suggests that default options affect customers’ decisions. However, the base configurator is the normal version without default options.
the process and context of product configuration are different from This is the configurator used as the control group. For the other
the product selection process studied in previous research. More versions, each attribute has a default option. To eliminate the effect
factors are involved in the configuration process. of option difference on customers’ choices, we randomly assign
Product type - Products can be classified into two categories: the default options for each experiment participant. It means that
utilitarian products and hedonic products [14]. For utilitarian for difference customers, the default options encountered in the
products, customer choices are based purely on the functional configuration tasks are different as well. This configurator is used
requirements. A certain domain knowledge or expertise is needed to investigate consumers’ decision behaviour. The default option
to finish the configuring task. For hedonic products, customers’ for each attribute is also randomly selected for each experiment
choices are made based their subjective preferences. For example, subject. This could offset the influence of choice on consumers’
the corresponding attributes may be colour, shape or design. selections.
Customers’ preferences for these attributes are subjective. In our In the experiment, a participant is randomly assigned to one of
research, we ask whether product type mediates customers’ the four configurators. After the configuring task, the participant is
selection of default options. directed to another configurator with a different product type and
Expertise - Experts have more experience and knowledge of the configurator type. For example, if the first randomly assigned
product, and therefore they may not be affected by the default configurator is configurator III, which is a traditional watch
option because they know what they want to purchase. Unlike configurator, then the next configurator the participant encounters
experts, novice customers have less knowledge about the product, is configurator II, which has different product type and
so they are easily affected by the default option. configurator type. Before each configuring task, the participant
Number of choices - it has been acknowledged that the number completes a pre-experiment survey for each product. The pre-
of choices may also affect consumers’ decisions. For example, if experiment survey is used mainly to determine the relative
an attribute has a large number of choices, the cost of evaluating importance that customers concern about each attribute and their
degree of expertise with the utilitarian product. The detailed
questions of the survey are shown in Figure 3. The experiment can
be summarised as in Figure 4.
Figure 2(a). Screenshot of the watch configurators, with default options
Figure 3(b). Screenshot of the pre-experiment survey (partial) of laptop
configurators to determine customers’ degree of expertise
Figure 4. Experiment process
4 EXPERIMENTAL RESULTS AND
DISCUSSION
Figure 2 (b). Screenshot of the PC configurators, with default options
Table 1. Configurators used in the experiment. 4.1 Basic statistics
Base Configurator w/ One hundred forty participants are recruited from a university in
configurator w/o default options Hong Kong. Each experiment subject receives 30 Hong Kong
default options dollars as compensation for his or her time and effort. We check
Laptop (utilitarian I II customers’ choice distribution with and without default choices.
product)
The purpose is to see whether the default choices lead to a
Watch (hedonic III IV significant difference in consumers’ behaviour.
product)
The statistics on the choice distribution are shown in the
following table. If the default options have no effect on customers’
decisions, the distribution of customers’ choices should not be
significantly different for the two types of configurators, i.e., with
and without default choices. A chi-square test is used to check the
difference between the distributions. The p-value of the test result
is shown in the last column.
Table 2. Consumers’ choice distribution for watch attributes
Attribute Number Attribute choice Attribute choice P-value
of distribution distribution of chi-
Attribute (with default (w/o default square
choices option, 40 option, 52 test
subjects) subjects)
Frame 3 (15, 14, 11) (21, 14, 17) 0.501
Figure 3(a). Screenshot of the pre-experiment survey of watch
configurators
Band 6 (3, 9, 3, 11, 4, (2, 6, 0, 21, 3, 0.004 as mentioned in the previous section. The independent variables
10) 20) are the number of choices, the order of the attributes, the concern
about each attribute and the customers’ expertise (only for the
Calibre 2 (10, 30) (21, 31) 0.047 laptop, the utilitarian product). The numbers of choices for the two
types of products are shown in the second column of Tables 1 and
Outer 8 (8, 5, 6, 2, 8, 2, (9, 9, 6, 2, 10, 8, 0.014 2. The relative importance that customers accord to each attribute
5, 4) 7, 1) is elicited from the pre-experiment survey. We use the pre-test
Arm 2 (19,21) (15, 37) 0.009 survey to elicit information about the customers’ concern about
each attribute. A Likert scale ranging from 1 to 7 is used to allow
customers to specify their degree of concern. ‘1’ corresponds to the
least degree of concern, and a larger number means a higher degree
Table 3. Consumers’ choice distribution for laptop attributes
of concern. A sample question for the watch configurator is ‘How
Attribute Number of Attribute Attribute P-value of
chi-square
concerned are you with the calibre compared to other parts of a
Attribute choice choice
test watch?’ Regarding expertise, we designed a basic knowledge test
choices distribution distribution
(with default (w/o default for laptops containing 10 multiple-choice questions. The number of
option, 49 option, 47 correctly answered questions is used as the measure of the
subjects) subjects) customer’s degree of expertise.
