Conjoint Analysis

Amar Saxena
[email protected]
+91.993.002.2910 25th Oct 2024
 A Short Survey … to start our session !!
• We would be grateful if you could answer a short survey for a
restaurant. The aim is to understand the importance of various
attributes when deciding which restaurant to go to.

• Please rate the importance of these attributes on a 10-pt scale

1: Least Important 10: Very Important

Customer Service
Quality of Food
Ambience
Food Options
Price

Slide
Slide22
Buyers want all of the
desirable features at the Sellers want to maximize
lowest possible price profits by: minimizing costs;
providing products/services
that offer greater overall
value than the competition

How to determine the real

preferences of the buyers?
Slide33
 Market Place – Different Perspectives, Different Goals
• Buyers want all of the most desirable features at the lowest
possible price

• Whereas, sellers want to maximize profits by:

– minimizing costs of providing features
– providing products/services that offer greater overall value than
the competition

• Typical market research role is to focus first on the demand side

of the equation
• After figuring out what buyers want, next assess whether it can
be built/provided in a cost-effective manner

But what is the best way to determine buyer preferences?

Slide4 4
Slide
 How to Learn What Customers Want?
• Ask Direct Questions about preference:
What brand do you prefer? What Interest Rate would you like?
Preferred Annual Fee? What Credit Limit would you like?

• Answers are often trivial and

unenlightening. Often have
very low discrimination.
(e.g. you will prefer low fees to
high fees, higher credit limits to
low credit limits)

Answers definitely are not very actionable

Slide
Slide55
Conjoint Analysis
Conjoint Analysis within marketing was developed back in the
1970s. The word conjoint derives from the idea that buyers
evaluate an overall product or service based on its multiple
conjoined attributes (features)

Conjoint
1. Joined together; combined.
2. Of, consisting of, or involving two or more combined or
associated entities; joint.
--The American Heritage College Dictionary

A research technique that lets you determine the

combination of features your product (or service) must have,
to appeal to a specific market or market segment.
Slide
Slide66
 What is Conjoint Analysis?
• Conjoint analysis questionnaires present respondents with
realistic tradeoffs, such as...

• If you choose left, you prefer Power.

• If you choose right, you prefer Fuel Economy.

Rather than asking directly about preference for Power vis-à-vis

Fuel Economy, realistic tradeoff scenarios are presented and
preferences are inferred from the product choices.

Slide
Slide77
UNDERSTANDING CONJOINT

Slide88
Suppose we want to market a new mobile phone. What are important factors?

We know from experience and from talking with consumers that there are three
important product features:
Internal Memory RAM Price

We further know that there are alternatives for each of these features, for instance:
Internal Memory RAM Price
256GB 6GB ₹ 10,000
128GB 4GB ₹ 20,000
64GB 3GB ₹ 30,000

What would be market’s “ideal” mobile phone for consumers?

And what will be the “ideal” mobile phone from a cost of manufacturing
perspective?
Assuming that it costs less to produce a phone that has lesser features.

Here’s the basic marketing issue: We’d lose our shirts selling the first mobile
phone and the market wouldn’t buy the second. The most viable product is
somewhere in between, but where? Conjoint analysis lets us find out where.
Traditional research might start by considering the rankings of the various
features of mobile phone.
Rank Internal Memory Rank RAM
1 256GB 1 6GB
2 128GB 2 4GB
3 64GB 3 3GB

This information doesn’t tell us anything that we didn’t already know about
which mobile phone to produce.
Now consider the same two features taken conjointly. The table here shows the
ranking of all the 9 possible products for two buyers assuming price is the same
for all combinations.

Both buyers agree on the most and least preferred mobile phone. But as we can
see from their other choices, Buyer 1 tends to trade-off mobile phone internal
memory for RAM, whereas Buyer 2 makes the opposite trade-off.
The knowledge we gain in going from 1st figure to 2nd figure is the essence of
conjoint analysis. If you understand this, you understand the power behind
this technique.

