Special Topic in Business Analytics BAI 3399

Chapter 1: Introduction
----------MARKETING---------------------------------------------------------------------------------------
- Marketing Research: any technique or a set of practices companies use to collect
information to understand their target market better.
--It can be Qualitative (Understanding/exploratory), or
Quantitative(conclusive/decision-making).
- Strategic Marketing: This consists of Segmentation (homogeneous groups), Targeting
(choosing a segment), Positioning (occupying an image in the market)
- Marketing Operations: Price, Place, Product, Promotion
P.S.: the difference between descriptive (mean, std, variance) and inferential (Hypothesis
testing, ANOVA, regression) statistics. Verbatim is data collected as texts during qualitative
research.
Marketing Management is the practical application of marketing techniques to identify,
create, keep, and get customers.
The 4 P’s are decision variables, under the control of the company whereas the 4 C’s are the
driving forces of picking the marketing strategy.
Profitability is how much you make from a sale after you subtract everything it costs to
make that product, except NGO (Non-Governmental Organizations) like amnesty and social
marketing.
CRM customer relation management which is a database about customers like Salesforce
and HubSpot.
Pricing Strategies: penetration, skimming(écremage), economy, premium.
P.S.: The top of mind is a genius positioning strategy (thirst=Coca-Cola)
People are not always rational in their decisions, so we need to understand the customers.
(having the “best” product is just not enough)
- What value do we derive from consumption?
Objective: 1- Utilitarian
Subjective: 4- Esthetic (& ethical?)
3- Symbolic
2- Experiential
Willingness to pay is the maximum amount of money a customer is willing to pay for a product
or service. It is a common metric measured in pricing research studies.
Trade marketing is a strategy focused on wholesalers, retailers, and distributors rather than
consumers, with the goal of increasing demand with supply chain partners and getting products
in front of consumers.
The distribution panel provides continuous tracking of product sales to consumers based on
information gathered at the retail point-of-sale. (to follow-up)
Promotion can be emotional, cognitive, or behavioral. The nudge theory is based upon the idea
that by shaping the environment, also known as the choice architecture, one can influence the
likelihood that one option is chosen over another by individuals.
Sales funnel management describes the process of optimizing the customer journey from first
contact to purchase. The AIDA model describes the four stages a consumer goes through before
making a purchasing decision. The stages are Attention, Interest, Desire, and Action.
-----MARKETING ANALYTICS IN TODAY’S WORLD--------------------------------------------------------------
New challenges for companies:
- A rapidly saturated basic market demand.
- Increased direct, indirect, and peripheral competition.
- Products with shortened life cycles.
- Consumers are increasingly better informed and solicited.
Decisions to make:
• Advertising design/ Segmentation/ Campaign effectiveness/ Targeting
• Sales channels/ Positioning/ Budgets/ Marketing Mix/ Market size
Decision environment:
• Information overload
• Insight penury
• Fast! Faster! Fastest!
Response Models:
• Issue: Is company getting the best efficiency and effectiveness on its promotional spending
across US markets?  Approach: Used market response analysis and resource allocation to
conclude that Heinz: was misallocating spending/ could substantially reduce overall spending
without sacrificing national market share.
• Issue: Running out of good sites for typical full-service Marriott hotels  Approach: Conjoint
analysis to determine customer preferences, critical information for hotel design.
• Issue: Sales force size expanding approximately 10% per year: was size and
allocation the best?  Approach: Used resource allocation to determine the optimum sales
force size and allocation across products and markets.
• Issue: Forecasting the time path of sales of DirecTV before introduction  Approach: Used
Bass diffusion model. Market size estimate from customer survey/ Diffusion parameters
estimated from managerial judgments and analogous products (cable TV).
• Issue: Exelon wanted to differentiate itself from other second-tier companies  Approach:
Used positioning analysis to determine that:
– Customer preferences were more associated with the characteristics and offerings of second
tier companies, especially price.
– Analysis identified opportunity for re-position towards more customer focus, reliability, and
value.
• Issue: Soliciting donors is expensive, and donors are more and more sensitive to wise
spending of donation money Approach: Used customer choice model to identify best
potential donors, based on past data, and solicit only those donors who are likely to be
profitable.
-What is a model?
A model is a stylized representation of reality that is easier to deal with and explore for a
specific purpose than reality itself. Types: verbal, graphic, box and arrow, mathematical.
Kinds of models: descriptive/predictive
-Why use a model? Gain additional insight/Explore more options/Help reach group consensus,
support group decisions/Make more consistent, better decisions.
-How to build one? Specify (Variables, Relationships, interactions, dynamics) Calibrate
(Statistical estimation with real data (econometric approach), Judgmental calibration (tribal
wisdom approach)) Validate (Global fit (R², model fit), Variable significance (correct signs, t-
tests), Face validity (does it make sense?) Apply (Unique vs. multiple objectives? Short term vs.
long term?)
Ps.: Models are not to be trusted, they are to be used, they do not make decisions. Managers do.
Tertium Quid is a third thing that is indefinite and undefined but is related to two definite or
known things.
-----REGRESSION REMINDER AND BASICS---------------------------------------------------------------------
1. **Role of Linear Regression**: Linear regression models the relationship between the
dependent variable Y and one or more independent variables X using a linear equation.
2. **Dependent Variable (DV)**: Denoted as Y, it's the variable being predicted.
3. **Independent Variable (IV)**: Denoted as X, it's the variable used to predict the dependent
variable.
4. **Mediator**: In linear regression, we usually focus on direct relationships between IVs and
DV. However, if there's a mediator, it could be introduced in more complex models like
mediation analysis.
5. **R-squared (R^2 )**: R^2 is calculated as the proportion of the variance in the dependent
variable that is predictable from the independent variables.
Where SSE is the sum of squares error and SST is the total sum of squares.
6. **t-test**: The t-test assesses the statistical significance of individual regression coefficients
(slopes). The formula for the t-test statistic for a coefficient beta i is:
Where hat{beta_i} is the estimated coefficient and SE (hat{beta_i}) is the standard error of
the coefficient.
7. **Intercept**: The intercept beta_0 is the value of the dependent variable when all
independent variables are zero. It's also calculated as part of the linear regression.
8. **Slope**: For a simple linear regression with one independent variable, the slope beta_1 is
calculated using the formula:
Where bar{X} and bar{Y} are the means of the independent and dependent variables,
respectively.
9. **Sum of Squares Error (SSE)**: SSE is calculated as the sum of the squared differences
between the actual values of the dependent variable and the predicted values from the
regression model.
10. **Sum of Squares Regression (SSR)**: SSR is calculated as the sum of the squared
differences between the predicted values of the dependent variable and the mean of the
dependent variable.

