MARKETING ANALYTICS

SPRING 2021
INTRODUCTION
version 2.0
‘Mike’ Minhi Hahn
[email protected]
I. Guideline for the Online Lecture

• Invitation to the lecture: By e-mail on Sundays & Tuesdays

• Recommend Video On & Audio Mute during the lecture

• Prepare to take memos in each class

• Using BeachBoard News and Content

• Upload File, Lecture File and “No Textbook but Recommended”

Guideline for the Online Lecture (continued)
• Questions: In class, e-mails, chats (Answer: In class, e-mails)

• Q&A Assignments every class

• Office Hours: 3:15-4:00 p.m. or by Appointments

• Preparations of Excel, Excel Tool Pack and Enginius Software

• Recorded Class Videos will be uploaded on Wednesdays

• Exercises & ‘Happiness Breaks’: For class stimulations

II. Topics of Lecture File 1 (LF1)

• LF1-1: INTRODUCTION (1/20 W)

• What is M.A.? / Why M.A.? / Marketing Analytics, Spring 2021

• LF1-2: MODELS, DATA and ANALYTIC TOOLS (1/25 M)

• Models/ Data/ Tools for Marketing Analytics / Preparations (Enginius)

• LF1-3: CONSUMER INSIGHTS (1/27 W)

• Consumer Insights/ Check Preparations/ Formation of Teams (Breakout)

4
 SIDE

JOEY GALLO
(Defensive Shift)

5
 (1) Defense against Joey Gallo

 Traditional versus Today’s data based defense in baseballs

▪ From: Mean, Dispersion and Distribution among a group of persons

▪ To: Mean, Dispersion and Distribution of each person

▪“Big data is about analyzing outliers, afterall.” (Greg Allenby)

6
RECOMMENDATIONS

7
 (2) Analysis of Market Baskets

• Traditional versus Today’s data based analysis of consumers

• Univariate or multivariate distribution among a group of persons

• Multivariate (including Bivariate) distribution of each person

=> Understand “Associations” among persons or items

• Analysis of interactions among people (among products)

• Analyzing “Closely connected” People or Things

8
SIDE

9
(3) Contests: Human beings vs. Computers

• Huge amount of available data + Speedy computer-aided tools

• Go (Baduk, Weiqi): A huge number of alternatives but No Uncertainty

• AlphaGo (or AlphaZero)

• Computer Aided Tools

• So, computer or human being?

=> Who will win games of marketing?
10
• Why M.A.?

Today’s Marketing Decision Environment

• Rich _________________ data

• Rich _________________ data

• Intelligent and Speedy ______________________ to analyze data

11
• Today’s Marketing Decision Environment: Integration

• Rich individual level data

• Rich (multi-objective, multi-variate) association (interaction) data

• Intelligent and speedy computer-aided tools to analyze data

=> Require marketers to take advantage of analyzing such data

using computer aided tools to be competitive

12
So What is Marketing Analytics?

• My Definition:

Marketing Analytics is a field of study dealing with

collecting and managing marketing ______ and

developing and applying

marketing models and _______________ tools

for the purpose of making better marketing __________.

13
III. Marketing Analytics Spring 2021 by Mike
• Position:
• An introductory course: No Prerequisites

• Decision Perspective: Getting insights and applying to marketing decisions

• Excel, Excel Add-ins (Data Analysis),

and Excel-based Web-based Software (ENGINIUS)

• Focus on the models and tools that can help make better decisions in

new product design (Conjoint Analysis),

market segmentation (Cluster & Discriminant Analysis),
positioning (Perceptual Mapping), and
generating consumer insights (Regression, Predictive Modeling)

14
• Side: Diverse Perspectives of Data Analytics

• Decision perspective: Developing Strategies***

• Prediction perspectives: Recommendations, Predictive Analysis
• Data preparation & analysis perspective: Managing Data
• Modeling perspective: Developing Models
• Software perspective: Algorithms, Software, Packages, Programming
• AI perspective: Machine learning, Text mining, Visualizing data
• Measures perspective: Developing and applying marketing measures
• Statistical Perspective: Applying typical and advanced statistical methods

15
 Appendix
Emergence of Marketing Analytics
Management Data Collection Modeling Analyze Big Data

Marketing Marketing Marketing Marketing

Management Research Science Analytics

Management Psychology Operations Research Computer S. (IT)

Economics Sociology Econometrics Statistics
1900 - 1950 - 1970 - 1990 -

16
Objectives
• Refer to the Syllabus.

• To motivate students to become Marketing Analytics friendly

• Exercise with excel based analytic tools

• Give opportunities to enjoy generating marketing insights from outputs

of running marketing models

[Lecture, Class Exercises, Case Analysis & Discussion, Term Projects]

• (Informally) To make you a little happier.