Monitor 5 (19, 19, 8, 2, (6, 27, 9, 5, 0) 0.000 Because the responses are binary variables, logistic regression is
1) used to identify the relationship between independent variables and
Resolutio 3 (6, 35, 8) (8, 31, 8) 0.64 responses. The result is shown in Tables 3 and 4. For the laptop,
n the utilitarian product, expertise is an independent variable. For the
Screen 2 (12, 37) (20, 27) 0.011 watch, the hedonic product, the selection of attributes does not
depend on customers’ expertise; only subjective preferences
Operating 4 (16,12, 13, (18, 5, 17, 7) 0.014 matter. Thus, expertise is not considered in the regression model of
System 8) the watch. Model 1 includes all of the independent variables and all
CPU 6 (4, 17, 16, 4, (2, 7, 18, 12, 0.001 of the first-order interactions between independent variables. A
5, 3) 6, 2) stepwise procedure is then conducted to remove the insignificant
RAM 9 (4, 4, 10, 1, (3, 3, 8, 6, 10, 0.000 factors one by one from the model according to the p-value in the
10, 8, 2, 5, 4, 8, 1, 4) regression until only the significant variables remain.
5)
Graphics 5 (11, 18, 13, (8, 11, 20, 4, 0.066 Table 4. Relationship between response and different variables - laptop
Card 2, 5) 4)
Independent Variables Model 1 (logistic Model 2 (logistic
regression) regression, stepwise
Hard disk 7 (7, 11, 4, 8, (7, 7, 10, 5, 8, 0.210
result based on
8, 5, 6) 6, 4) model 1)
Battery 6 (5, 8, 12, 8, (10, 4, 8, 11, 0.071 Expertise 0.693* 0.715**
5, 11) 3, 11) (0.384) (0.321)
Concern about attribute -0.339 -0.234***
Based on the tables, we can see that for most attributes, the (0.425) (0.0802)
distributions of customer choices are significantly different, as the Sequence of -0.353
corresponding p-value is small. This means that default options configurator (0.614)
affect customers’ decisions during the configuring process. We Number of choices 0.198 0.402*
notice that only the watch frame in watch, screen resolution and (0.424) (0.227)
hard disk in laptop don’t have significant difference between the Expertise * Concern 0.0022
base configurators and the default option-based configurators. (0.0483)
After further investigation, we found that the choices for these Expertise * Sequence 0.0082
(0.0478)
three attributes either have very strong dominance relationship in
Expertise * Number of -0.1053** -0.1028**
terms of customer preferences (screen resolution or hard disk), or
Choices (0.0515) (0.0504)
very heterogeneous customer preferences (watch frame, the
Concern * Sequence 0.0074
choices can be found in Figure 2). For the former case, customers (0.0494)
tend to choose the clearly superior choices regardless of the default Concern * Number of 0.0093
options. For the latter case, customers’ choices are purely Choices (0.0547)
determined by the preferences. Default options can hardly change Sequence * Number of 0.0613
their intrinsic preferences. Choices (0.0991)
*: p-value<0.1; **: p-value<0.05; ***: p-value<0.01
Remark: the numbers represent the coefficients of the corresponding
4.2 Which factors affect customers’ decisions? independent variables in the logistics regression. The numbers in the
parentheses are the standard deviation of the corresponding coefficients.