Chapter 2: Segmentation
Segmentation is the process of categorizing customers into groups (a.k.a. segments,
clusters). So that customers within a segment are similar enough to be treated similarly
(economies of scale, coherence), yet different enough from customers in other segments
(worthy of differentiation).
A perfect segment is: • Distinct from the other segments: If two segments have similar
customers, why treat them separately?
• Homogeneous: Otherwise, customers within that segment cannot be treated the same way,
which defeats the purpose of segmentation
• Identifiable: A segment is not conceptual only, it’s also operational and if you can’t
identify/target members of a segment, its value is low
• Substantial: Not necessarily in terms of number of customers, though and worth being treated
as a separate segment
• Useful, operational, and informative: Too many segments kill the purpose of segmentation.
What’s in it? What can I learn from it? How can it help my marketing strategy?
P.S.: There is no such thing as an optimal segmentation. But some segmentations are good and
appropriate for some specific purposes.
The uses of segmentation:
(Short Term): Salesforce allocation, call planning, and direct marketing targeting, channel
assignment, communication program. Pricing, and Today’s competitors and my current relative
advantage to the customer.
(Long Term): Company strategy and positioning development, product development, find new
markets, niches, identifying and satisfying customer needs in different ways, emerging needs,
identifying a small group of today’s customers who might be representative of future broad-
base needs, or future potential.