17
Class Schedule & Lecture Files
• Part I: Introduction, SW Preparations, Becoming Friendly (Classes 1,2,3)
• Part II: Regression Analysis (Classes 4,5,6,7,)
• Part III: Conjoint Analysis (Classes 8,9,10,11,12)
• Part IV: Segmentation with Cluster Analysis (Classes 13,14,15,16)

• Part V: Positioning with Perceptual Maps (Classes 19,21,22)

• Part VI: Predictive Modeling Analysis (Classes 23,24,25)

• Others: Review for exams (Classes 17, 29)

Mid-term (Class 18), Final Exam (Class 30),
Group Session for Term Project (Class 20),
Term Project Presentations & Feedback (Classes 26,27,28)

18
Term Project and Case Analysis
• Will form five teams of about 4 students (Class 3)*

• Term project: Be ambitious! Enjoy! Pick an “interesting topic”

• Five case analysis reports. Each team will present one in class.

• Test Marketing Case (Regression), Forte Hotel (Conjoint),

Ontela PicDeck (Segmentation), ISBM (Segmentation), Infiniti G20 (Positioning),

• Grading Criteria of Term Projects & Cases:

(1) Teamwork, Efforts, & Fun, (2) Solid Application of Information & tools,
(3) Marketing Implications, (4) Creativity and Pleasant Surprises.

19
Some appealing topics in this era? Big Picture?

• Customer needs that will increase drastically

• New business models that may fulfill growing needs/wants

• Business that has to change the way of doing business

• Business that can take advantage of factors that became much

affordable in this era

• Others …
Comparison of MA versus MR

Marketing Analytics Marketing Research

• Focus on Data Analytics • Focus on Data Collection
• Only Quantitative Research • Include Qualitative Research
• Focus on generating outputs & • Focus on basic Data Analysis
interpreting the outputs Fundamentals (Cross-Tabulation,
• Main tools: Regression, Conjoint Correlation, Hypothesis Testing)
Analysis, Cluster Analysis,
Perceptual Mapping
• Improve Decision Making • Improve Decision Making
 Prep 1: Office 365 Excel & Data Analysis ToolPak Add_Ins

File => Option => Add-Ins => Analysis ToolPak

Once installed: Excel => Data => Analyze (Data Analysis)

22
23
Prep 2: Getting Access to ENGINIUS programs

24
25
Prep 3: Buying Ontela PicDeck (B) Case

Purchasing
Ontela PicDeck (B) Case

27
Activating free Data Analysis Tool

 Click the Microsoft Office Button , and then click Excel Options.
 Click Add-Ins, and then in the Manage box, select Excel Add-ins.

 In the Add-Ins available box, select the Analysis ToolPak check box, and
then click OK.

Ontela PicDeck (B) Case:

 https://hbr.org/store/case-studies
 Search Ontela PicDeck (B): Customer Segmentation, Targeting and
Positioning

28
Today’s Question(s)
 What is Marketing Analytics?

 Why taking advantage of Marketing Analytics becomes a critical competitive

advantage today?

 Why many firms fail to achieve such advantage despite their efforts to utilize
Marketing Analytics?

29
Today’s Question(s)
 What is Marketing Analytics? => Definition

 Why taking advantage of Marketing Analytics becomes a critical competitive

advantage today?
=> Individual Level Data
Connection Data
Models and Computer Aided Tools

 Why many firms fail to achieve such advantage despite their efforts to utilize
Marketing Analytics?
=> Data challenge
=> Data Analyst challenge
30
Summary
 Today’s development of computer aided models and tools and availability of
individual level and interaction (or association) data require marketers to take
advantage of them to make better decisions than their competitors.