Next, let’s figure out a set of values for Internal Memory and a second set for
RAM for Buyer 1 so that when we add these values together for each mobile
phone and reproduce Buyer 1's rank orders.
One possible scheme:

We could have picked many other sets of numbers - and they would have worked.
So there is some arbitrariness in the magnitudes of these numbers even though their
relationships to each other are fixed.
Next suppose that table below represents the trade-offs Buyer 1 is willing to
make between RAM and Price. Starting with the values we just derived for
RAM, the table shows a set of values for Price that when added to those RAM
reproduce the rankings for Buyer 1 in the table.
We now have a complete
set of values that capture
Buyer 1's trade-offs.
Referred to as “utilities”
or “part-worths”)

Let’s see how we could use

this information to
determine which mobile
phone to produce. Suppose
we were considering one of
two mobile phones.

The values for Buyer 1

when added together give
us an estimate of his
preferences. Applying these
to the two mobile phones
we are considering, we get
the results in this table.
We’d expect buyer 1 to prefer the RAM mobile phone over the
Internal Memory mobile phone since it has the larger total value.
It’s easy to see how this can be generalized to several different
mobile phones and to a representative sample of buyers.
Products/Services – Composed of Features/Attributes
The Basic Premise
• Buyers value products or services based on the sum of the parts.

• Credit Card:
– Brand + Interest Rate + Annual Fee + Credit Limit

• On-Line Brokerage:
– Brand + Fee + Speed of Transaction + Reliability of Transaction +
Research/Charting Options

Can we learn how buyers value the separate components of an offering?

If so, then we are in a better position to design products that improve
profitability.

Slide 16
Slide 16
Why Does Conjoint Analysis Work Well?

• Asks respondents to make tradeoffs similar to those

in the real market.

• Respondents are discouraged from telling that all

features are equally desirable.

• When respondents are forced to make difficult

tradeoffs, we learn what they truly value

Slide 17
Slide 17
 How Conjoint Analysis Is Used in Industry
• New Product Introduction
• Pricing Research
• Brand Equity Research
• Market Segmentation
• Product Positioning/Line Extensions
• Employee Research (benefits, retention, etc.)

• Other interesting uses: Litigation (assessment of damages),

environmental impact studies, capital budgeting, patient/
physician communications, job search/hiring, agriculture etc.

Conjoint analysis may be considered almost anytime

to study situations in which consumers face tradeoffs!
Slide 18
Slide 18
• Conjoint Analysis is a Decompositional model
– Only the respondent’s overall preference is required

Majority of the other techniques are Compositional models

– E.g. Regression, Discriminant

• A separate conjoint model can be created for each and every

respondent

Slide 19
Slide 19
AN EXAMPLE

Slide 20
20
 What credit card to offer?
• Imagine that a credit card company is interested in how
consumers trade off various aspects of credit card offerings:
specifically, the brand, annual fee and credit limit.

An Attribute List
• The first step in conducting a conjoint analysis study is to
develop a list of attributes and levels of interest.
Annual Fee: Credit Limit:
Brand:
No annual fee $1,000 credit limit
Visa
$10 annual fee $2,500 credit limit
MasterCard
$20 annual fee $5,000 credit limit
Discover

So there will be 3 x 3 x 3 = 27 possible credit card combinations.

With conjoint analysis, we do not have to ask respondents to evaluate all
possible combinations.
How do we choose an efficient subset of possible cards to ask respondents?
Slide 21
Slide 21
 Experimental Design Plan
• This is done by FRACTIONAL FACTORIAL DESIGN.
– This feature is in all the softwares. Can be used to develop the
product combinations on which to get data from respondents.
• Efficient tradeoff questions lead to precise estimates of
respondents' preferences.
No $10 $20 $1,000 $2,500 $5,000
Master-
Visa Discover Annual Annual Annual Credit Credit Credit
card
Product Profiles Fee Fee Fee Limit Limit Limit
Card # 1 ✓ ✓ ✓
Card # 2 ✓ ✓ ✓
Card # 3 ✓ ✓ ✓
Card # 4 ✓ ✓ ✓
Card # 5 ✓ ✓ ✓
Card # 6 ✓ ✓ ✓
Card # 7 ✓ ✓ ✓
Card # 8 ✓ ✓ ✓
Card # 9 ✓ ✓ ✓
Slide 22
Slide 22
• Each row in this table represents a product concept we'll ask
respondent to evaluate. The X’s indicate what levels each card
uses. Card # 1
• For example, the first card is: Discover
$20 Annual Fee
$2,500 Credit Limit
• Some other cards would be –
Card # 3 Card # 8 Card # 6
Visa Mastercard Discover
$10 Annual Fee $10 Annual Fee No Annual Fee
$5,000 Credit Limit $2,500 Credit Limit $5,000 Credit Limit
• Make a card for each product combination.
• All the cards should be exactly the same.
• All attributes should be similar in the card. No attribute should
have greater importance or Visually depicted.
Slide 23
Slide 23
• Now ask the respondent to evaluate the nine cards from the
design plan showed earlier.

