Data Analytics for Marketing: A practical guide to analyzing marketing data using Python

Data Analytics for Marketing

Copyright © 2024 Packt Publishing

All rights reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, without the prior written permission of the publisher, except in the case of brief quotations embedded in critical articles or reviews.

Every effort has been made in the preparation of this book to ensure the accuracy of the information presented. However, the information contained in this book is sold without warranty, either express or implied. Neither the author, nor Packt Publishing or its dealers and distributors, will be held liable for any damages caused or alleged to have been caused directly or indirectly by this book.

Packt Publishing has endeavored to provide trademark information about all of the companies and products mentioned in this book by the appropriate use of capitals. However, Packt Publishing cannot guarantee the accuracy of this information.

Group Product Manager: Kaustubh Manglurkar

Publishing Product Manager: Heramb Bhavsar

Book Project Managers: Farheen Fathima and Shambhavi Mishra

Senior Editor: Rohit Singh

Technical Editor: Rahul Limbachiya

Copy Editor: Safis Editing

Proofreaders: Rohit Singh and Safis Editing

Indexer: Subalakshmi Govindhan

Production Designer: Joshua Misquitta

DevRel Marketing Executive: Nivedita Singh

First published: May 2024

Production reference: 1120424

Published by Packt Publishing Ltd.

Grosvenor House

11 St Paul’s Square

Birmingham

B3 1RB, UK.

ISBN 978-1-80324-160-9

To Andreia Lopes, my partner. You are my safe harbor; this book would not have become a reality without your support during the endless hours I thought about giving up. You make me want to be better, and I would not be where I am without you. Love you!

– Guilherme Diaz-Bérrio

Contributors

About the author

Guilherme Diaz-Bérrio is the Head of Marketing Analytics at Kindred Group, one of the 10 largest gambling operators. He helps improve marketing efforts across various platforms. His career started in finance at a hedge fund and moved through the automotive industry at BMW Group and BMW Financial Services, before coming to Kindred Group. He graduated with a degree in economics from ISEG, University of Lisbon, and has additional training in data science and econometrics. He is also the co-founder of Pinemarsh, a data analytics and digital marketing consulting firm.

I want to thank my editor, Rohit Singh, for his help and patience in reviewing my drafts. I also want to thank Birjees Patel and Deepesh Patel, for taking a chance and inviting me to write this book. I would like to thank the reviewers, Devanshu Tayal, Shubham Gupta, and Michael Van Den Reym. Their feedback was incredibly helpful in getting to the final drafts. Finally, I would like to thank Krishna Bhaskaran, for giving me the opportunity to work in marketing analytics and giving me the foundations of what I know today.

About the reviewers

Devanshu Tayal is a highly accomplished data scientist with a master’s degree from BITS, Pilani, India. His extensive expertise in data science is evidenced by his contributions to a wide range of industries. Devanshu is deeply committed to mentoring and guiding aspiring data scientists and is an avid researcher of emerging technologies in the field. He is a strong advocate for diversity and inclusion and has shared his insights through various publications. Devanshu is frequently invited to deliver guest lectures at universities throughout India, and his contributions as a technical reviewer have been acknowledged in multiple books. His comprehensive knowledge and experience in the field make him an asset to any team or project.

Shubham Gupta, an accomplished technology leader and a staunch advocate for data-driven decision-making, possesses a vast wealth of expertise spanning analytics, business intelligence, strategic planning, and cutting-edge innovative solutions. His deep comprehension of both technological intricacies and business dynamics equips him to guide business stakeholders from a wide array of industries toward making well-informed, data-backed decisions, thus streamlining operations and fostering substantial growth.

Furthermore, Shubham’s significant involvement as a judge for numerous prestigious tech awards highlights his unwavering dedication to promoting excellence and driving innovation within the tech industry.

Michael Van Den Reym is a seasoned professional in the field of digital analytics and search engine optimization, with a rich background in enhancing online visibility for diverse businesses. Currently, he’s working at iO, the biggest full-service digital agency in Belgium. Michael has been a speaker on data-driven marketing at industry conferences such as MeasureCamp, MeasureFest, and BrightonSEO. Michael’s work primarily revolves around leveraging data-driven insights using Python and data visualization tools to create strategies that significantly improve digital marketing outcomes.