 The purpose of this course is to make the students become familiar with and
enjoy using models and computer aided tools in making marketing decisions.

 Lecture => Exercises with examples/cases => Case analysis & discussion

 Importance of the Term Projects

31
 Team Formation [email protected]

OFFICE HOUR: MW 1:00-2:00

OFFICE: CBA RM 336

 All the teams should submit all 5 case analysis reports. [Stage 1]
 Each team will be assigned randomly to present one of the case analyses. [Stage 2]

Members The case to make Presentation

Team 1: [ ] CableTV (Regression) [2/10]

Team 2: [ ] Forte Hotel (Conjoint Analysis) [3/1]
Team 3: [ ] Ontela PicDeck (Segmentation) [3/8]
Team 4: [ ] ConneCtor (Segmentation) [3/15]
Team 5: [ ] Infiniti G20 (Positioning) [4/12]
32
IV. Models, Data & Analytic Computer-Aided Tools

• What is Data? Can we directly generate insights from data?

• What is Model? What is the role of Models?

• What are the Computer-Aided Analytic Tools? Why do we

develop and use them?

• Preparations for Week 2

Marketing Analytics: A Competitive Advantage?

• Of course: Take advantage of rich, individual level, interaction data

Take advantage of competent computers getting better & better
(5.8 => 17.3% of Marketing Budget expected in 3 years*)

• Well...: Has not lived up to its promise?*

Data Challenge: Data is Data!
Data Analyst Challenge: Lack of competent MA people

* Source: Mela & Moorman, “Why Marketing Analytics Hasn’t Live Up to Its
Promise?” Harvard Business Review, May 2018

34
• Mela & Moorman, “Why Marketing Analytics Hasn’t Live Up to Its Promise?”
Harvard Business Review, May 2018
• In 3 years, analytics budgets over marketing budgets will increase from 5.8% to 17.3%

• Top marketers evaluation of effects of analytics on company wide performance has

increased from 3.8 to 4.1 out of a 7 point scales.
• Too much data but too little information
• Companies do not have the the right talent to leverage the marketing analytics

• Most important is knowing which tool, of the many available, is best for which problem
• Communicate insights not facts => structure the analysis in a decision framework
• Some good at math and coding, some good at framing issues, developing explanation
s and connecting to business implications => Need teamworks

35
What to do?

• Know which tool is the best for which problem ***

• Communicate insights not facts ***

=> structure the analysis in a decision framework

• Need teamworks ***

36
Newton’s Data and Gravity Theory

Gravity Model: F = G x (m1 x m2) / r2

37
• Harmonizing models, data, and computer-aided tools

=> Analyze data (Theory oriented vs. Data oriented)

=> Generate insights (With or without Computer-aided tools)

(Generate information competitors do not easily get)

=> Make better decisions

Where’s the Model?

38
IV-1. Data
• Facts and statistics collected together for reference or analysis

• Observations (personal, mechanical), Communications, Experiences,

Inferences

• Premise: Some existing information or knowledge will be represented

or coded in some form suitable for better usage or processing

• Is data useful? => Is data insightful?

39
SIDE: Data from Consumers
• Qualitative Inferences (Qualitative Data & Inferences)
• Focus Groups/ In-depth Personal Interviews/ Motivation Studies

• Direct (quantitative) measurements => Quantitative Data

• Attitude scales/ Purchase Intentions/ Choices/ Behavior measures (like clicks)

• Indirect Estimations applying Quantitative Models

• Conjoint Analysis => “partworths”
• Regression/ Logistic Regression/ Choice Model analysis => Coefficients, Elasticities
• Value-in-use assessments => Economic Value

40
Characteristics of objects are measured to be analyzed

Measurement is

an assignment of numbers 135 (or 60)

to characteristics weight
of objects, persons, states, or events Mike Minhi Hahn
according to certain rules. Pound (or Kg)

41
• Is the power relationship between producers and distributers affecting
the productivity of marketing channels?

• What type of contents are effective for twitters?

• What does the data say?

• Will computer find out distinctive patterns from the data?

=> To address these issues, we will need “MODELS"

42
• An era of Big Data?