• Data to be collected by either of the following ways:

– Use a 1 to 10-point rating scale, where 1 means that the offer is
really terrible, and 10 means that the offering is very excellent.
or
– Rank all the profiles

• Please ask the respondent to answer carefully, as the results

depend on how well the task is performed.

Slide 24
Slide 24
 Your Conjoint Results

Part Worth Utilities:

Brand Annual Fee Credit Limit

Visa 6.67 No annual fee 9.00 $1,000 credit limit 3.67
MasterCard 7.00 $10 annual fee 5.00 $2,500 credit limit 6.33
Discover 4.33 $20 annual fee 4.00 $5,000 credit limit 8.00

Importance Scores:
Brand 22
Annual Fee 42
Credit Limit 36

Slide 25
Slide 25
 How Does Conjoint Analysis Work?
• Product features are varied to build many product concepts
• Respondents are asked to rate/rank the product concepts
(dependent variable)
• Based on the respondents’
evaluations of the product
concepts, we figure out how
much unique value (utility)
each of the features added
Regress dependent variable on
independent variables; betas
equal part worth utilities.

Slide 26
Slide 26
 Implementation – 4 Key Questions
Attributes to consider?
• What are the important attributes that could affect preference?

Levels of each attribute?

• How will the respondents know the meaning of each attribute
(factor)?

The profiles. Ranking Vs Rating.

• What do the respondents actually evaluate?

• How many profiles do they evaluate?

Slide
Slide 2727
 Steps to Conjoint Analysis Studies...
• Attribute List Formulation. A business problem is defined and
an attribute (features) list is developed to study the problem.
• Levels at which each Attribute will be measured.
• Create Profiles – combination of various levels of attributes.
• Data Collection. Ask respondents to express the trade-offs they
are willing to make among product features by rating or sorting
among hypothetical product concepts.

What do we show? What do we ask for?

• Product Features • Product liked by the
Camera, Processor, consumers
Color, Storage, Size • Products selected by
Etc. them from the options
• Price shown.
Slide
Slide 2828
 Steps in Conjoint Analysis

What do we show? What do we ask for?

X Y
• Product Features • Product liked by the
▪ Camera consumers
▪ Processor • Products selected by
▪ Color them from the options
▪ Storage shown.
▪ Size
▪ Etc.
• Price

Slide 29
 Steps to Conjoint Analysis Studies...
• Utility Calculation. A set of preference values or part worth
"utilities" is derived from the data; they reflect the trade-offs
each respondent made.
• Willing to Pay
• Market Simulation. Utility values are used to predict how
buyers will choose among competing products and how their
choices are expected to change as product features are varied.
– The value of conjoint analysis is its ability to estimate choice
behavior for a wide range of market scenarios.

Slide
Slide 3030
Insight into the output

• Part Worth Utilities

• Importance Scores

Slide 31
Slide 31
 Creating an Attribute List
• Selection of attributes is extremely important
– Should NOT be based on the important attributes of the product
– Focus on attributes important for managerial decision making

• Attributes should be relevant – they influence consumers’ utility.