Currently, Michael is working on his first book, Fundamentals of SEO for Business, which revolves around search engine optimization for marketing professionals.

Table of Contents

Preface

Part 1: Fundamentals of Analytics

1

What is Marketing Analytics?

What is analytics?

An overview of marketing analytics

Why should we bother with marketing analytics?

Exploring different types of analytics

Descriptive analytics

Diagnostic analytics

Predictive analytics

Prescriptive analytics

Walking through the maze of tools and techniques

Beyond simple pivot tables

Why Python?

Modern challenges in the world of privacy-centric marketing

The importance of data engineering and tracking

Don’t moonlight as a data engineer

Web tracking is hard, and it is becoming harder

Summary

Further reading

2

Extracting and Exploring Data with Singer and pandas

Technical requirements

What is ETL, and why should you care?

Data pipelines

What is Singer?

Summarizing data and EDA

Primer on descriptive statistics

Percentiles, quantiles, and distributions

Measures of central tendency

Measures of variability

Dealing with common data issues

Bill Gates walks into a bar

Missing values and data imputation

Digging deeper into variable transformations

Data standardization or scaling

Power transformations

Summary

Further reading

3

Design Principles and Presenting Results with Streamlit

Technical requirements

Types of dashboards and their design

Understanding the design concepts of a dashboard

Thinking about how to best present data

Thinking a bit about processing information

Generating effective filters, dimensions, and metrics

Filters

Dimensions

Metrics

Getting your data into Streamlit and generating a basic dashboard

Starting out with Streamlit

Creating a marketing data dashboard with Streamlit

Summary

Further reading

4

Econometrics and Causal Inference with Statsmodels and PyMC

Technical requirements

What is a linear regression?

What is a model?

What are the assumptions of a linear regression?

Exploring different types of regression models

What we can do when the assumptions break down

How to do a linear regression

What is logistic regression?

Objectives of logistic regression models

Odds of an event

What is causal inference?

Correlation, causation, and key drivers

A more practical application

A small detour through the backdoor

Watch out for colliders

Summary

Further reading

Part 2: Planning Ahead

5

Forecasting with Prophet, ARIMA, and Other Models Using StatsForecast

Technical requirements

What is forecasting?

Why forecasting is important

Types of times series data

Exploratory data analysis

What to forecast

Weekly, daily, and sub-daily data

Time series of counts

Prediction intervals for aggregates

Long and short time series

Transformations

What types of patterns are present?

Time series decomposition

Time series features

Basics of time series forecasting

Simple methods

Fitted values and residuals

Correlation and forecasting

Variable selection in time series regression models

Advanced forecasting methods

Extending regression models to time series

ETS models

ARIMA models

The Prophet model

Which model to use

Summary

Further reading

6

Anomaly Detection with StatsForecast and PyMC

Technical requirements

What is an anomaly?

Techniques to detect anomalies

Anomaly detection with STL decomposition

Twitter’s t-ESD algorithm for anomaly detection

Isolation forests for anomaly detection

Forecasting as an anomaly detection tool

Practical implementation with StatsForecast

Using rates of arrival to identify change points

Pros and cons of using rates of arrival for change point detection

Summary

Further reading

Part 3: Who and What to Target

7

Customer Insights – Segmentation and RFM

Technical requirements

Understanding the sources of customer dynamics

Analyzing customer dynamics – unveiling segmentation and RFM

Delving deeper into what segmentation is

Clustering

Classification

Discriminant analysis and classification

Exploring RFM

Approaches and techniques – independent versus sequential sorting

A practical example of RFM analysis

Profitability evaluation

ROMI after RFM

Results of using RFM for targeting

Summary

Further reading

8

Customer Lifetime Value with PyMC Marketing

Technical requirements

Diving deeper into CLV

CLV in practice

Using CLV to calculate acquisition costs

CLV and prospects

CLV and incremental value

What’s wrong with the CLV formula?