• “Big data is the new reality. In fact, data is only getting better, faster, and
richer. It’s here to stay and you’d better capitalize on it.”
(Bari, Chaouchi, and Jung, Predictive Analytics for Dummies, 2nd ed., 2017)

• Is it the Size (Amount)?

• Is it the Variety?
▪ Is it the Measurement Level?
▪ Is it the Measured Objects?
▪ Is it the Measurement Tools?
=> Still, data is data.

43
Can data alone provide insights?

An example of “Natural Experiment” (An exploratory analysis)

$ Amount

Sales Level

Ad Level

1.5 year Time

Stopped Resumed
Ad Ad

44
An example of getting insights directly from Data (An exploratory analysis)

Sales of A New Drug

Sales

0 Time from the Launch

A comment in
a nine o’clock news
45
• Data usually includes errors.

• Data could be right or wrong.

• Data could have various types. (words, pictures, numbers, ...)

• Data could be subjective or objective.

• Data is data. Nothing more or nothing less than that.

• Can we transform data into relevant and insightful information?

46
Iris recognition of ring shaped region

Biometrica: “Mathematical Pattern Recognition”

ID Touch: Apple’s Fingerprint recognition

47
IV-2. From Data to Model
• Anheuser Busch’s Classic Advertising Experiments

Experiment 1 Experiment 2
Sales
compared
to
Normally
Expected
Ad Level

-25% Normal +50% -100% -50% Normal +50% … +300%

48
• WHAT IS YOUR MODEL or THEORY

on the data observed in the Anheuser Busch Experiments?

[Can we propose a model for the observations?]

49
 Role of Models

Description
DATA Decision Evaluation

Prediction Decision Rules

Model or Theory

50
What may be a model of “perceived quality of a course”?

Usefulness

Workload Perceived
Quality of a Course
Instructor

Content

Fun Models can enhance generating insights from data.

Provide the perspective for understanding data.
51
• My Definition of Model

• A model is a meaningful representation of a real world

phenomenon

- +
- +

52
• George Box

“Essentially, all models are wrong, but some are useful.”

(Box and Draper, 1987)

Is this a model of
KNOWLEDGE?

Somerset Maugham

53
• Models can be of a variety of forms

• Implicit (Mental), Explicit Models

• Graphic, Verbal, Mathematical Models

• Descriptive, Predictive, Decision (Normative) Models

54
• What determines your company sales?

Total Demand in the Market in 2020

Competitors’ Markeitng Budget in 2020

Sales in 2020
Our Marketing Budget in 2020

Our Sales in 2019, Total Demand in 2019

(Other omitted factors)

e.g., S2000 = D2000 x (Budget/Industry Budget) x (S2019/D2019)

55
• A simple Response Model: A Linear Model

Company Sales at t = a

+ b x Total Demand in the Market at period t

+ c x Company Marketing Budget at period t

+ d x Competitors Marketing Budget at period t

+ e x Company Sales at period t-1

+ Errors
56
 • A simple Response Model: A Linear Model Constant or Intercept

Company Sales at t = a
Dependent Variable
+ b x Total Demand in the Market at period t

+ c x Company Marketing Budget at period t

Coefficient

+ d x Competitors Marketing Budget at period t

Independent Variable
+ e x Company Sales at period t-1

+ Errors
57
 APPENDIX
• Marketing Response Models

MARKETING EFFECT = f (MARKETING EFFORT,

etc.)

Awareness, Image, Attitude, Loyalty Product

Purchase Intention, Purchase, Place
Purchase (choice) probability Price
Search, Trial, Recommendation, Promotion
Word of mouth, Social Interactions, Communication
Sales, Market Share, Service
Growth, Entertainment
… ...
58
IV-3. Computer-Aided Analytic Tools

• Taking advantage of “smart, intelligent, speedy” computers

• Every analytic tools has some “model” to apply.

• From logical inference models to simplistic heuristic models

e.g., Optimization models => Simulation Models

Conceptually Designed Models => Data Oriented Models
Supervised Learning Models => Unsupervised Learning Models

59
Summary
• Data is data.

• Models determine how we analyze data. They provide the perspective,

insights we bring to the data analysis.

• Data brings the objectivity into the analysis of real world phenomena.

• Analytic tools provide efficient ways to analyze data utilizing models.