Because we want to study the effects of different factors on the
selection of default options, it is natural to use a binary variable as
Based on the result shown in Table 4, we find that the degree of
an indicator that indicates whether the participant selects the
expertise is moderately significant in affecting customers’
default option in the configuring task for configurators II and IV,
decisions about default choices. The interaction of degree of they do not choose the default option. This finding is identical to
expertise and number of choices is significant in affecting the case of the laptop. However, in contrast to the laptop
customers ’ decisions to choose the default options. Through a configurator, the sequence of the attribute in the configuring
stepwise procedure, we can eliminate the insignificant independent process is significant. We think the reason is that for the laptop
variables one at a time. This leads to model 2, which consists only configurator, the numbers of choices for different attributes are
of the significant independent variables. We find that the degree of quite similar. However, for the watch configurator, the number of
expertise, degree of concern about each attribute, and the choices ranges from 2 to 24. Thus, the sequence is significant in
interaction between degree of expertise and number of choices are the customer’s decision. In addition, it is observed that customers
significant in affecting customers’ decisions. In particular, the tend to choose the default options that are presented early. We also
coefficient of expertise is positive. This means that if a customer’s find that the interaction between concern and number of choices is
expertise is greater, he or she is more likely to choose the default also significant in affecting the choices.
options. This finding seems different from previous study in [9]. It
should be noted that we use logistic regression to identify the
relationship between the independent variables and the choice of 5 CONCLUSION
default options. In [9], the authors study the relationship between Product configurator design has been widely studied in the area of
the number of selected default options and the expertise degree. engineering. Very little work investigates the effect of default
Thus the research questions are different. This can explain the options on consumer decision making during the configuring
difference of the experiment findings. process. This paper studies whether default options have a
The sign of the coefficient of degree of concern is negative, significant effect on people’s decisions in the context of product
indicating that if a customer is more concerned with an attribute, customisation. In the settings of product configurators, a default
then he or she is less likely to choose the default options. The choice is highlighted for each product attribute. During the
coefficient of number of choice is positive, meaning that if an experiment, we find that some respondents accept the default
attribute has more choices, customers are more likely to choose the choices and others reject them. It is of primary interest to study
default option. It has been acknowledged that when more choices which kinds of products and what type of attributes are influenced
are presented, the burden of choice is much higher. In this most by the default options. Through a set of empirical
situation, customers may stay with the default option to save time experiments, we show that customers’ choices are significantly
and effort in product configuration. influenced by default options. For utilitarian products, we also note
that expertise, concern for the product attribute, number of choices
Table 5. Relationship between response and different variables - watch and the interaction between expertise and number of choices
Independent Model 1 (logistic Model 2 (logistic significantly mediate the default options’ effect on customers’
Variables regression) regression, stepwise
choices. However, for hedonic products, concern about the product
result based on model
attribute, order of configuration and the interaction between
1)
concern and number of choices are significant factors. From
Concern about 0.17 -0.218*
attribute (0.337) (0.129) companies’ perspective, customers are more likely to select the
Number of Choices 0.078
default options. This could potentially benefit customisers and
(0.145) improve the operations of the company.
Sequence -0.333 -0.334*** This research still has some limitations. The number of subjects
(0.446) (0.112) can be larger and the subjects have similar background. Thus, only
Concern * Number of -0.0182 -0.016*** lab experiment is used to conduct the research. To provide more
Choices (0.017) (0.00422) convincing research outcome, field experiment will be carried out.
Concern * Sequence 0.021 In addition, the methods on quantifying the expertise degree of the
(0.089) subjects is very sensitive to the discrimination of the questions in
Number of Choices * -0.0315 the pre-survey test. In our future work, we plan to recruit more
Sequence (0.053) participants and further polish the questionnaire to quantify the
*: p-value<0.1; **: p-value<0.05; ***: p-value<0.01 degree of expertise more accurately. Furthermore, the order of
Remark: the numbers represent the coefficients of the corresponding configuration may be a significant factor as well. In the future
independent variables in the logistics regression. The numbers in the study, we plan to randomise the configurating order for the
parentheses are the standard deviation of the corresponding coefficients. research.