– Qualitative surveys like FGD and IDPs can help.
• Attributes should be independent of each other – i.e. they
should measure independent aspects of the product.
– Brand, Speed, Color, Price, etc.
If attributes are interrelated, then certain combinations might be highly
unrealistic and confusing to the respondents (e.g. price and quality)
• Do not use too many attributes
– Creates longer questionnaire  Difficulty in data collection
Slide 32
Slide 32
 Creating an Attribute List
• Each attribute has varying degrees, or "levels.“ (at least 2 levels)
– Brand: Coke, Pepsi, Sprite
– Speed: 5 pages per minute, 10 pages per minute
Each level is assumed to be mutually exclusive of the others
• Should be easy to communicate
• Attributes should discriminate – ability to differentiate between
competitive offerings in marketplace.
• Attributes can be on a nominal or interval scale
– Attributes like brand and color are known as Nominal or Categorical
– Attributes like price and speed are on interval scale, and are known as
Quantitative attributes

• Do not use: Subjective Attributes/ Ambiguous Level Descriptors

– E.g. subjective attributes like style, aesthetic appeal
– Subjective levels like good, modern, low price etc.
Slide 33
Slide 33
 Issues in identifying the attributes
• Understand the inherent associations in the mind of the
consumers (Focus Groups can throw light here!!)
– E.g. Price-Quality relationship.

• Inherent correlation between attributes

– Existing relationship between attributes
• E.g. High Horsepower cannot go with high mileage
– One attribute could indicate presence or absence, but the other
indicates amount
Create super-attributes – create a new attribute e.g. performance

Slide 34
Slide 34
 Formulating Attribute Levels
• Levels should be mutually exclusive: a product concept can have
one and only one level of each attribute
Attribute: Add-on features
– level 1: Sunroof
– level 2: GPS System
– level 3: Video Screen
The attribute levels are not mutually exclusive. They can all be there in
the same car. If you define levels in this way, you cannot determine
the value of providing two or three of these features at the same time

• Levels should have concrete/unambiguous meaning.

– "Very expensive" vs. "Costs Rs 57.5 per kilo"
– "Weight: 5 to 7 kilos" vs. "Weight is 6 kilos“

• There should not be too large a gap between levels

Slide 35
Slide 35
• Balance the number of levels across attributes.
– Don’t include too many levels for any one attribute. The usual
number is about 3 to 5 levels per attribute.
E.g. The temptation is to include many, many levels of price, so we
can estimate people’s preferences at each price point
– If an attribute (say, Attribute A) is at many levels (say 8). Other
attributes (Attributes B, C & D), are at lesser levels (say 3), then the
Attribute A will automatically come out to be very important.
– For attributes like price, a better approach is to ask for a few levels,
and treat it as a linearly related variable.
• The levels should span a range that is larger than in reality, but
not substantially.
• Levels should be generally acceptable.

Slide 36
Slide 36
Attribute 1: Brand Attribute 5: Weight

1. IBM 1. Weighs 3 pounds

2. Dell 2. Weighs 5 pounds
3. Compaq 3. Weighs 7 pounds
4. Toshiba
5. Acer Attribute 6: Operating System

Attribute 2: Display 1. Windows XP Professional

2. Windows XP Professional + Microsoft
1. 14-inch display Office
2. 15-inch display 3. Windows XP Home
3. 17-inch display
Attribute 7: Exterior
Attribute 3: Processor Speed
1. Black composite exterior
1. 2.0 GHz Processor 2. Silver aluminum exterior
2. 2.4 GHz Processor 3. Gunmetal titanium exterior
3. 3.0 GHz Processor
Attribute 8: Price
Attribute 4: Battery Life
1. $800
1. Battery life: 3 hours 2. $1,200
2. Battery life: 4 hours 3. $1,600
3. Battery life: 5 hours 4. $2,000
5. $2,400
Create Profiles & Data Collection

• Full Profile
• Fractional Factorial Design

• Ranking of the profiles

• Rating of the profiles

• Least restrictive set of statistical assumptions

• Sample Size

Slide 38
Slide 38
 Part Worth Utilities
• Numeric values that reflect how desirable different features
are.

Feature -> Vanilla Choc ₹ 15 ₹ 25 ₹ 35

Utility -> 2.5 1.8 5.3 3.2 1.4

• Higher utilities are better.

• In some cases, part worth utilities can be positive or negative
values. The interpretation is the same (higher numbers are
better).

Slide 39
Slide 39
 Interpreting Part Worth Utilities
• Part-worths are estimated on an Interval scale
– Interval scales support addition/subtraction operations, but not
multiplication/division
– No meaningful zero point.
– So it is NOT appropriate to say that a part worth utility of 1.0 is
twice as preferred as a utility of 0.5.