Issue 1

Issue 2

Issue 3

Issue 4

Issue 5

Beyond the CLV formula

The BTYD model

The Pareto/NBD model

The BG/NBD model

Implementing the BTYD model using PyMC Marketing

Predicting the expected number of purchases for a new customer

Estimating the CLV

Summary

Further reading

9

Customer Survey Analysis

Technical requirements

Steps in customer survey analysis

Questionnaire construction

Principles of questionnaire design

Types of questions

Asking questions

Questionnaire design—layout

Response formats

Reliability and validity

Reliability and classical measurement theory

Standard error of measurement

Using scales with high reliability

How to do sampling

Types of sampling

Probability versus quota sampling

Sample size for estimating population mean

Response rate

Control charts

Customer loyalty and NPS methodology

Issues with NPS

Potential loss of revenue

Advocacy, purchasing, and retention loyalty

Factor analysis

Summary

Further reading

10

Conjoint Analysis with pandas and Statsmodels

Technical requirements

An introduction to conjoint analysis

The fundamentals of conjoint analysis

Setting up a conjoint study

Step 1 – select the product attributes to be included

Step 2 – select the product attribute levels

Step 3 – create product profiles

Step 4 – collect data from target customers

Step 5 – estimate the utility of each product attribute and levels using regression analysis

Conducting conjoint analysis in Python

Determining the value of a product attribute

Choice-based conjoint analysis

Reporting findings

Summary

Further reading

Part 4: Measuring Effectiveness

11

Multi-Touch Digital Attribution

Technical requirements

An introduction to attribution models

Heuristic attribution models

The implementation of different heuristic attribution models

Algorithmic attribution models

Shapley value attribution

Fractribution

Summary

Further reading

12

Media Mix Modeling with PyMC Marketing

Technical requirements

Understanding MMM

MMM versus MTA versus lift analysis and A/B testing

Steps toward implementing MMM

Data collection

How much data to collect

Modeling

How to measure the adstock effect

Saturation and diminishing returns

Which comes first?

Selecting a model

Experimenting and calibrating

A synthetic data example of MMM

Synthetic data generation

Modeling

Model results

Summary

Further reading

13

Running Experiments with PyMC

Technical requirements

What makes a good experiment?

A/A testing

Type I and Type II errors

p-values

Common pitfalls

Delving deeper into some pitfalls

Conversion rate

Uplift modeling

Experimentation

Observational studies

Quasi-experiments

Difference in differences

Synthetic control and causal impact

Summary

Further reading

Index

Other Books You May Enjoy

Preface

When I first started as a marketing data analyst, I felt lost. I already knew how to program in Python, and the basics of statistics and econometrics, but marketing analytics is a surprisingly deceptive field. It feels easy, but due to the nature of the data we work with, it involves more complex models than you initially thought; correlation is often confused with causation, and sometimes, you just feel like you are flying blind. Either that or you feel like a glorified pivot table maker. A lot of my knowledge then came from trial and error: testing techniques, reading up on new methods, and making mistakes… a lot of mistakes.

When I started managing and hiring a team of analysts, I sometimes felt that it would make my life easier if I could just point them to a book that gave them the basics instead of spending hours in one-on-one sessions explaining methods or techniques. This book is my attempt at that: a summary of the fundamentals of marketing analytics, above simple pivot table making.

Marketing analytics is an incredibly complex field, and it is impossible to encapsulate all of it in one book. This book aims to give you, the reader, a grounding understanding of the techniques and tools most used in marketing analytics. The aim is to provide a practical, no-nonsense approach. You will have to have some basic understanding of the theoretical aspects surrounding tools and models so that you know what you are using and why, but we will quickly shift to a practical approach. The ultimate goal of this book is to equip you with the practical knowledge to get operational as a marketing data analyst quickly. This book will present open source libraries that allow you to derive insights quickly and use examples of common questions you will face daily as a marketing analyst.

There are gaps and techniques we will not explore. Marketing analytics is an ever-evolving field, and writing a book to cover everything would take more than 5,000 pages. Each chapter could be, and sometimes is, its own book. This book aims to equip you with fundamental knowledge, give you an overview of what is available, and provide some understanding of how to apply it. It also aims to give you the biggest asset an experienced analyst can have – knowing what to look for and research when facing a new problem. This last point is, for me, the most important. If this book achieves nothing else, let it be that it provides you with the compass to find the tools and techniques you need in your daily life. Even if that means you will disagree with me on a specific technique, that will be a win.

