CONJOINT ANALYSIS: A STRATEGIC TOOL FOR
PRODUCT RESEARCH
Ighomereho, O. S.
Department of Economics and Business Studies
Redeemer's University, Ogun State, Nigeria
E-mail: [email protected]
ABSTRACT
Despite efforts by companies to offer products that meet customers' needs, a large
percentage of them still fail. One of the reasons for product failure is negligence on
the part of the company to find out what customers want and how they want it. The
fulfillment of customers' needs and wants in a profitable way requires that
companies understand the attributes of their product(s) that are most valued by the
customers. Such information can lead to the creation of optimal value propositions.
This study considered how conjoint analysis could be used to aid this process. It
discussed the role of conjoint analysis in the determination of buyers' responses to
a product during concept testing and test marketing and also for the modifications
of existing products. It also demonstrated how Microsoft Excel could be used for
conjoint analysis by companies when developing new products or when managing
existing products in the face of intense competition.
Keywords: customer, new products, existing products, conjoint analysis, company,
competition.
INTRODUCTION
Globally, companies are faced with numerous difficult decisions directed at
assessing and maximizing future profitability, sales, and market share for new
product entries or modifications of existing products given the current offering of
competitors. Decisions about product attributes including price play a significant
role in the success or failure of new products (McNally and Schmidt, 2011).
Developing the "right" new product is critical to a company's success and is often
cited as a key competitive dimension (Chao and Kavadias, 2008). Companies often
spend a lot of resources developing new products only to find that consumer
adoption is much lesser or much slower than expected. Sometimes the problem is
that the final product fails to deliver on its promises.
However, the problem is that the company has not properly estimated the
product's market potential. Product research enables management to select more
effective, more efficient, less risky and more profitable alternatives that can
maximize sales (Cooper and Schindler, 2006). Product research is a component of
marketing research that provides information and intelligence on the attributes of a
good or service that could satisfy a recognized need or want of consumers. It
involves concept testing, determination of optimal product design, package tests, test
marketing, product modification, brand positioning and repositioning (Kotler and
Keller, 2006). Conjoint analysis is one of the techniques that could be useful for
such purpose. The objective of conjoint analysis is to determine the combination of a
limited number of attributes that is most influential on customers' choice or decision
making. Achieving marketing success with a new product requires that
International Journal of Economic Development Research and Investment Vol 2 No. 3, Dec. 2011 1
market needs and preferences be determined (Kotler and Keller, 2006). This is
applicable to both local and foreign market. By using conjoint analysis, a company
can answer questions such as: which product attributes are important or unimportant
to the consumer? What levels of product attributes are the most or less desirable in
the consumer's mind? Should pricing or other attributes of current products be
modified in response to competition? Each of these identified management problems
may be addressed and solved using the technique of conjoint analysis. If the
management of a company knows what customers either local or international really
value, then they know where to put their strategic efforts.
Despite the extensive use of conjoint analysis in American and European
companies (Sawtooth Software, 2002), it is relatively not used by marketing
researchers in developing countries (Kotri, 2006). Using the technique of conjoint
analysis requires a thorough knowledge of statistical data analysis such as logistic
regression, multinomial logit and probit which may discourage its use. In order to
make conjoint analysis easy for in-house use by companies, the application of
Microsoft Excel was introduced by Dobney (2000). So far, the demonstration of
conjoint analysis with excel used ranking and rating approach of conjoint analysis
(Dobney, 2000) with none on choice-based conjoint. To fill this gap, this study
demonstrated how to use Excel for choice-based conjoint analysis. It also described
the approach of conjoint analysis and discussed the role of conjoint analysis in new
product development and in the management of existing products.
