Conjoint
Analysis Example:
Pricing a New Product
Pricing Strategy & Analysis
Federico Rossi
Purdue University
Krannert School of Management
Spring 2022
© Copyright Federico Rossi, 2022
Case
• Imagine a company is interested in marketing a new
cleaner to remove spots from carpets
• Management is interested in 5 attributes of the new
product which are expected to influence the
purchase decision of consumers:
– Package Design, Brand name, Price, Good Housekeeping
Seal, Money‐Back Guarantee
• Evaluate the optimal price and other characteristics
for this new product
1
Step 1: Identify the research purpose
and research questions
Find the optimal pricing for the new product
Step 2: Specify attribute variation and
levels
• Use consumer input to identify the important
attributes to be included.
• Find realistic ways to communicate the attributes and
levels to respondents.
– Audio‐visual aids
– Demonstrations
– Reference products
• Attributes and levels must be actionable.
• Profiling current products in the market.
2
Attributes and Levels of Carpet Cleaner
Attributes Levels
Package Design A, B, C
Brand Name K2R, Glory, Bissell
Price $1.19, $1.39, $1.59
Good Housekeeping Seal Yes, No
Money‐back Guarantee Yes, No
Total number of possible product combinations = 108
Step 3: Design of Stimuli
• Feasibility of the “full factorial” (all possible
combinations of attributes and levels) design
• “Fractional factorial” designs
• Orthogonal (zero correlations across attributes)
fractional factorial designs
3
0. RAW DATA: Orthogonal Fractional Design for
Carpet Cleaner
Respondent
Package Brand
Price $ GHKS MBG Evaluation
Design Name
(rank)
1 A K2R 1.19 No No 6
2 A Glory 1.39 No Yes 8
3 A Bissell 1.59 Yes No 2
4 B K2R 1.39 Yes Yes 17
5 B Glory 1.59 No No 5
6 B Bissell 1.19 No No 16
7 C K2R 1.59 No Yes 7
8 C Glory 1.19 Yes No 12
9 C Bissell 1.39 No No 10
10 A K2R 1.59 Yes No 1
11 A Glory 1.19 No Yes 11
12 A Bissell 1.39 No No 4
13 B K2R 1.19 No No 15
14 B Glory 1.39 Yes No 13
15 B Bissell 1.59 No Yes 14
16 C K2R 1.39 No No 9
17 C Glory 1.59 No No 3
18 C Bissell 1.19 Yes Yes 18
0. RAW DATA: Fractional Factorial Designs
A subset of 18 product combinations is selected for use
such that attributes are uncorrelated (orthogonal).
• Why 18?
– To allow sufficient degrees of freedom to estimate
the model for each individual respondent.
• Which 18?
– Commercial software available to pick subset.
4
1. CREATING DUMMIES: Dummy Variable Coding
Dummy Variables
Pack B Pack C
Pack A 0 0
Pack B 1 0
Pack C 0 1
• Yields a total of eight dummy variables for the five
attributes: (3‐1) + (3‐1) + (3‐1) + (2‐1) + (2‐1) = 8
1. CREATING DUMMIES: Dummy Variable Coding
Profile Pack B Pack C Glory Bissell $1.39 $1.59 GHKS MBG
1 0 0 0 0 0 0 0 0
2 0 0 1 0 1 0 0 1
3 0 0 0 1 0 1 1 0
4 1 0 0 0 1 0 1 1
5 1 0 1 0 0 1 0 0
6 1 0 0 1 0 0 0 0
7 0 1 0 0 0 1 0 1
8 0 1 1 0 0 0 1 0
9 0 1 0 1 1 0 0 0
10 0 0 0 0 0 1 1 0
11 0 0 1 0 0 0 0 1
12 0 0 0 1 1 0 0 0
13 1 0 0 0 0 0 0 0
14 1 0 1 0 1 0 1 0
15 1 0 0 1 0 1 0 1
16 0 1 0 0 1 0 0 0
17 0 1 1 0 0 1 0 0
18 0 1 0 1 0 0 1 1
5
2. CORRELATION BETWEEN ATTRIBUTES
Check Correlation Matrix
Pack B Pack C Glory Bissell 1.39 1.59 GHKS MBG
Pack B 1.0
Pack C -0.5 1.0
Glory 0.0 0.0 1.0
Bissell 0.0 0.0 -0.5 1.0
1.39 0.0 0.0 0.0 0.0 1.0
1.59 0.0 0.0 0.0 0.0 -0.5 1.0
GHKS 0.0 0.0 0.0 0.0 0.0 0.0 1.0
MBG 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.0
3. PATH WORTH ESTIMATES
Run Linear Regression for Single Respondent
Regression Statistics
Multiple R 0.99153625
R Square 0.983144135
Adjusted R Square 0.968161144
Standard Error 0.952579344
Observations 18
ANOVA
df SS MS F Significance F