Throughout the book, we will use Python and its rich data analysis and statistics package ecosystem.

Who this book is for

There are some assumptions I have about who you are as a reader. Although an attempt is made to explain the most complex Python code snippets, you need to have a basic understanding of Python and be comfortable with it. By comfortable, I mean you know what a function is, how to define it, how to import a module, and the basic language syntax. Another requirement is you should not be afraid of mathematics. This point is contentious, but some chapters will have some formulation and theory before we get into actual code. Some people might disagree, but while copilots such as GitHub Copilot or ChatGPT can help you produce the code, you still need fundamental theoretical knowledge. In fact, with code copilots becoming better and better, most likely, the distinction between a good and an average analyst will be the theoretical grounding they have. I will attempt to give you the basic toolbelt of math techniques early on, starting from how to calculate a mean, but this book assumes you are comfortable with high-school-level mathematical notation.

This book is primarily aimed at data analysts who want to understand the full suite of techniques available to them in marketing analytics. You can also be a marketing professional aiming to move to the analytical side, but if this is you, I advise you to first brush up on the basics of Python programming, math, and statistics.

What this book covers

Chapter 1

, What is Marketing Analytics?, delves into what we mean by marketing analytics, breaking down the types of analytics, from descriptive to prescriptive, what value they add to the business, and what questions each of them answers.

Chapter 2

, Extracting and Exploring Data with Singer and pandas, gives you a brief introduction to ETL and how to extract and handle marketing data, ingestion, and Exploratory Data Analysis (EDA). We will cover the fundamentals of descriptive statistics and go through common data transformations to ensure data normality.

Chapter 3

, Design Principles and Presenting Results with Streamlit, takes us through how to properly design a dashboard for marketing data, from design principles to actual implementation. This is instrumental in displaying our results in a presentable way to non-technical audiences.

Chapter 4

, Econometrics and Causal Inference with Statsmodels and PyMC, deals with the fact that, as a marketing analyst, you usually do not have the luxury of big data to feed into machine learning models. The data will be sparse or low-frequency time series or panel data, which will prevent you from brute-forcing your way through. You need a solid understanding of econometrics and the principles of causality to answer common questions your stakeholders will have.

Chapter 5

, Forecasting with Prophet, ARIMA, and Other Models Using StatsForecast, digs deeper into forecasting. Forecasting is one of the fundamental tasks of a marketing data analyst. It is also one of the most complex fields in statistics. You should understand which models to apply, when to apply them, and what to avoid. We will review the most common models, from ARIMA to ETS, and what are the common pitfalls in forecasting time series.

Chapter 6

, Anomaly Detection with StatsForecast and PyMC, describes how to perform anomaly detection. In the daily life of an analyst, you will be tasked, more often than not, with finding anomalies before they create business impact. You will also have to understand how to deal with low-frequency data and derive anomalies while avoiding false positives.

Chapter 7

, Customer Insights – Segmentation and RFM, helps us discover how to segment customers and create valuable profiles for better marketing. We’ll explore customer segmentation and RFM scoring.

Chapter 8

, Customer Lifetime Value with PyMC Marketing, builds upon the previous chapter by showing how to assign a value to our customers and segments to optimize our marketing efforts, and to evaluate the ROI of our activities by estimating how much customers are worth.

Chapter 9

, Customer Survey Analysis, describes customer satisfaction analysis through surveys. Analyzing customer satisfaction is an integral part of customer satisfaction management, which is an important part of CRM. We’ll go through how to analyze survey data to derive insights, how to calculate samples, and the pitfalls of NPS.

Chapter 10

, Conjoint Analysis with pandas and Statsmodels, starts with a description of what conjoint analysis is and what it is used for. We’ll cover some of the techniques used to derive useful insights, customize your product offering with conjoint analysis, and explain how to build the analysis from the ground up.