• You cannot compare one level from one attribute with one level
from another attribute

• You CAN compare differences between two levels of one

attribute versus two levels of another attribute

Slide 40
Slide 40
 Attribute Importance
• A measure of how much influence each attribute has on
people’s choices
• Best minus worst level of each attribute, as percentage

• Importance are directly affected by the range of levels you

choose for each attribute.
• Importance run from 0 to 100 and are Ratio scaled, meaning
that you can apply multiplication/division operations. An
importance of 20 is twice as important as a 10.

Slide 41
Slide 41
 Issues with Attribute Importance
Some weaknesses with traditional conjoint importance scores:
• The importance of an attribute is directly affected by the range of
levels you choose to include in the study.
– So, you can make an attribute such as Price more important by
studying it on a wider range of prices.
– Do not take the importance number as-is, without consideration of
the way it is computed and the range associated with each
attribute. To say that brand makes up 30% of the decision, or that
Price has 40% importance makes little sense on its own.
• An importance score takes the full difference between best and worst
levels, irrespective of whether the "worst" utility is associated with the
expected worst level (such as the highest price level)
• An attribute of no consequence to a respondent will still take on a
positive importance score.
Slide 42
Slide 42
 Important Rule for Attribute Importance
• Importance can be very misleading if they are computed from
average population (or group) utilities.

• Compute importance for each individual separately, and then

average those results across all people.

Slide 43
Slide 43
MARKET SIMULATION

Slide 44
44
 Intro to Market Simulation
• Conjoint analysis recognizes that people make trade-offs, and
that different people make different trade-offs. If you know
what those trade-offs are, you have a powerful tool for
predicting choice behavior.

• From tradeoff questionnaires, we obtain part worth utility

values that summarize respondent preferences.
• Let us again take the example of ice-creams:

Feature -> Vanilla Choc ₹ 15 ₹ 25 ₹ 35

Utility -> 2.5 1.8 5.3 3.2 1.4

Slide 45
Slide 45
 Market Simulators ("What-If" Simulators)
• Offer a more powerful and intuitive way to present the results
of conjoint analysis projects to management

• Say, just two possible ice cream cone options are available in
the market. Which cone would this respondent choose?

Option 1: Option 2:
Vanilla Chocolate
₹ 25 ₹ 15

• We simply add up the part worth utility values for the levels
associated with each option. The option with the highest utility
is said to be "chosen."

Slide 46
Slide 46
• Respondent in the example "chooses" the higher utility option
2, the ₹ 15 Chocolate cone!

• Repeat for every respondent ... and summarize the results

across the sample.

35%
65%

Vanilla @ ₹ 25 Chocolate @ ₹ 15

Slide 47
Slide 47
 A Powerful "What-If" Tool
• With many attributes and many product concepts, in competition with
one another, many possible market scenarios can be tested.
• No need to go back and interview respondents about every new choice
scenario.
– Simply add the utility of each potential product offering to project the
likely choice of the customer.

• The resulting choice shares ("Shares of Preference") range from 0% to

100% and are simple for managers to understand.

"A good market simulator is like having all of your respondents gathered in one room for the sole
purpose of voting on product concepts and competitive scenarios (defined in terms of the attribute
levels you measured) you show them. You walk into the room, show them a market scenario (i.e.
products A, B and C), and they vote for the one(s) they prefer. Many potential products and market
situations could be evaluated, and your captive audience would never get tired, ask for lunch breaks,
or require you to pay them by the hour."
Slide 48
Slide 48
 Why Conduct Simulations? Competitive Effects
Matter!
• Let's consider that you conducted a conjoint analysis study and found that 80%
of the market prefers round widgets, and 20% prefers square ones.
(So the part worth utility for round widgets vastly exceeds that for square widgets).

What should you do? Which should you take to market?

• In the absence of any other information, round would be the logical choice
• But what if there are 10 competitors in the market, ALL only offering round
widgets?
• Market simulation allows to conduct what-if games to investigate issues such
as new product design, product positioning, and pricing strategy.
• Simulations can reflect real-world behavior
– Represent idiosyncratic preferences of segments and individuals
(remember, you don’t have to appeal to the "fat" part of the market to
carve out a profitable business)
• Results expressed in terms that make sense to management and are
actionable
Slide 49
Slide 49
 Assumptions of Conjoint Simulators
• Market simulators are very powerful tools. Use them wisely. Be
aware that they make a number of assumptions.

