MARKETING

ANALYTICS

Introduction to Marketing Analytics

Topics to be covered

 Introduction to Marketing Analytics

 Defining the Right Business Problems
 Data Sources
 Data Types
 Predictors versus Target Variable
 Modeling Types
 The 7-Step Marketing Analytics Process
Introduction to Marketing Analytics

 A primary responsibility of marketing is to properly manage

the wants and needs of customers. This can be accomplished
through strategic decisions about products, pricing,
distribution, and communications that are based on insights
from marketing analytics. In this section, the discussion will
center on:
 Marketing Analytics Defined
 Analytics Levels
(continued…)

 Marketing analytics uses data, statistics, mathematics, and

technology to solve marketing business problems. It involves
modeling and software to drive marketing decision making.
 Today, however, the availability of large amounts of data,
improvements in analytics techniques, substantial increases
in computer processing power, and affordability have made
marketing analytics more practical and available to a much
larger audience.
(continued…)

Google Search Trends for

the Terms Marketing
Analytics, Supply Chain
Analytics, Financial
Analytics, and HR Analytics
(2020 estimated)
(continued…)

 Analytic Levels: There are basically four analytic levels:

Defining the Right Business Problems
 One of the most critical steps in the analytics process is to start with an
understanding of the real business question the data needs to address.
Knowing the right business question leads to smarter decisions that drive
impact.
 In an effort to define the right business problem, it can be useful to follow
the SMART analytics principles. The SMART principles can be used as a
goal-setting technique. The acronym stands for specific, measurable,
attainable, relevant, and timely.
Data Sources

 Data sources consist of both primary and secondary data. Primary data
is collected for a specific purpose. Companies conduct surveys, focus
groups, interviews, observations, and experiments to address problems
or answer distinct questions.
 In contrast, secondary data relies on existing data that has been
collected for another purpose. While secondary data might not address
specific or current problems, it could be useful in formulating ideas
about how to ask the right questions or to design future data collection
initiatives. At the same time, internal and external secondary data
sources can be useful in exploring current business questions.
Data Types

 Types of Data: Data is facts and figures collected, organized,

and presented for analysis and interpretation. Data is
available in two main forms:
 Structured and
 Unstructured.
(continued…)

 Structured Data: Structured data is made up of records that

are organized in rows and columns. This type of data can be
stored in a database or spreadsheet format. It includes
numbers, dates, and text strings that are stored in a clearly
defined structure. The data is easy to access and analyze
using descriptive, predictive, prescriptive, and AI data
analytics techniques.
(continued…)

 Unstructured Data: Unstructured data includes text,

images, videos, and sensor data. The data does not have a
predefined structure and does not fit well into a table format
(within rows and columns). Examples of this type of data
include voice recording from customer service calls, text,
images, video recording, social media conversations, and the
Internet of Things sensor data.
 Both structured and unstructured data are important in
executing marketing analytics. As noted, structured and
unstructured data come in different formats and
measurement types.
(continued…)

 Four Levels of Measurement: The following figure illustrates

four primary scales:
(continued…)

 Nominal Scale: A scale whose numbers serve only as labels or tags for
identifying and classifying objects. When used for identification, there
is a strict one-to-one correspondence between the numbers and the
objects.
 Consider the following examples of nominal scale.
(continued…)

 Ordinal Scale: A ranking scale in which numbers are assigned to

objects to indicate the relative extent to which some characteristic is
possessed. Thus it is possible to determine whether an object has more
or less of a characteristic than some other object.
 Example: Indicate the level your economic status:

Low (1) Medium (2) High (3)

(continued…)

 Interval Scale: A scale in which the numbers are used to rate objects
such that numerically equal distances on the scale represent equal
distance in the characteristic being measured.
 Example: The Fahrenheit temperature scale is an example of the
interval level of measurement.
 Named after its developer, Rensis Likert, the Likert scale is widely used
interval scale in the field of marketing.
(continued…)

 Ratio Scale: The highest scale. It allows the researcher to identify or

classify objects, rank-order the objects, and compare intervals or
differences. It is also meaningful to compute ratios of scale values.
 Examples of the ratio scale of measurement include wages, units of
production, weight, changes in stock prices, distance between branch
offices, and height. Money is also a good illustration.
(continued…)
 Data Measurement: Data measurement can be categorized in a variety
of ways based on the type and means for collection.
(continued…)

 The two main types of data measurement most often

explored in the remaining chapters are numerical and
categorical.
Predictors versus Target Variable
 Types of Variables: Variables are characteristics or features that
pertain to a person, place, or object. Marketing analysts explore
relationships between variables to improve decision making. Although
the example in the following figure only uses two variables, companies
often use multiple variables at the same time as inputs to systems that
process data and use it to predict dependent variables.

In practical marketing analytics applications, variables are designated as

independent (predictor) variables or dependent (target) variables.
Modeling Types: Supervised Learning versus Unsupervised Learning

 Depending on the nature of the business problem being addressed,

different types of algorithms can be used resulting in different types of
modeling: Supervised Learning and Unsupervised learning.

 Supervised learning suggests that the target variable of interest is

known and is available in historical dataset. The historical dataset can
also be referred to as labeled data (because the target variable is
available) and is divided into a training dataset, validation dataset, and
an optional testing dataset (holdout sample).
(continued…)

 When the target variable is continuous (or metric), supervised learning

is referred to prediction. When the target variable is categorical (or
non-metric), supervised learning is called classification.
(continued…)

 Unsupervised learning has no previously defined target variable. The

goal of the unsupervised learning is to model the underlying structure
and distribution in the data to discover and confirm patterns in the data.

 Supervised and unsupervised learning can be used together to gain

more insight,
The 7-Step Marketing Analytics Process
(continued…)

 Step-1: Business Problem Understanding: Most marketing analytics

models are developed when a business identifies a problem. The idea is
to develop a model using analytics problem better and design a
solution. One of the key elements in this step is to question whether the
problem the business is presenting is the correct problem.
(continued…)

 Step-2: Data Understanding and Collection: Here the marketing

analysts’ job is to identify where the data is stored, its format, and how it
can be combined to understand the question at hand. Marketing analyst
must have a good understanding of the types of data and sources of
data.
(continued…)

 Step-3: Data Preparation and Feature Selection: Data in different

formats is combined in this step. To do so, the unit of analysis and the
target and predictor variables identified. Three most important tasks are
data cleaning, data examining, and data modification.
(continued…)

 Step-4: Modeling Development: In this step, the analyst selects the

methods to use. The choice depends on the variable type and
availability and the business question addressed. Different models
should be tried to identify the one that provides the best accuracy,
speed, and quality.
(continued…)

 Step-5: Model Evaluation and Interpretation: This step ensures the

modeling is accurately performed and provides the best prediction for
the future. If the validation results show high accuracy, then the model
can be recommended evaluating an optional testing dataset to assess
how well the final selected model might perform on new data.
(continued…)

 Step-6: Model and Results Communication: Understanding the

differences in the perspectives of problem solving skills is critical
because most people may not have a clear understanding of the
modeling techniques used. It is key, therefore, for the analyst to present
the model in a way that other people can understand, particularly
management.
(continued…)

 Step-7: Model Deployment: The model completion and execution step

is not finished until it has been implemented and is running on real-time
records to offer decisions or actions.
Thank You!