Chapter 11

, Multi-Touch Digital Attribution, explains in detail what digital attribution is. Marketing attribution is a fundamental problem in marketing analytics. How to attribute outcomes to marketing channels will change the conclusions you derive from channel evaluation. This chapter will describe the most common attribution methods and how to build them.

Chapter 12

, Media Mix Modeling with PyMC Marketing, describes the fundamental issue of understanding how to use Media Mix Modeling to optimize your marketing activities. Understanding a marketing channel’s performance is important, but of critical importance in modern marketing analytics is understanding how channels interact with each other. The answer to this question allows us, as analysts, to advise marketing teams on optimal budget allocation.

Chapter 13

, Running Experiments with PyMC, starts by explaining the fundamentals of what an experiment is. Running experiments in marketing is a fundamental technique for optimization and efficiency. We’ll go through the fundamentals of how to run experiments and how to analyze the outcome, while avoiding the most common pitfalls.

To get the most out of this book

The code provided for this book comes in the form of Jupyter Notebooks, with the exception of Chapter 3

, which is a Python file. All chapters will teach you how to install the required packages for each section.

You should already have a Python distribution installed on your development machine, either via the main Python Website or using Anaconda. An alternative is to use Google Colab.

All source code was developed and tested on MacOS (64-bit) and Google Colab.

If you are using the digital version of this book, we advise you to type the code yourself or access the code from the book’s GitHub repository (a link is available in the next section). Doing so will help you avoid any potential errors related to the copying and pasting of code.

Download the example code files

You can download the example code files for this book from GitHub at https://github.com/PacktPublishing/Data-Analytics-for-Marketing

. If there’s an update to the code, it will be updated in the GitHub repository.

We also have other code bundles from our rich catalog of books and videos available at https://github.com/PacktPublishing/

. Check them out!

Conventions used

There are a number of text conventions used throughout this book.

Code in text: Indicates code words in text, database table names, folder names, filenames, file extensions, pathnames, dummy URLs, user input, and Twitter handles. Here is an example: To begin, we need to import the required libraries: streamlit, pandas, numpy, matplotlib, and seaborn.

A block of code is set as follows:

import pandas as pd

import numpy as np

import seaborn as sns

pd.set_option('display.float_format', lambda x: '%.2f' % x)

df = pd.read_csv('data/ trafficsources.csv')

df.describe()

When we wish to draw your attention to a particular part of a code block, the relevant lines or items are set in bold:

# Define the model

with pm.Model() as model_pymc:    alpha = 1.0/daily_sales.sales.mean()      lambda_1 = pm.Exponential(lambda_1, alpha)    lambda_2 = pm.Exponential(lambda_2, alpha)

Any command-line input or output is written as follows:

pip install tap-googleads

Bold: Indicates a new term, an important word, or words that you see onscreen. For instance, words in menus or dialog boxes appear in bold. Here is an example: Analytics can be split into four areas or pillars: descriptive, predictive, diagnostic, and prescriptive analytics.

Tips or important notes

Appear like this.

Get in touch

Feedback from our readers is always welcome.

General feedback: If you have questions about any aspect of this book, email us at [email protected]

and mention the book title in the subject of your message.

Errata: Although we have taken every care to ensure the accuracy of our content, mistakes do happen. If you have found a mistake in this book, we would be grateful if you would report this to us. Please visit www.packtpub.com/support/errata

and fill in the form.

Piracy: If you come across any illegal copies of our works in any form on the internet, we would be grateful if you would provide us with the location address or website name. Please contact us at [email protected]

with a link to the material.

If you are interested in becoming an author: If there is a topic that you have expertise in and you are interested in either writing or contributing to a book, please visit authors.packtpub.com

.

Share your thoughts

Once you’ve read Data Analytics for Marketing, we’d love to hear your thoughts! Please click here to go straight to the Amazon review page

for this book and share your feedback.

Your review is important to us and the tech community and will help us make sure we’re delivering excellent quality content.

Download a free PDF copy of this book

Thanks for purchasing this book!

Do you like to read on the go but are unable to carry your print books everywhere?

Is your eBook purchase not compatible with the device of your choice?