Managing segmentation: Takes 7 steps:

1. Define segmentation problem: Internal assessment and planning (Objective(s) of
segmentation? (critical step!), Resources?, Constraints?) Database review (Primary data
available? Secondary data available?)
2. Identify data needs: What kind of data is most suited to answer your business problem: Socio-
demographic, Firmographic, Past behavior, Purchase data, Scanner data, Deep needs, Benefits
sought, Perceptions. (slide 98)
3. Conduct market research:
• Qualitative study (Interviews, sources, materials, “Deep needs” identification
• Quantitative study (Sample design (which respondents/informants?), Questionnaire
development, Data collection)
4. Build segmentation database: Assemble different data sources, with two types of data:
 Segmentation variables (bases)
– Characteristics that tell us why segments differ (e.g., needs, wants, benefits, solutions to
problems, usage situation, usage rate, decision processes, past behavior)
 Discriminant variables (descriptors)
– Characteristics that tell us how to find, reach, identify segments (age, income, education,
profession, lifestyles, media habits, use occasions, industry, size, location, organizational
structure)
5. Define market segments/6.Describe market segments: Two core approaches to
segmentation: Need-based segmentation and Choice-based segmentation.
• How many segments?
• How are they defined? (Segmentation variables)
• How can they be described, reached? (Discriminant variables)
7. Implement results

Need-based segmentation is when customers are grouped together based on their

similarities of Needs, Wants, Lifestyles, Behavior. (slide 104-119)
The dendogram shows how much distance separates the next two closest segments.
• The value on the y-axis indicates how much distance we need to travel to join two segments
• i.e., how much information is lost when two segments are grouped together
----Once segments are formed, describe them by their means
----Name the segments
----Often, surveys include additional information about respondents, not related to their needs,
but informative, called discriminant variables (descriptors)
• They do not describe what they want or why they buy They describe who they are and how
they can be reached:
– B2C: Age, Sex, Profession, Income, Number of children, Magazines
they read, TV shows they watch, etc.
– B2B: Industry, Revenues, Sales, Number of employees, etc.
P.S.: Segments should not be formed on these variables!
(the rest of the slides are graphs)

Chapter 3: Targeting
Targeting is which segment to go after? (with what product/offering, at what price,
communicating primarily on what benefits, etc.)
-3 targeting strategies: Undifferentiated, differentiated, and focused.
-The appeal of a segment is based on:
• Overall appeal of that segment (Size, Growth rate, Margins, revenues, Volumes) aka:
“External/objective” strength
• Relative advantage of the firm to serve this segment (Access, channels, Brand, reputation,
Product, service fit, Production capabilities...) aka: Internal/subjective” strength
Both dimensions (internal and external strengths) have characteristics that make them difficult
to assess because they are complex, the dimensions can be weighted differently, and the need of
subjective assessment.
The GE McKinsey Matrix has been widely used to measure the relative appeals of business units
or divisions and a multi-dimensional version of the much simpler BCG Matrix.
• Assessed by:
– Market attractiveness (“external” appeal)
– Business unit strength (“internal” appeal)
The GE/McKinsey matrix in 6 steps:
1. List segments, and estimate size of that segment
2. List dimensions that drive segment attractiveness (Size, volume, growth, margins, current
competition...)
3. List dimensions that underlie the competitive advantage of the firm (Product fit, access, brand
reputation, current penetration...)
4. Rate each segment on each dimension (1=worst, 5=best) (Forces group discussion, builds
consensus)
5. Weigh each dimension (1=least important, 5=most important) (Again, forces group
discussion, builds consensus)
6. Compute simple summed multiplications (sigma(ratings)*weights), and plot.