THE APPROACH OF CONJOINT ANALYSIS
Green and Srinivasan (1978) defined Conjoint analysis as any decompositional
method that estimates the structure of preferences given overall evaluation of a set of
alternatives that are pre-specified in terms of levels of different attributes. In this
study, it is defined as a survey method of data collection and analysis for eliciting
preferences for a product. It is based on the premise that the relative values of
attributes considered jointly can better be measured than when considered in
isolation. Its critical assumption is that preference for an object is a function of the
specific attributes of the object rather than the object per se (Min, 2007). Conjoint
analysis was introduced to marketing 40 years ago in a seminal paper by Green and
Rao (1971) where the conjoint measurement theory developed in psychology by
Luce and Turkey (1964) was adapted to the solution of marketing problems. Since
then, it has become an important marketing research tool that is being used
extensively in marketing to analyze consumer trade-offs, understand how customers
make purchase decisions and predict consumer behaviour as well as determine how
people value different features that make up an individual product for the purpose of
providing products that better meet customers' needs (Green and Srinivasan, 1978,
Green, Carroll and Goldberg, 1981, Green and Srinivasan, 1990, Chen and
Hausman, 2000 and Green, Krieger and Wind, 2001).
Steps in Developing a Conjoint Analysis Study
According to Ryan and Farrar (2000), the application of conjoint analysis involves
the following stages:
International Journal of Economic Development Research and Investment Vol 2 No. 3, Dec. 2011 2
Stage 1: Identification of Attributes: These are the dimensions on which products
are described. The first step in conjoint analysis is to identify and choose objective
attributes that describe the product such as colour, size or price. If a policy question
is being addressed, the attributes will be predefined. Where the attributes are not
predefined, literature reviews, group discussions and individual interviews will be
necessary to identify the attributes. Orme (2010) states that attributes should cover
the full range of possibilities for the product and they should be independent with no
overlapping meaning.
Stage 2: Assigning Levels to the Attributes: Levels represent the different
realizations of the attributes that we find in the marketplace. This step involves
choosing the options for each attribute. The levels must be plausible and actionable.
Levels assigned to the attributes may be cardinal (for example, the attribute of size,
may have levels of 250g, 500g or 1000g) or ordinal (for example, small, medium or
big).
Stage 3: Design Product Profiles: At this stage, products are defined as a
combination of levels of different attributes. When the attributes are many, the
product profiles that will be used should be a subset of the possible universe of
product profiles. This is because all the profiles generated cannot be included in the
questionnaire.
Stage 4: Select the Presentation Medium: Choose the form in which the
combinations of attributes are to be presented to the respondents. According to Mora
(2011), the presentation of profiles is not restricted to text. Other options include
verbal presentation and pictorial presentation.
Stage 5: Select the Technique to be used to analyze the Collected Data: The part-
worth model is the model used to express the utilities or the measure of desirability
of the various attributes levels. This can be estimated with different techniques such
as ordinary least square regression analysis and logistic regression.
CONJOINT ANALYSIS METHODOLOGY
Conjoint analysis requires respondents to make a series of trade-offs between
different levels of product attributes. Since the introduction of conjoint analysis,
three principal approaches have emerged, the full profile conjoint, adaptive or hybrid
conjoint and choice-based conjoint (Bakken and Frazier, 2006). These approaches
vary in the methods of eliciting preferences and estimating utility. In a full profile
conjoint, product alternatives are presented as complete profiles consisting of one
level for each of the attributes. Respondents consider each of the profiles one-at-a-
time and either rate them on some preference measure or rank order them. With
ranking, the respondents are asked to list the profiles in order of preference, that is,
from most preferred to least preferred using cards. Researchers later found out that
the rating method which requires respondents to give each trade-off scenario a score,
of say 1 to 10, indicating the level of appeal gave better results (Krieger et al, 2004).