Don’t worry, now with every Packt book you get a DRM-free PDF version of that book at no cost.

Read anywhere, any place, on any device. Search, copy, and paste code from your favorite technical books directly into your application.

The perks don’t stop there, you can get exclusive access to discounts, newsletters, and great free content in your inbox daily

Follow these simple steps to get the benefits:

Scan the QR code or visit the link below

https://packt.link/free-ebook/9781803241609

2. Submit your proof of purchase

3. That’s it! We’ll send your free PDF and other benefits to your email directly

Part 1: Fundamentals of Analytics

In this part, we will go through the fundamentals of analytics, introducing marketing analytics as a discipline. We will be focusing on data extraction, ingestion, and exploratory data analysis, followed by techniques for effectively presenting results and building dashboards for non-technical audiences. The subsequent discussion shifts toward econometrics and causal inference, providing a foundational understanding of statistics and equipping you with the skills to construct, test, and evaluate statistical models, emphasizing their significance and application in marketing.

This part contains the following chapters:

Chapter 1

, What is Marketing Analytics?

Chapter 2

, Extracting and Exploring Data with Singer and pandas

Chapter 3

, Design Principles and Presenting Results with Streamlit

Chapter 4

, Econometrics and Causal Inference with Statsmodels and PyMC

1

What is Marketing Analytics?

Half the money I spend on advertising is wasted; the trouble is I don’t know which half.

– John Wanamaker, the forefather of marketing

In this chapter, we will attempt to cover the fundamentals of marketing analytics as a role and discipline. As a marketing analyst, you are faced with common questions during your day-to-day activities. For example, How did this campaign perform? or How can you optimize your budget to achieve a result?.

In this chapter, we will break down the types of analytics (from descriptive to prescriptive), the value they add to a business, and the questions each of them answers.

You will learn about the following topics:

What is analytics?

An overview of marketing analytics

Exploring different types of analytics

Beyond simple pivot tables

Why Python?

Modern challenges in the world of privacy-centric marketing

The importance of data engineering and tracking

By the end of this chapter, you will understand what marketing analytics is and what it is supposed to measure. You will have a firm grasp of the different types of analytics and why simply using a spreadsheet, while tempting, is sometimes not enough. You will also have an understanding of the importance of data engineering and web tracking.

But before we delve into the tools and techniques that are required of you, to achieve your results, we first need to unpack what we mean by analytics in general and marketing analytics in particular.

What is analytics?

Like any buzzword, analytics can often be overused and hard to define from an exact source. According to the Oxford Dictionary, the textbook definition of analytics is the systematic computational analysis of data or statistics, in order to describe, predict, and improve business performance. Gartner defines it more broadly as statistical and mathematical data analysis that clusters, segments, scores, and predicts what scenarios are most likely to happen.

Analytics is commonly known to branch out into four pillars or areas: descriptive analytics, diagnostic analytics, predictive analytics, and prescriptive analytics.

In essence, analytics is the act of extracting meaningful and actionable insights from data by using a set of techniques and tools paired with domain knowledge. Raw data, however large it may be, will not be a silver bullet in your quest for insights in marketing. Neither will advanced techniques and a lack of domain knowledge of how the field you are analyzing operates. It is only in the joining of these three aspects—domain knowledge, data, and techniques—that you will be able to do your job.

An important caveat about analytics is that it is neither reporting nor data science. Your primary job is not to produce a stable, automatically updated dashboard or a machine learning pipeline. In analytics, it helps you to have the right mindset and try to achieve reproducible code for follow-up analysis, or have a stable pipeline. But as an analyst, that is not your primary goal; it is just nice to have in order to achieve your end result. Your primary goal is speed and accuracy; that is, you need to produce meaningful insights that teams easily rely on in a reasonable amount of time with reasonable accuracy.

While it may seem controversial, this distinction bears some thought. Often, analytics will be folded into one of the two extremes. Either it is viewed as simple reporting and/or BI work, meaning you will lose the ability to generate actionable insights due to the rigid nature of datasets and dashboard architecture required. Or, it is viewed as data science, which means you will often use complex models that require a lot of data to learn and are lacking in interpretability. Analytics stands in the middle, although with blurry frontiers:

Figure 1.1 – Business analytics as the intersection of several skills

Now let's gain an overview of what is marketing analytics.

