Marketing Strategy Research - BM05
Introduction - Lecture 1 (30.10.2023)
After data is obtained, a strategy can be built. We therefore focus on the tools to use on marketing
research that will then bring us to a marketing strategy
Marketing strategy research emphasis
1. importance of data-driven marketing
2. complex and dynamic marketing analytics landscape
solutions:
○ linking analytical tools with marketing strategy - process of finding tools (5
techniques we will learn)
5 techniques learned
1. Linear regression => market responses model
2. Conjoint analysis => new product design/ product innovation (preferences elicitation)
3. Bass model => new product diffusion: when have new product, you wanna predict how's
gonna perform in market place
4. Cluster analysis (classification) => segmentation (targeting and positioning)
5. Multidimensional scaling => Positioning (how you are related from competitors,
differentiated in the market)
○ hands-on experience of marketing analytics
■ can be reached through: real marketing data; real managerial problems;
strategy recommendations
○ 4 Principles of data-driven marketing
■ The principles are generic and applicable to almost all data-driven marketing
situations. 4 principles:
1. any statistical analysis is to reduce information loss
2. causation cannot be learnt from data
3. prediction does not care about statistical significance
4. practical usefulness triumphs statistical criteria
Addressing the challenges
a. focus on the essentials
b. drill down approach - going deep into the tools - coding from scratch
i. e.g. cluster analysis - 5 steps cluster analysis
1. select distance measure
2. select a clustering procedure
3. decide the number of clusters
4. validate the clustering
5. interpret clusters
c. avoiding technical details - you are not expected to learn technical details
,Market Response Model - Lecture 2 (6.11.2023)
Predictive Modeling
Intended to help scholars and managers understand how consumers individually and collectively
respond to marketing activities, and how competitors intract > how to predict market response?
Data - Prediction Machine - Prediction machine
- the objective of the “prediction machine” is to find functional relationship between input
(data) and output (prediction)
seek to establish “functional relationships” > different forms
- Linear Regression: y = a + bx (a = intercept; b = slope)
Toy example (linear regression)
using price to predict sales
X: price (IV) = Input
Y: Sales (DV) = Output
Sales = a + bPrice
, a. a = intercept = starting point
b. b = slope/coefficient = direction of the line
What is a good prediction?
Principle 1: Any statistical analysis is to reduce information loss
How to draw the line in scatterplot?
Choose a line to minimize the differences
1. square Δy (differences)
2. sum up over all point
5 steps to perform linear regression
1. examining the data
When IVs are highly correlated (e.g. age and
income) it’s fundamental to check
multicollinearity - cause it’s harder to
estimate coefficient
Introduction - Lecture 1 (30.10.2023)
After data is obtained, a strategy can be built. We therefore focus on the tools to use on marketing
research that will then bring us to a marketing strategy
Marketing strategy research emphasis
1. importance of data-driven marketing
2. complex and dynamic marketing analytics landscape
solutions:
○ linking analytical tools with marketing strategy - process of finding tools (5
techniques we will learn)
5 techniques learned
1. Linear regression => market responses model
2. Conjoint analysis => new product design/ product innovation (preferences elicitation)
3. Bass model => new product diffusion: when have new product, you wanna predict how's
gonna perform in market place
4. Cluster analysis (classification) => segmentation (targeting and positioning)
5. Multidimensional scaling => Positioning (how you are related from competitors,
differentiated in the market)
○ hands-on experience of marketing analytics
■ can be reached through: real marketing data; real managerial problems;
strategy recommendations
○ 4 Principles of data-driven marketing
■ The principles are generic and applicable to almost all data-driven marketing
situations. 4 principles:
1. any statistical analysis is to reduce information loss
2. causation cannot be learnt from data
3. prediction does not care about statistical significance
4. practical usefulness triumphs statistical criteria
Addressing the challenges
a. focus on the essentials
b. drill down approach - going deep into the tools - coding from scratch
i. e.g. cluster analysis - 5 steps cluster analysis
1. select distance measure
2. select a clustering procedure
3. decide the number of clusters
4. validate the clustering
5. interpret clusters
c. avoiding technical details - you are not expected to learn technical details
,Market Response Model - Lecture 2 (6.11.2023)
Predictive Modeling
Intended to help scholars and managers understand how consumers individually and collectively
respond to marketing activities, and how competitors intract > how to predict market response?
Data - Prediction Machine - Prediction machine
- the objective of the “prediction machine” is to find functional relationship between input
(data) and output (prediction)
seek to establish “functional relationships” > different forms
- Linear Regression: y = a + bx (a = intercept; b = slope)
Toy example (linear regression)
using price to predict sales
X: price (IV) = Input
Y: Sales (DV) = Output
Sales = a + bPrice
, a. a = intercept = starting point
b. b = slope/coefficient = direction of the line
What is a good prediction?
Principle 1: Any statistical analysis is to reduce information loss
How to draw the line in scatterplot?
Choose a line to minimize the differences
1. square Δy (differences)
2. sum up over all point
5 steps to perform linear regression
1. examining the data
When IVs are highly correlated (e.g. age and
income) it’s fundamental to check
multicollinearity - cause it’s harder to
estimate coefficient