 Computers are now faster and versatile than ever in analyzing a huge
amount of data. Marketers take advantage of such computers in making
decisions.

60
Summary
• What is Marketing Analytics? => Definition

• Why taking advantage of Marketing Analytics becomes a critical competitive advant

age today?
=> Individual Level Data
Interaction (Connection) Data
Models and Computer Aided Tools

• Why many firms fail to achieve such advantage despite their efforts to utilize Mark
eting Analytics?
=> Data challenge
=> Data Analyst challenge

61
• Today’s development of computer aided models and tools and availability
of individual level and interaction (or association) data require marketers to
take advantage of them to make better decisions than their competitors.

• The purpose of this course is to make the students become familiar with
and enjoy using models and computer aided tools in making marketing
decisions.

• Lecture => Exercises with examples/cases => Case analysis & discussion

• Importance of the Term Projects

62
V. Consumer Insights
• A deep understanding of customer
• Customer feelings, beliefs, and motivations

• Could give a smaller company an edge against bigger competitors

• Customers are not going to tell us

• Importance of looking beyond simply what the customer says or does

• Infer what motivates the consumer to say or do something

• The key is understanding deeper than what our competitors have

(From Kellogg Case, Gaming the Gamers)

63
WHY?

• Managers need consumer insights that can help predicting & c

ontrolling:

• Consumer choices among different product categories

• Consumer choices within a product category

• Consumer choices among different product categories and within

a category in the next related purchase occasion

64
• A Big Question:

• Marketers have Rich Data at individual consumer level

Smart and speedy computer aided tools

Old and newly developed models

• Quantitative Data Analytics: Is it the best way to generate consumer insights

?

• If yes, when?

65
Getting Consumer Insights

Qualitative Research Quantitative Research

“Consumer Insights” “Understanding Consumers”

from one Consumer? from “Big Data Analytics”

66
SIDE: Data from Consumers
• Qualitative Inferences (Qualitative Data & Inferences)
• Focus Groups/ In-depth Personal Interviews/ Motivation Studies

• Direct (quantitative) measurements => Quantitative Data

• Attitude scales/ Purchase Intentions/ Choices/ Behavior measures (like clicks)

• Indirect Estimations applying Quantitative Models

• Conjoint Analysis => “partworths”
• Regression/ Logistic Regression/ Choice Model analysis => Coefficients, Elasticities
• Value-in-use assessments => Economic Value

67
• Focus Group Study: Still one of the most popular marketing
research methods

68
What is your position?

• Consumers are homogeneous.

• Consumers are heterogeneous.

• Consumers are homogeneous and heterogeneous.

69
What is your position?

• Consumers are homogeneous. => A sample of one is enough?

• Consumers are heterogeneous. => A “big” sample is necessary!

• Segmentation => Targeting => Positioning

• Consumers are homogeneous and heterogeneous. => “Well...”

• “Comfortable underwear” vs. “Attractive underwear”

70
Consumer Insights with “Persona”
• Persona: the aspect of someone's character that is presented to or
perceived by others. (Bing)

• An example from Gaming the Gamers (HBS case)

• Solitary Gamers

• Genre Specific Gamers

• Multiplayer Gamers

71
So when do we rely on marketing analytics?
(Integration)

• Heterogeneity => Segmentation, Targeting and Positioning models, CLV

• Outliers => Statistics, Data mining

• Automatic and Routine Identification of Patterns => Data mining

• Endless calculations => “Evaluation of alternatives” models
(e.g., media scheduling models)

• Scientific decision making => Decision aid models

• Hypothesis testing => Statistics

72
 [email protected]
Team Formation OFFICE HOUR: MW 11:30-12:20 or by Appointment
OFFICE:

• All the teams should submit all 5 case analysis reports. [Stage 1]
• Each team will be assigned randomly to present one of the case analyses. [Stage 2]

Members Case to be presented

1
2
3
4
5
73
Today’s Questions:
1. How do we generate marketing insights from data?

• My Definition of Marketing Analytics (from the last class lecture):

Marketing Analytics is a field of study dealing with

collecting and managing marketing data and
developing and applying marketing models and
computer-aided tools
for the purpose of making better marketing decisions.