An overview of marketing analytics

The quote at the beginning of this chapter illustrates one of the fundamental questions of the marketing manager in their day-to-day activities. The best way to evaluate where to spend and target their efforts to achieve their ultimate target is to obtain new customers or retain current ones.

Marketing analytics is nothing more than the application of analytical methods to said goal, bringing a quantifiable way of guiding investment or consumer targeting decisions. As with any new and growing domain, it is hard to pin an exact definition of it, but we can define it as a technology-enabled and model-supported approach to harness customer and market data to enhance marketing decision making. Being a domain in the larger field of data analytics, it looks to use mathematics and statistics together with computational tools and techniques to find meaningful patterns and knowledge in data. In this book, we will strive to focus on only the most relevant techniques and models to solve fundamental questions in marketing analytics.

Standard techniques frequently used in marketing include media mix modeling, pricing and promotion analyses, sales force optimization, and customer analytics such as segmentation or lifetime value estimation. The optimization of websites and online campaigns now frequently works hand in hand with the more traditional marketing analysis techniques, coupled with attribution modeling and media mix modeling, to understand channel interactions and optimal budget allocation.

These tools and techniques will allow you to support both strategic marketing decisions—such as how much to spend on marketing and how to allocate budgets across a portfolio of brands and the marketing mix—and more tactical campaign support in terms of targeting the best potential customer with the optimal message in the most cost-effective medium at the ideal time.

The past decades have seen an explosion of data in a digital format, with some estimates pointing to a jump from 6 percent to around 90 percent. That, together with massive improvements in computational tools such as faster databases, inference algorithms, and easier programming languages for statistics means the dramatic improvement and evolution of marketing analytics in recent times. But one might wonder why we should be concerned with marketing analytics, or why it should be regarded as an independent sub-field of greater analytics.

Why should we bother with marketing analytics?

Any business that employs analytics, of any kind, expects that it will improve the performance of said business. Marketing analytics is no different. Evidence supports the claim that marketing analytics improves business performance, be it in the form of increased sales, profits, or market share.

One study states that for a one-unit increase in marketing analytics deployment (measured on a scale of 1–7), there is an increase of 8 percent in return on assets (ROA) for the business, accounting for $180 million in net income. Businesses in highly competitive industries gain even more; an increase of 21 percent on ROA.

Let’s see a simple example of what marketing analytics can do with a mail coupon campaign. Kroger, an American retailer, conducts regular direct mail coupon campaigns. These campaigns to customers delivered a redemption rate of 70 percent within six weeks of mailing compared to an average of 7.93 percent for other companies. How? According to the analytics company working with Kroger, Demographics can tell you nothing about it. Just because I am the same age as you, live next door, and have 2.2 children does not mean we have the same preferences. What they do is study each customer to see what drives their behaviors individually. Do they have kids, do they skew toward healthy or fun, do they prefer organic or convenience foods, and where are they price sensitive? Is this across all products or only some? We tell our retailing customers there is no silver bullet. Take data from customers and look at the decisions the business is making and look at their impact on the consumer.

Having discussed the what and why of marketing analytics, we need to take a small detour to explain the different types of analytics in the analytical maturity model to better understand what to apply in each step.

Exploring different types of analytics

As we have seen earlier, analytics is a broad term covering four different pillars in the modern analytics ladder. Each plays a role in how your business can better understand what your data reveals and how you can use those insights to drive business objectives.

The following diagram will help you visualize how the pillars relate to one another:

Figure 1.2 – The analytics maturity model

The first step in the process is to always understand the fundamental questions you are trying to answer. All analytical questions can be boiled down into the following categories:

What happened and when did it happen?

Why and how did it happen?

What will happen in the future?

How can I make something happen?

These categories will define the different areas of analytics involved, which will inform our decision about what tooling and techniques to apply.

Analytics can be split into four areas or pillars: descriptive, predictive, diagnostic, and prescriptive analytics.

You can think of the pillars by remembering