Regression 8 476.3333333 59.541667 65.61734694 4.49243E-07
Residual 9 8.166666667 0.9074074
Total 17 484.5
Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 95.0% Upper 95.0%
Intercept 6.5 0.635052896 10.235368 2.94732E-06 5.063409447 7.936590553 5.063409447 7.936590553
Pack B 8 0.549971941 14.546197 1.47264E-07 6.755876086 9.244123914 6.755876086 9.244123914
Pack C 4.5 0.549971941 8.1822356 1.84812E-05 3.255876086 5.744123914 3.255876086 5.744123914
Glory -0.5 0.549971941 -0.9091373 0.386975857 -1.744123914 0.744123914 -1.744123914 0.744123914
Bissell 1.5 0.549971941 2.7274119 0.023323127 0.255876086 2.744123914 0.255876086 2.744123914
1.39 -2.833333333 0.549971941 -5.151778 0.000601697 -4.077457247 -1.58920942 -4.077457247 -1.58920942
1.59 -7.666666667 0.549971941 -13.940105 2.12822E-07 -8.91079058 -6.422542753 -8.91079058 -6.422542753
GHKS 1.5 0.476289672 3.149344 0.01174995 0.422557085 2.577442915 0.422557085 2.577442915
MBG 4.5 0.476289672 9.4480319 5.72823E-06 3.422557085 5.577442915 3.422557085 5.577442915
6
4. AVERAGE PATH WORTHS
Average parameters across individuals
5. ATTRIBUTE IMPORTANCE
Sample Level Attribute Importance
Sample Level Attribute Importance
0.4
0.35
0.3
0.25
0.2 Sample Level Attribute Importance
0.15
0.1
0.05
0
Attributes
7
6. MARKET SEGMENTATION BY MBG
Willingness to Pay
• Utility gain going from $1.39 to $1.19 = 2.83 utils
– 2.83 utils cost $.20
• Utility gain going from $1.59 to $1.19 = 7.67 utils
– 7.67 utils cost $.40
10
6 Consumer
Utils
4 preference is
2
non‐linear in price
0
1.19 1.39 1.59
Price
The conversion from dollars to utils can now be used to
compute the willingness to pay for other attributes
8
7. WILLINGNESS TO PAY
• Utility gain going from Package A to Package B = 8 utils
– If 2.83 utils costs $.20, then how much does 8 utils cost?
• Utility gain going from Package A to Package B = 8 utils
– If 7.67 utils costs $.40, then how much does 8 utils cost?
One can generate a range for consumer’s willingness to pay
for different attribute levels
Can be interpreted as conservative & best case prices
Applying Conjoint Results to Market
Simulation
• Conjoint results help understand consumer
preferences
– Importance of attributes
– Impact of different levels of an attribute
• Other big use is “what‐if” analysis to predict
the market impact of
– Adding a product to an existing set of products
– Deleting a product from an existing set of product
– Changing an attribute
9
From utility to purchase
The model allows us to derive the utility consumers
obtain from the available alternatives:
UA = 4.5, UB = 5.2, UC = 4.2
• How do we go from utilities to final choices?
First‐Choice Rule Share of Preference Rule
Pick the alternative generating the Add a random error to the utilities. This
highest utility. error captures other events not
considered by the model. Each alternative
Choose alternative B. has a probability of being chosen:
exp 𝑈
𝑃𝑟𝑜𝑏 𝑐ℎ𝑜𝑜𝑠𝑒 𝐵
∑ 𝑈
exp 5.2
0.54
exp 4.5 exp 5.2 exp 4.2
8. MARKET SIMULATION
• What if the following three products were offered in
the market? What would be their share?
– Product 1: Package B, Glory, $1.39, No seal, No MBG
– Product 2: Package C, Bissell, $1.39, Yes seal, Yes MBG
– Product 3: Package C, K2R, $1.59, Yes seal, Yes MBG
FIRST CHOICE RULE SHARE OF PREFERENCE RULE
Prod1 Prod2 Prod3 Prod1 Prod2 Prod3
0 1 0 0.018 0.980 0.002
0 1 0 0.002 0.997 0.001
0 1 0 0.018 0.982 0.000
10
Summary
• Conjoint Analysis can be used to conduct:
– Pricing
– Concept testing
– Market Segmentation
• During Lab session you will be given a similar case
for practice.
Review – “LINEST” function of MS Excel
11