The ranking and rating exercise will reveal each respondent’s priorities and
International Journal of Economic Development Research and Investment Vol 2 No. 3, Dec. 2011 3
preferences (Bakken and Frazier, 2006). The drawback in this approach is that the
number of attributes that can be used is highly restricted usually 4 to 6. That is why
Green and Srinvasan (1990) argued that full profile conjoint should be used when
the number of attributes included in the conjoint study is small (up to six). The
adaptive or hybrid conjoint analysis was introduced by Richard Johnson at Sawtooth
Software in 1985 as a reaction to the number of attributes problem especially, the
limitation of full profile conjoint in terms of the number of attributes that it can
handle (Bakken and Frazier, 2006). It combines stated preferences and attribute
importance ratings with a paired comparison tasks. Rather than ask respondents to
evaluate all attributes at the same time as in full-profile, it reduces the levels in each
profile and asks respondents to evaluate a few profiles which are referred to as
partial profiles using pair-wise rating. With this approach, the researcher can handle
up to 30 attributes (Green and Srinivasan, 1990). For rating and ranking-based
conjoint methods, the basic weighted additive model can be stated as follows:
J M
βjm · xjm+εk
rk = β0 + ∑ ∑
j=1 m=1
Where:
rk = response for option k;
β0 = intercept or constant
βjm = partworth utility of level m of attribute j;
xjm = 1 if option k has level m on attribute j else xjm = 0
εk = error term
The partworth utilities are estimated, usually by applying multiple
regression, such that the sum of squares between empirically observed responses rk
(ratings or rankings) and estimated responses ri is minimal. In more recent years, it
has become common practice to use choice-based conjoint analysis because it
closely resembles real life decisions. It is the application of discrete choice modeling
to understanding consumer decision making. The groundwork for modeling choice
among multiple alternatives was laid by Dan McFadden in 1973 (Sawtooth
Software, 2008). In 1983, Louviere and Woodworth introduced an approach that
used only a choice task by integrating the mathematics of discrete choice modeling
with conjoint analysis (Sawtooth Sotfware, 2008). This became the basis of choice-
based conjoint analysis in the 1990s. The main characteristic distinguishing choice-
based conjoint analysis from other types is that respondents do not rank or rate a
series of profile; they simply observe a set of available alternatives and chose the
most preferred alternative. Respondents could be asked to evaluate two or more
profiles at a time and to indicate their preferences by making a choice (Hauser,
2002). In addition, choice-based conjoint analysis enables the researcher to include a
"none" option for respondents which might read "I wouldn't choose any of these".
The utility associated with a profile is given as:
Uij = βxij + eij
Where:
Uij = the utility of respondent i associated with profile j (this could be profile A or B)
International Journal of Economic Development Research and Investment Vol 2 No. 3, Dec. 2011 4
β = a vector of parameters to be estimated
xij = a vector of attributes of profile j presented to respondent i
eij = the stochastic portion of the utility function
Respondent i would choose profile A over profile B if UiA › UiB, and the probability
of such choice is Pi (A) = Prob {βxIA + eiA ≥ βxIB + eIB}
THE ROLE OF CONJOINT ANALYSIS IN PRODUCT DEVELOPMENT
A company develops new products to increase sales and profits, to respond to
changing customer needs, to gain competitive advantage, to meet technological
changes and to diversify risk. However, developing and introducing new products is
frequently expensive and risky (Pride and Ferrell, 2008). A major breakthrough of
conjoint analysis in product research is in new product development. It could be
used to measure, analyze and predict customers' responses to new products and also
to estimate the price customers will be willing to pay. Developing the "right" new
product is therefore crucial and if a new product must succeed, it must consist of the
desired attributes that the customers want. It should be able to satisfy their needs and
it should be planned.
The planning process involves the stages of new product development which
include idea generation, ideas screening, concept development and testing, business
analysis, prototype development, test marketing and commercialization (Bearden,
Ingram and LaForge, 2007). The two critical stages where conjoint analysis could be
useful are concepts testing and test marketing. Concept development and testing
involve a description of the proposed product including its features and its probable
price and presenting it to appropriate target consumers through a survey (Kotler and
Keller, 2006). This allows companies to model and test different product options to
evaluate likely market preferences and potential share, revenue and profit, all based
on what customers' really value.
It provides opportunity for companies to determine customers' initial reaction
to a product idea before investing resources in its production. The result of concept
testing can help a company better understand the product attributes and benefits that
are most important to potential customers. Test marketing is when the product
prototype is made available in certain geographical areas considered to be
representative of the market to study consumers' response to it. The aim is to
determine the extent to which potential customers will buy the product. This enables
a company to put new products and their supporting marketing programmes through
validating tests prior to full-scale product launches.