• Because each brand and product feature is equally available in

the conjoint survey, we assume equal distribution.
• Other key assumptions include:
– Perfect information
– No out-of-stock conditions
– Equal time on market to achieve market potential
– Equal effectiveness of sales force and advertising
– We have interviewed the right people
– Each respondent is in the market to buy
– We have included the right attributes and levels
– No switching costs
Slide 50
Slide 50
 Variations in Market Simulation
• Begins by computing the overall desirability of each of multiple
product options, simply by adding the part worth utility values.
• Assume that the total utility (after summing across the part-
worths for each product alternative) for three alternatives for
a single respondent was

• There are many ways to convert those overall utility scores for
product alternatives to probabilities of choice (or "shares of
preference"). Common methods are:
– Maximum Utility Rule
– Bradley-Terry-Luce (BTL) Rule
– Logit Rule
Slide 51
Slide 51
 Variations in Market Simulation
• Maximum Utility Rule:
Simply assume that each respondent will choose the product alternative
with the highest utility. In the example above, Product A would capture
100%. Repeat the same for all respondents, and summarize the overall
sample's shares of preference.
• Bradley-Terry-Luce (BTL) Rule:
The likelihood of choosing a product alternative is equal to its total utility
divided by the sum of all utilities for products in the simulation scenario.
In the example, the likelihood of this respondent choosing Product A is equal
to 3.2 / (3.2 + 2.6 + 0.3) = 52.5%. Product B's choice likelihood is 2.6 / (3.2 + 2.6
+ 0.3) = 42.6%, etc.
• Logit Rule:
Just like the BTL rule, except that product utilities are first "exponentiated"
prior to applying the rule. "Exponentiation" means to take the antilog of the
utility, or to raise the constant "e" to the power of the utility.
Under the Logit rule, the probability of this respondent choosing product A
is equal to 24.53 / (24.53 + 13.46 + 1.35) = 62.3%.
Slide 52
Slide 52
 Market Share – Logit Rule
Market Share =
• Step 1 – Calculate utility of all the options
• Step 2 – Exp(Utility)
• Step 3 – Market Share is the proportion of number found in
Step 2.
EXHIBIT TN-2: CURRENT VIDEO STREAMING INDUSTRY MARKET SHARES

Streaming Option Utility EXP(Utility) Percentage of

Market
FilmCast HD 0.27 1.31 25.3%
FilmCast Ultra HD −0.04 0.96 18.6%
Videosource 0.46 1.58 30.6%
Webflickstream 0.28 1.32 25.6%
Total 5.18

Slide 53
 Willingness to Pay
• Price vector has to be there

• Calculate the range of utility for the Price attribute

• Divide by the price differential
• This ratio is the “price per point”
• Multiply this ratio with the utility of any level of other attributes
to get the Willingness to Pay.
Partworth utility for
WTP for level m level m of attribute n
of attribute n Utility vector for
the price attribute

Slide 54
TYPES OF CONJOINT

Slide 55
55
Flavours of Conjoint Analysis

Slide 56
 Various Response Types
• Rating-based conjoint: Respondents rate the product
alternatives shown. This can be on a scale of 0 to 100.
Alternative: Constant Sum.
• Best-worst conjoint (MaxDiff Analysis): Respondents are asked
to indicate which option is the best and which option is the
worst among the alternatives.
• Ranking-based conjoint: Respondents rank alternatives from
best to worst.
• Constant-sum conjoint (Chip-allocation): Respondents are
exposed to a number of different scenarios (tasks), and are
asked tο indicate how many times would they purchase one of
the projected alternative choices, for a constant number of
future purchases. The purchase volume is predefined before
the start of the experiment.

Slide 57
 Various Response Types …
• Volumetric Conjoint: Respondents are exposed to a number of
different scenarios (tasks), and are asked to provide the exact
quantity that they would purchase each of the illustrated
alternative choices.

• Choice-based conjoint (CBC): Respondents are asked

to choose which option they will buy or otherwise
choose. This is the most theoretically sound, practical,
and popular method of conjoint analysis.