• Quantitative versus Qualitative Approach

74
Today’s Questions:
2. Why Marketing Analytics?

• Rich individual level data

• Rich (multi-objective, multi-variate) association data

• Intelligent and speedy computer-aided tools to analyze data

=> Availability of big data or “rich data” requires marketers to take a

dvantage of analyzing such data using computer aided tools to be
competitive.
• Marketers try to generate distinctive “deep” consumer insights to develop
superior competitive strategies. (qualitatively or quantitatively)

• Sometimes we rely on in-depth investigation of one or few persons and generate

insights, (e.g., developing “persona”)

• We may rely on analyzing observations of relatively a big number of persons to under

stand differences in different groups of customers and/or to choose potential target
consumers. Then, we may analyze their association data to develop competitive
marketing strategies.

• We will learn the most popular marketing analytic tools including conjoint analysis,
cluster analysis. perceptual mapping, regression analysis, choice (predictive)
modeling and cross tabulation analysis.

76
 APPENDIX
Customer Engagement Marketing
(Kotler & Armstrong, Principles of Marketing)

“...today’s companies are using online, mobile, and social media to refine their
targeting and to engage customers more deeply and interactively...

The new marketing is customer engagement marketing—fostering direct

and continuous customer involvement in shaping brand conversations,
brand experiences, and brand community...
Its goal is to make the brand a meaningful part of consumers’ conversations
and lives.”

(16th edition, Global edition, 2016, p.42)

77
 Appendix:

Modern Consumers => Era of customer engagement marketing

• They are empowered.

• They are repeat purchasers.

• They are older and experienced.

• They are dominantly female.
• They harmonize right & left brains.
• Gen-Z are new species.

• They are “whole human with mind, heart, and spirit”

(Marketing 3.0, Kotler et al., 2010)

78
Appendix: So What is Marketing Analytics?
• My Definition:

Marketing Analytics is a field of study dealing with

collecting and managing marketing data and

developing and applying

marketing models and computer-aided tools

for the purpose of making better marketing decisions.

79
 APPENDIX

• Data mining is the process of finding anomalies, patterns, and correlations

within large data sets to predict outcomes. (SAS Institute)

• Machine learning is a method of data analysis that automates analytical

model building. It is a branch of artificial intelligence based on the idea that
systems can learn from data, identify patterns, and make decisions with
minimal human intervention. (SAS Institute)

AI > Machine Learning > Deep Learning

• Predictive Analytics is the process of refining the data resources using

business knowledge to extract hidden value from those newly discovered
patterns by data mining. (Bar, Chaouchi & Jung 2017)
80
Appendix: Marketing Analytic Tools to Deal with in this semester

• Conjoint Analysis for A New Product Design [Enginius]

• Consumer tradeoffs among different attributes and different attribute levels

• What benefits do consumers consider the most important?
• Prediction of choices and market shares

• Cluster Analysis for Market Segmentation [Enginius]

• How do we identify groups of heterogeneous customers?

• What are the characteristics of different groups of customers?
• Which groups should we target at?
• Often needs Factor Analysis and Discriminant Analysis

81
• Perceptual Mapping for Positioning [Enginius]

• How consumers perceive the positions of alternative products?

• Which groups of consumers are most favorable with our product?
• Which product is the closest to the ideal product consumer perceives?

 Choice Model for Predictive Modeling [Enginius]

• How much does an attribute contribute to the preference or choice of a

consumer?
• What may be choice probability or market share for a product alternative?
• Estimating the latent classes of consumers (segmentation)
• How valuable is a consumer segment for a marketer?

82
• Regression Analysis [Excel Regression]

• What attributes contribute to the prediction of a marketing effect?

• Association between categorical or metric variables and a metric variable
• If we increase a marketing effort, how much impact will it have on
a marketing effect?
• How can we do the conjoint analysis with the regression analysis?

83
 APPENDIX
• Some other valuable computer aided tools

• Machine Learning (Clustering, Classification, Regression, )

• Deep Learning Models (Neural Network Models)
• Store location modeling
• Media selection and scheduling models
• Communication budget allocation models
• Sales force resource allocation models
• Forecasting models (e.g., New product forecasting, Box-Jenkins)
• Evaluation of promotional effects models
• Pricing models

84