Usually, when testing the viability of new products, potential consumers are
asked to indicate how important some attributes are to them. The technique of
conjoint analysis has been proved to be better than other approaches to
understanding consumer preferences and decision-making such as Contingent
Valuation, ordinary surveys and focus group estimates because it provides
opportunities for respondents to answer survey questions as if they were placed in a
real market situation (Hauser and Rao, 2002 and Kotri, 2006). It also estimates the
relative importance of different attributes and the various levels of the attributes of a
product which allows the policy maker to observe the individual impact of each
International Journal of Economic Development Research and Investment Vol 2 No. 3, Dec. 2011 5
attribute on the overall benefit (Krieger Green and Wind, 2004). It can produce
results that may not be obtained from compositional approach where respondents are
asked to directly state their assessment of the importance of the attributes (Orme,
2010).
THE ROLE OF CONJOINT ANALYSIS IN MANAGING EXISTING PRODUCTS
Conjoint analysis is also useful for managing existing products in order to overcome
intense competition in the business environment. Whenever a new product succeeds,
competing products are bound to spring up and these products could have a
significant impact on profits or market share if a company does not make any
change in its products overtime (Kotler and Keller, 2006). Also, the markets are
highly dynamic. What was a profitable product yesterday may not be tomorrow
because customers' attitudes and preferences change overtime (Pride and Ferrell,
2008). For a company to maintain it market share, it must seek for ways of
improving the product by finding out the attributes that are currently appealing to the
consumers. Conjoint analysis could be used to find out if customers are satisfied
with the current product or they want a change in some of the product attributes. In
designing the choice-based conjoint questionnaire, the current product is displayed
consistently with prospective versions of the product. Analysis of the responses will
indicate the action to be taken.
CONJOINT ANALYSIS USING EXCEL
In this demonstration, the choice-based conjoint was used because it tries to mimic
the actual purchase decision process for products within a competitive context. Also,
it includes the "none" option which indicates a buy or non-buy decision. By
selecting that option, a respondent can contribute information about the decrease in
demand to be expected if the product becomes unattractive in some ways. In this
case, the proposed new product is a multipurpose soap.
Table 1: Attributes and Levels of the Proposed Soap
Attributes Levels
Type Liquid, Solid
Size Small, Big
Use Washing/Bathing only, Multipurpose
Scent With scent, No scent
With four attributes and two levels each for three attributes and three levels for one
attribute, it is possible to form twenty-four possible product profiles (2 X 3 X 2 X 2).
Since they are few, they can all be tested but if they are many say 108, then
orthogonal selection will be necessary. In this case, respondents were shown twenty-
four hypothetical offers formed by combining varying levels of the attributes and
combined to form twelve choice items.
Table 2: A Sample of a Conjoint Profile
If you were to buy soap and these were your only options, which would you choose?
Attributes Option 1 Option 2
Type Liquid Liquid None
Size Small Small I wouldn't choose any of these
Use Bathing only Multipurpose
Scent With scent Without scent
International Journal of Economic Development Research and Investment Vol 2 No. 3, Dec. 2011 6
They were then asked to make a choice. The respondents' answers will indicate
whether the product concept (multipurpose soap) has a broad and strong consumer
appeal or not. The objective is to derive the utility values that consumers attach to
varying levels of soap attributes. When preference data are collected from a
sufficient sample of target customers, the data could be used to estimate the market
share of the product. A pilot study of 10 respondents is shown in the excel sheet
below:
Table 3: Excel Demonstration of Choice-Based Conjoint Analysis
A B C D E F G H I J K L M
1 Type Size Use Scent A/L TNTC TNTO Utility
2 Solid Liquid Small Big Wash Bath Multipurpose Scent No
3 8 4 5 7 4 4 0 10 2
4 7 5 5 7 2 5 1 10 2 Solid 80 120 0.67
5 8 4 4 8 4 3 1 11 1 Liquid 40 120 0.33
6 9 3 6 6 2 6 0 9 3 Small 50 120 0.42
7 8 4 4 8 1 7 0 11 1 Big 70 120 0.58
8 10 2 4 8 2 5 1 9 3 Washing 26 80 0.32
9 5 7 5 7 3 5 0 10 2 Bathing 50 80 0.63
10 8 4 6 6 4 4 0 10 2 Multipur 4 80 0.05
11 9 3 7 5 2 6 0 9 3 Scent 100 120 0.17
12 8 4 4 8 2 5 1 11 1 No Scent 20 120 0.17
A/L = Attribute Level, TNTC = Total Number of Time Chosen , TNTO = Total Number of Time Occurred,
Table 3 shows the Excel analysis of customers' preference for soap. The
figures from row 3 to row 12 were derived from the number of times those attribute
levels were chosen by the respondents. In the demonstration above, there are 10
respondents. To get the total number of times each attribute level was chosen, add
up the number of times each attribute level was chosen by the 10 respondents, that
is, for the attribute level of solid soap, we have 8+7+8+9+8+10+5+8+9+8 = 80.