For the watch configurator, the attributes are not technical. The
selection is based purely on appearance, and no knowledge is ACKNOWLEDGEMENTS
required for the configuring task. Therefore, there is no individual
This research is supported by Hong Kong Research Grants Council
variable to quantify the degree of expertise. Based on model 1, we
(Project No. UGC/FDS14/E02/15, for data collection) and (Project
find that none of the individual variables are significant. Through a
No. UGC/FDS14/E07/17, for data analysis).
stepwise procedure, the original regression model can be modified
to model 2, in which all of the variables are significant. The degree
of concern is moderately significant. Configuring sequence and the REFERENCES
interaction of concern with number of choices are significant in
affecting customers’ decisions to choose the default options. We [1] E. Garbarino and S. M. Johnson, ‘The Different Roles of Satisfaction,
also notice that all of the signs of the coefficients are negative. Trust, and Commitment in Customer Relationships’, Journal of
Marketing, 63(2), 70–87, (1999).
Therefore, when customers are more concerned with the attribute,
[2] F. S. Fogliatto, G. J. Da Silveira and D. Borenstein, ‘The mass
customization decade: An updated review of the literature’,
International Journal of Production Economics 138(1), 14-25,
(2012).
[3] Y. Wang, and M. M. Tseng, ‘Attribute selection for product
configurator design based on Gini index’, International Journal of
Production Research, 52(20), 6136-6145, 2014.
[4] Y. Wang, and M. M. Tseng, ‘Adaptive Attribute Selection for
Configurator Design via Shapley Value’, Artificial Intelligence for
Engineering Design, Analysis and Manufacturing, 25 (1), 189–199,
(2011).
[5] T. Randall, C. Terwiesch and K. T. Ulrich, ‘User Design of
Customized Products’, Marketing Science, 26(2), 268–280, 2007.
[6] M. Mandl, A. Felfernig, J. Tiihonen, and K. Isak, ‘Status Quo Bias in
Configuration Systems’, 24th International Conference on Industrial,
Engineering and Other Applications of Applied Intelligent Systems
(IEA/AIE 2010), Syracuse, New York, 105-114
[7] E. J. Johnson, S. Bellman, and G. L. Lohse, ‘Defaults, framing and
privacy: Why opting in-opting out’, Marketing Letters, 13, 5–15,
(2002).
[8] C. L. Brown, and A. Krishna, ‘The skeptical shopper: A
metacognitive account for the effects of default options on choice’,
Journal of Consumer Research, 31, 529–539, (2004).
[9] J. Wang, L. Cheng and W. Han, ‘The effect of default option on
customer decision behavior in product customization’, Proceedings of
10th International Conference on Service Systems and Service
Management, Hong Kong, 2013
[10] R. Thaler, and C. R. Sunstein, ‘Libertarian Paternalism’, American
Economic Review, 93 (2), 175-179, 2003.
[11] J. Park, and M. R. Banaji, ‘Mood and heuristics: The influence of
happy and sad states on sensitivity and bias in stereotyping’, Journal
of Personality and Social Psychology, 78, 1005–1023, (2000)..
[12] A. Tversky and D. Kahneman, ‘Judgment under Uncertainty:
Heuristics and Biases’, Science. 185(4157),1124-31, 1974.
[13] D. Jannach and L. Kalabis, ‘Incremental prediction of configurator
input values based on association rules - A case study’. In:
Proceedings of the International Workshop on Configuration
(ConfWS 2011 at IJCAI 2011). Barcelona, Spain, 2011
[14] R. Batra, O. T. Ahtola, ‘Measuring the hedonic and utilitarian sources
of consumer attitudes’, Marketing Letters 2(2), 159-170, (1991).
[15] J. Levav, M. Heitmann, A. Herrmann, and S. S. Iyengar, ‘Order in
Product Customization Decisions: Evidence from Field
Experiments’, Journal of Political Economy, 118(2), 274-299, 2010
[16] D. J. MacInnes, C. Moorman and B. J. Jaworski, ‘Enhancing and
measuring consumers’ motivation, opportunity and ability to process
brand information from ads’, Journal of Marketing, 55, 32-53, 1991.
[17] R. W. Olshavsky. ‘Towards a More Comprehensive Theory of
Choice’, in NA - Advances in Consumer Research Volume 12, eds.
Elizabeth C. Hirschman and Moris B. Holbrook, Provo, UT:
Association for Consumer Research, 465-470, (1985).
[18] A. Tversky. ‘Elimination by aspects: A theory of choice’,
Psychological Review, 79(4), 281-299, (1972).