Marketing Analytics in Modern Business

• Essentials of Marketing Analytics was published in 2022.

Introduction to Marketing Analytics

• This chapter discusses an analytics framework
• The chapter also discusses relevant marketing analytics concepts
• Industry best practices are also covered

Marketing Analytics Defined

• Marketing analytics uses data, statistics, mathematics, and tech to solve marketing issues.
• Modeling and software help drive marketing decisions
• Marketing Analytics is the fastest growing field of analytics applications
• The impact and benefits of marketing analytics are increasingly clear

Analytics Level and Their Impact on Competitive Advantage

• Higher data management is needed as organizations use more advanced techniques
• Increased analysis maturity is also required for a competitive advantage

Analytic Levels

• Descriptive analytics explain or quantify the past, using data queries, reports, and statistics
• Predictive analytics is used to build models to explain the future using the past
• Historic sales can predict future sales
• Prescriptive analytics identifies the optimal course of action or decision
• UPS route optimization and Amazon's price optimization are examples
• Artificial Intelligence (AI) and cognitive analytics mimic human-like intelligence for specific tasks
• AI analytics uses machine learning to understand new data
• Hitachi uses AI to discover patterns typically undetected by humans

Defining the Right Business Problems

• Understanding what the customer does requires deep investigation of the customer's path
• You need to know how, where, and why they search along their journey
• Knowing their satisfaction levels it also key
• Problem identification uncovers strategic options
• Problem identification will help improve market share and customer relations
• You'll be able to position the company to make the most of innovation
• To arrive at answers, you must also understand the intent behind the question from the customer
• You must also include stakeholder input using discovery methods
• Discovery can begin with questions relating to the what, who, where, when, why, and how

SMART Principles

• The SMART principles can be a goal-setting technique.
• For SMART, The goal should be clearly Specific
• Progress should be Measurable
• The goal should be Achievable
• It should also be Relevant by solving the analytics problem and aligning with business objectives
• The goal should be Timely, with a frame determined for completion
• Its important to analyze the success of the project and if it will make an impact

Data Sources

• Data consists of both primary and secondary information
• Primary data is collected for a specific reason
• Secondary data relies on existing data collected for another purpose
• Examples of secondary data include public datasets, online & mobile sites, channel partners, commercial brokers, corporate information and even government sources

Types of Data

• Structured data is made of records in rows and columns
• Structured data can be stored in a database or in a spreadsheet formula
• Structured data includes numbers, dates, and text strings
• Structured data is easy to access and analyze.
• Images, videos, text, and sensor data is unstructured
• Unstructured data does not fit into a table
• Advanced analytics are required to analyze unstructured data
• Technology has advanced to help improve manipulation and exploration of this data

Data Measurement

• Numerical data is either discrete (integer) or continuous
• Discrete data is measured in whole numbers
• Continuous data can include values with decimals
• Categorical data are values selected from categories.
• Binary data can have two values
• Nominal data has no meaningful order
• Ordinal data has inherent meaningful values

Metric Measurement Scales

• Metric scales can be measured as intervals or ratios.
• Both scales possess constant units of measure
• The distance between each point of the scale are equal.
• Interval variables do not include an absolute zero.
• Ratio scales have an absolute zero point and can be discussed in terms of multiples

Predictors versus Target Variable

• Variables are features that pertain to a person, place, or object.
• Weather can impact ice cream sales, where conditions are independent
• The weather influences or drives the dependent, target, or outcome variable which would be the ice cream sales.

Modeling Types: Supervised versus Unsupervised Learning

• Supervised learning is used when a target variable is known
• A training dataset helps "learn” the variable
• A validation dataset assesses the algorithm's accuracy
• A testing dataset evaluates the final selected algorithm.
• Apply the algorythm to new unlabeled data
• If the target variable is continuous, results are a prediction
• If categorical, supervised learning is called a classification.
• Use unsupervised learning when there's no variable
• The goal is to model data to discover and confirm patterns
• Association analysis offers suggestions based on past purchases.
• Cluster analysis groups customers based on key variables.

The 7-Step Marketing Analytics Process

• The 7-step marketing analytics process is iterative that helps improves the modeling cycle
• Each step is important to achieve a successful outcome.

Step 1: Business Problem Understanding

• Marketing analytics models are used when a business identifies a problem.
• Analytics help to understand and design the solution.
• Questioning it to ensure that its the right problem is the correct
• You need to understand/solve, how stakeholders will use results and who will affected and if this is an ongoing issue

Step 2: Data Understanding and Collection

• Identify where the data is, its format, and how to combine to understand the question
• Interview stakeholders and examine databases to confirm there is an issue
• Next, sample databases from records to analyze any issues
• As an example: examine purchases and customer returns
• Marketing analysts must understand different sources of data
• The origin of data will affect the decision

Step 3: Data Preparation and Feature Selection

• Combine data in different formats
• Identify the unit of analysis which is the variable
• Visually and statistically examine the data
• Deal with missing data
• Merge data from different sources
• Other features are refined
• Adjust the data formats
• Improve accuracy to help improve the model

Step 4: Modeling Development

• Select the method to use
• Choice depends on the variable
• You can use classification, prediction, clustering, or association
• Partition the data into datasets
• Training, validation, and testing
• The analyst should decide on appropriate modeling techniques
• Search for the model that provides accuracy, speed, and quality and is also simple and practical

Step 5: Model Evaluation and Interpretation

• Find the value of the algorithm
• The algorithm runs on the validation dataset
• High accuracy in validation leads the model to predict new cases

Step 6: Model and Results Communication

• The analyst can presents the models in a way other people can understand
• The analyst can approaches stakeholders early-on and ensure that they also fully understand the model
• Whether simple or complex, the model should be explainable in straightforward terms with appropriate visualizations.

Step 7: Model Deployment

• Implement the model on real records to help actions.
• You'll need to train the key stakeholders to action the system
• Ethics must be assessed
• This includes the privacy of subjects
• Also, will there be bias?
• And is the objective unfair?
• The analysts must follow ethics

Explore the impact of marketing analytics on business, its rapid growth, and its role in gaining a competitive edge. Learn the benefits of implementing effective analytics and how companies like Expedia and Spotify use it to achieve common goals.