Also, the column of "number of times each attribute level occurred" was derived by
counting the number of times each attribute level occurred in the questionnaire.
Based on the notion of balance, each attribute is expected to appear approximately
the same number of times. A careful look at table 3 shows that each attribute
occurred equally which is 240 times. The last column indicates the utility level
which is calculated by dividing "total number of times an attribute level was chosen"
by "total number of times it occurred". The utility value ranges between 0 and 1. The
higher the value, the more preferred the attribute level. The attribute levels that have
a high utility value (above 0.50) are decending as follows: scent, solid, bathing and
big. This indicates that the most preferred soap is a big solid bathing soap with scent
(table 3).
Table 4: Importance of each Attribute
Type 19.5%
Size 09.3%
Use 33.3%
Scent 37.9%
Table 4 shows the importance of each attribute. This was calculated by
dividing the difference between the best and worst level on each attribute by the sum
of the difference between the best and worst levels of all attributes. Given the
attributes tested, it shows that scent has a larger impact on decision making when
considering buying soap. This is followed by use, type and size. The preference
International Journal of Economic Development Research and Investment Vol 2 No. 3, Dec. 2011 7
indicated above could be used to determine the type of soap to produce. Since
multipurpose soap is not significant, the idea of introducing a multipurpose soap
should be dropped. This is because it will not contribute much to market share and
profitability. Incorporating these values in decision making ultimately may result in
decisions that better reflects the preferences of customers and increase in corporate
performance.
CONCLUSION
Understanding customer needs and designing appropriate products is a crucial
success factor in today's highly competitive market. The goal of product research is
to identify products that would maximize the market share and profit of a product
given a company's limited resources. This paper has discussed the potential of using
conjoint analysis for analyzing customers' preferences for new and existing
products. However, its use is subject to the limitation of the small number of product
attributes that can be effectively analyzed. When the attributes and levels are few, it
is possible to evaluate all possible combinations. However, when the product
attributes and levels are very many, fractional factorial designs and orthogonal
designs could be used to create a smaller design that is still large enough to estimate
the utility of each of the attribute levels or to estimate the effects the researcher is
interested in.
REFERENCES
Bakken, D. and Frazier, C. L. (2006). Conjoint Analysis: Understanding Consumer Decision
Making. In Rajiv G. and Marco V. The Handbook of Marketing Research: Uses, Misuses
and Future Advances. California: Saga Publications, Inc.
Bearden W. O., Ingram T. N. and LaForge R. W. (2007). Marketing: Principles and Perspective.
New York: McGraw- Hill Inc.
Chao, R. O. and Kavadias, S. (2008). A Theoretical Framework for Managing the New Product
Development Portfolio: When and How to use Strategic Buckets. Marketing Science, (54)5,
907-921.
Chen, K. D and Hausman, W. H. (2000). Mathematical Properties of the Optimal Product line
Selection Problem using Choice-Based Conjoint Analysis, Management Science, (46)2, 327-
332.
Cooper, D. R. and Schindler, P. S. (2006). Marketing Research. Irwin: McGraw-Hill.