Slide 58
 Traditional Full-Profile Conjoint Analysis
• Full-Profile refers to the idea that each product profile is
described using all the attributes involved in the study.
• Often referred to as "Card-Sort Conjoint Analysis," because
respondents actually received a deck of cards, with a conjoint
profile printed on each one.
– Respondents were asked to rank or rate the cards, such as shown
here for a Computer study on five attributes.
– The credit card example we showed earlier in this training module
is also an example of Traditional Full Profile Conjoint Analysis.

Slide 59
Slide 59
 Strengths & Weaknesses
• Traditional Full-Profile Conjoint Analysis is still used today, but
not as often as other approaches to conjoint analysis.

Strengths:
• Ability to handle a wide variety of problems, including pricing
research
• Simplicity of study execution and utility computation
• The technique leads to individual-level utilities

Weaknesses
• With more than about six attributes, it can become complicated
for respondents
• Sorting or rating cards doesn't come natural for many
respondents
• Since it uses a rating or ranking scale, the estimated part-worths
are not naturally scaled to be predictive of choice probabilities
Slide 60
Slide 60
 Choice-Based Conjoint Analysis
• In recent years, Choice-Based Conjoint, has flourished, and is
now the most popular approach in the world.
• Rather than ask respondents to rate or rank product profiles,
respondents are simply asked to choose their favored product
concept from a set of available options:

Slide 61
Slide 61
Choice-based Conjoint (CBC)
Options shown to respondents

Slide 62
 Adaptive Conjoint Analysis
• A computer-based form of conjoint analysis
• The interview has three main stages. Respondents are asked to:
– Rate or rank levels within each attribute, based on preferences.
– Rate the importance of each attribute, using a simple rating scale.
– Based on the previous answers (in steps 1 and 2), respondents
compare two product concepts at a time, described typically on
just 2 or 3 of the total list of attributes (a partial-profile display):

Respondents complete usually

20 to 30 such tasks.
The algorithm chooses tradeoff
questions to provide efficient
and informative data for
stabilizing the respondent's
part worth utilities.
Slide 63
Slide 63
 Which Method Should I Use?
Each brings unique benefits to different research problems:
• Traditional Full Profile Conjoint: Works well when the number of
attributes is no more than about six, and if the sample size is
particularly small (say, less than about 100).
• Choice-Based Conjoint (CBC): Argued to best mirror real-world
purchase processes and to have the highest validity of the three.
But, requires larger sample sizes to stabilize estimates. Also, it is
sometimes difficult for respondents to deal with more than
about six attributes in CBC questionnaires.
• Adaptive Conjoint Analysis (ACA): Often considered when there
are too many attributes to study using the other techniques. But,
not considered a good tool for pricing research, because it often
understates the sensitivity of respondents to price.
Slide 64
Slide 64
 Is Conjoint panacea for all issues?
Conjoint analysis simulates a purchase decision where a customer
compares similar products. It takes a very rational view that
customers decide which product is their favorite by comparing
individual attributes like price, brand, and features. Hence, conjoint
is only suited for following research scenarios:
1. That simulate a simple, well-considered purchase between
similar products.
2. Where only one person is involved in the purchase decision.
3. Where customer already knows what kind of product they need.
4. Where you know which attributes the customer use to compare
products.
5. Where those attributes have no overlap in meaning or possible
options.
Slide 65
 3 areas suited for Conjoint Analysis
(i) Calculating Willingness-to-Pay For Existing Features
https://youtu.be/1hWvsL-gOFo?feature=shared

(ii) Assembling The Optimal Product Mix / Bundling Strategy

https://youtu.be/0yMRSgZpYg0?feature=shared

(iii) Modelling Market Share Scenarios

Slide 66
 10 areas NOT suited for Conjoint Analysis
1. Simple Trade-Offs
2. One Category
3. Impulse Purchases
4. Complex Purchases
5. Intangible Attributes
6. No Benchmark Price
7. Emerging Opportunity
8. New Features
9. Attribute Quantity
10. Relationship Complexity

Slide 67
Slide 68
Conjoint Analysis Methods

Slide 69
Three Main Flavors of Conjoint Analysis

• Traditional Full-Profile Conjoint

• Choice-Based Conjoint (CBC), also known as Discrete Choice

Modeling (DCM)

• Adaptive Conjoint Analysis (ACA)

Slide 70
Slide 70