Chapter 4: Models and Scoring

Need-based segmentation observes segmentation variables/bases (needs, profile,
behavior) and groups customers based on similarities.
• The segments need to be: Manageable, Homogeneous, Distinct. (Once defined, describe and
name segments).
Data needs
DATA NEEDS: In need-based segmentation, we carefully divide data between segmentation and
discriminant data, and build segments based on the first only...
• ...Because we don’t want to build segments on the base of irrelevant data
• Need-based segmentation is appropriate to understand similarities of profiles but not always
appropriate to answer questions such as:
– Will they buy my product or not?
– What factors drive their choices?
– Which brand will they pick (out of many)?
Because similarities might not predict choices well
 Choice-based modelling is the process of categorizing customers into groups (a.k.a.
segments, clusters) based on their response likelihood (to buy/not to buy, to choose brand
A/B/C).
DATA NEEDS: In a typical choice-based segmentation, the model decides what’s relevant, and
everything becomes segmentation data (to predict likelihood of response)
CART (for classification and regression tree) is one of the most popular choice-based
segmentation tools.
1. Select the predictors (segmentation variables).
2. Select the response, to be predicted/explained.
3. Group all the respondents/customers into one big pool the “parent node”.
4. For each available predictor, one by one, split the population (the parent node) into
subgroups (the child nodes), and check to what extent the child nodes are more homogeneous
(within) more distinct (between).
5. Keep the split that works best measured by a statistical index, such as Gini index
6. Repeat for each child node
7. Stop when some criteria are met (No further improvement)
Choice modeling attempts to model the decision process of an individual or segment via
revealed preferences or stated preferences made in a particular context or contexts. Typically, it
attempts to use discrete choices in order to infer positions of the items on some relevant latent
scale.
Components of choice modeling:
• Observed choice: Buy / not buy (e.g., direct marketers)
• Predictors: Demographics, Attitudes, perceptions, Market conditions (price, promotion, etc.),
Past behavior, pattern of previous choices
• Link between the two:
– The model predicts customers’ probabilities of purchase...
Regression, intercept, slope, correlation coefficient...etc.
The logistic regression model Predictor (X) Response (Y) (Z-score and p value use).
 Applications of Choice models:
1- Calibrate model coefficient on a sample of customers.
2- Apply choice model to a larger list of customers.
3-Target customers based on predictors.
Dividing them into score classes (score= likelihood of donation*donation amount)
Score classes vs. choice -based segmentation:
- Similarities: they group customers based on the likelihood of choices, simple to use.
- Differences: score classes are more complex (it needs regular updates, completer and
more accurate, hard to intuitively implement).
Choice-based segmentation and need-based segmentation are two different approaches used in
marketing to understand and categorize consumer behavior and preferences. Here's a
breakdown of each:
IN CONCLUSION
- Choice-based segmentation focuses on observable behavior, such as purchasing decisions,
while need-based segmentation delves into underlying motivations or needs that drive
behavior.
- Choice-based segmentation is more data-driven, relying on actual consumer choices and
behaviors, while need-based segmentation often involves qualitative research to uncover
underlying needs.
- Choice-based segmentation is useful for understanding how consumers respond to existing
offerings and marketing efforts, while need-based segmentation helps identify unmet needs or
opportunities for innovation.

Chapter 5: Positioning
Positioning (unique place in the mind of the customer) captures the way customers
develop their perceptions and how to position the product depending on this perspective. In
addition, it evaluates the current offerings with the potential opportunities. Positioning is the
link between perceptions and preferences to explain behaviors.
P.S.: Soft Power is the ability to co-opt rather than coerce. It involves shaping the preferences of
others through appeal and attraction.
Positioning Map (perception map) is a 2D visualization to compare the product to its
competition, they are positioned along relevant dimensions (Red flag: avoid the center of the
map).
Gartner Magic Quadrants offer visual snapshots, in-depth analyses and actionable advice that
provide insight into a market's direction, maturity, and participants.
P.S.: White Paper is a document between B2B that includes terms and conditions.
--------------UNDERSTANDING CUSTOMER PERCEPTION-----------------------------------------------------
Correspondence Analysis Allows the reduction of an important number of variables into sub
dimensions by analyzing their correlation.
---P.S.: Correlation is a statistical measure that describes the extent to which two variables
change together. It indicates both the direction and the strength of the relationship between
two variables. There are several components and metrics associated with correlation:
1. Direction of the Relationship: Correlation can be positive, negative, or zero:
- Positive correlation: As one variable increases, the other variable also tends to increase.
- Negative correlation: As one variable increases, the other variable tends to decrease.
- Zero correlation: There is no systematic relationship between the variables.
2. Strength of the Relationship: The strength of the relationship is indicated by the correlation
coefficient, typically denoted by r. The correlation coefficient ranges from -1 to 1:
- r = 1: Perfect positive correlation.
- r = -1: Perfect negative correlation.
- r = 0: No correlation.
3. Components: Correlation involves two variables, often referred to as X and Y. Each pair of
observations (Xi, Yi) contributes to the calculation of the correlation coefficient.
- Pearson Correlation Coefficient r: Pearson correlation measures the linear relationship
between two continuous variables. It's calculated as the covariance of the variables divided by
the product of their standard deviations.
- Spearman Rank Correlation: Spearman correlation assesses the monotonic relationship
between two variables, i.e., whether the variables tend to change together in the same
direction, but not necessarily at a constant rate. It's calculated based on the ranks of the
observations-------------------------------------------------------------------------------------------------------------
-Linking Brands to Dimensions:

Multidimension scaled.
Attributes in positioning are characteristics of the products.
The difference between attributes and dimensions are the labels of axis.