Dobney, L. (2000). Conjoint Analysis Example using Excel, Retrieved from http/www.dobney.com
Green, P. E. and Rao, V. R. (1971). Conjoint Measurement for Quantifying Judgmental Data.
Journal of Marketing Research, 8: 355-363.
Green, P. E. and Srinivasan, V. (1978). Conjoint Analysis in Consumer Research: Issues and
Outlook. Journal of Consumer Research, Vol 5, September, 103-123.
Green P., Carroll J. and Goldberg, S. (1981). A General Approach to Product Design Optimization
via Conjoint Analysis. Journal of Marketing, Vol 43, 17-35.
Green, P. E. and Srinivasan, V. (1990). Conjoint Analysis in Marketing: New Development with
Implications for Research and Practice. Journal of Marketing, Vol. 54, 3-19.
Green P. E., Krieger, A. M. and Wind, Y. (2001). Thirty Years of Conjoint Analysis: Reflections
and Prospects. Interfaces, 31(3): S56-S73.
Hauser, J. R. and Rao, V. R. (2002). Conjoint Analysis, Related Modeling and Applications. In
Wind, Y. and Green, P. E. (eds), Market Research and Modelling: Progress and Prospects.
Norway: Kluwer Academic Publishers.
International Journal of Economic Development Research and Investment Vol 2 No. 3, Dec. 2011 8
Kotler, P. and Keller, K. L. (2006). Marketing Management. England: Pearson Education.
Kotri, A. (2006). Analyzing Customer Value using Conjoint Analysis: The Example of a Packaging
Company. Tartu: University of Tartu Press.
Krieger A., Green P. E. and Wind, Y. (2004). Adventures in Conjoint Analysis: A Practitioner's
Guide to Trade-off Modelling and Applications. Philadelphia: University of Pennsylvania.
Luce, R. D. and Tukey, J. W. (1964). Simultaneous Conjoint Measurement: A New Type of
Fundamental Measurement. Journal of Mathematical Psychology, 1, 1-27.
Min, D. (2007). An Incentive-Aligned Mechanism for Conjoint Analysis. Journal of Marketing
Research, Vol XLIV 214-223
McNally, R. C. and Schmidt, J. B. (2011). An Introduction to the Special Issue on Decision Making
in New Product Development and Innovation. Journal of Product Innovation Management,
28: 619-622.
Mora, M. (2011). Conjoint Analysis and Realism in Price Research. Retrieved from
http://relevantinsights.com/tag/choice-based-conjoint-analysis.
Orme, B. (2010). Getting Started with Conjoint Analysis: Strategies for Product Design and Pricing
Research. Madison: Research Publishers LLC.
Pride, W. M. and Ferrell, O. C. (2008). Marketing. Boston: Houghton Mifflin Company.
Ryan, M. and Farrar, S. (2000). Using Conjoint Analysis to elicit Preferences for Healthcare.
Business Management Journal. Vol 320, 150-153.
Sawtooth Sotfware (2002). Commercial Use of Conjoint Analysis in Europe: Results and Critical
Reflections. Technical Paper Series, Retrieved from www.sawtoothsoftware.com.
Sawtooth Sotfware (2008). The CBC System for Choice-Based Conjoint Analysis. Technical Paper
Series, Retrieved from www.sawtoothsoftware.com
Corrections
The errors noted in this article are corrected. The following are the descriptions:
1. In the last sentence of the 2nd paragraph in the introduction, the words are "really value"
and not "appreciate value".
2. Under conjoint analysis methodology, there is no negative sign in the first equation. It is
only addition.
3. The 2nd equation on page 4 was completely omitted. After the sentence "The utility
associated with a profile is given as ": This has been inserted.
4. On that same page 4, the second "Sawtooth Software" with no year is corrected
5. In table 3, the word under G is Multipurpose and not multiple. The abbrevation A/L is
corrected in the interpretation under the table 3 which was showing A/C. ln addition, the
figures for serial number 10 were misplaced. K=4, L=80 and M=0.05.
The errors are sincerely regretted.
Production Editor
International Journal of Economic Development Research and Investment Vol 2 No. 3, Dec. 2011 9
