lc06ad22c43660d948dc44ab011c94baa-s5741869699282365348-mf97ca59fdae299916f20f3ceb114fb1d.pdf

PRAISE FOR MARKETING ANALYTICS

‘With its focus on practicality, this book is an invaluable toolkit of
frameworks to drive consumer-centric analytics initiatives across marketing
organizations. It is unique in going beyond theoretical aspects and helping
practitioners apply analytics to understand consumer behaviour and
identify business opportunities. Grigsby’s extensive experience makes it a
must-read for marketing professionals of all levels.’ Anna Andrusova, Senior
Data Analyst, JCPenney

‘This is an excellent read for people in the industry who work in strategy and
marketing. It is one of the first books that I have read that covers the entire
spectrum from demand, segmentation, targeting, and how results can be
calculated. In an age where marketing is becoming more and more
sophisticated, this book provides the tools and the mathematics behind the
facts. Marketing Analytics is written with a scientific voice, but is very
readable, with the science wrapped into everyday activities, based on a
character we can all relate to, that are derived from these formulas,
ultimately driving ROI.’ Elizabeth Johnson, CEO, PathFormance

‘Grigsby’s book is the right blend of theory applied to the real-world large-
scale data problems of marketing. It’s exactly the book I wish I’d had when I
started out in this field.’ Jeff Weiner, Senior Director, Analytics, One10

‘An insightful, practical book for analytics marketing practitioners. It both
entertains and serves as a handbook for marketing analytics. With easy-to-
follow examples, Grigsby paints a clear picture of how to execute data
analytics and its role in the larger marketing and organizational goals.’ Craig
Armstrong, Director, Strategic Business Analysis, Targetbase

‘This is a great book for practitioners who have learned plenty of theories
and want to learn how to apply methodologies. It is also a great, easy-to-
read resource for anyone who does not have a deep theoretical background

but wants to learn how analytics work in real life.’ Ingrid Guo, VP, Analytics,
and Managing Director, Javelin Marketing Group (Beijing)

‘In Marketing Analytics, Mike Grigsby takes passionate marketing strategists
on a practical, real-life journey for solving common marketing challenges. By
combining the concepts and knowledge areas of statistics, marketing
strategy and consumer behaviour, Grigsby recommends scientific and
innovative solutions to common marketing problems in the current business
environment. Every chapter is an interesting journey for the reader.

What I like most about the book is its simplicity and how it applies to real
work-related situations in which almost all of us have been involved while
practising marketing of any sort. I also like how the author talks about
tangible measurements of strategic recommended marketing solutions as
well as how they add value to companies’ strategic endeavours. I highly
recommend reading this book as it adds a completely new dimension to
marketing science.’ Kristina Domazetoska, Project Manager and
Implementation Consultant at Insala – Talent Development and
Mentoring Solutions

‘Marketing Analytics, second edition is a must-read for students and budding
analytics professionals. The book illustrates concepts in statistics and
marketing with real-world examples and provides solutions without getting
too technical. It begins with basic statistical concepts required in the field of
marketing analytics, then illustrates the application of these concepts to
real-world business problems. It also touches upon concepts of big data
analytics and, most importantly, what really IS an insight. This book is
extremely conversational and entertaining to read and I’ve found myself
reaching for it on multiple occasions when I’ve encountered various
marketing-analytics-related problems, during both my student and
professional life.’ Akshay Kher, Analytics Practitioner

Second Edition

Marketing Analytics
A practical guide to improving consumer
insights using data techniques

Mike Grigsby

Publisher’s note
Every possible effort has been made to ensure that the information
contained in this book is accurate at the time of going to press, and the
publisher and author cannot accept responsibility for any errors or
omissions, however caused. No responsibility for loss or damage
occasioned to any person acting, or refraining from action, as a result
of the material in this publication can be accepted by the editor, the
publisher or the author.

First published in Great Britain and the United States in 2015 by Kogan
Page Limited as Marketing Analytics: A practical guide to real marketing
science

Second edition published in 2018

Apart from any fair dealing for the purposes of research or private study,
or criticism or review, as permitted under the Copyright, Designs and
Patents Act 1988, this publication may only be reproduced, stored or
transmitted, in any form or by any means, with the prior permission in
writing of the publishers, or in the case of reprographic reproduction in
accordance with the terms and licences issued by the CLA. Enquiries
concerning reproduction outside these terms should be sent to the
publishers at the undermentioned addresses:

2nd Floor, 45 Gee Street
London EC1V 3RS
United Kingdom

c/o Martin P Hill Consulting
122 W 27th St, 10th Floor
New York NY 10001
USA

4737/23 Ansari Road
Daryaganj
New Delhi 110002

India

www.koganpage.com

© Mike Grigsby, 2015, 2018

The right of Mike Grigsby to be identified as the author of this work has
been asserted by him in accordance with the Copyright, Designs and
Patents Act 1988.

ISBN 978 0 7494 8216 9
E-ISBN 978 0 7494 8217 6

Typeset by Integra Software Services, Pondicherry
Print production managed by Jellyfish
Printed and bound by CPI Group (UK) Ltd, Croydon, CR0 4YY

http://www.koganpage.com/

CONTENTS

Cover
Title Page
Copyright
Contents
List of Figures
List of Tables
Foreword to the first edition
Foreword to the second edition
Preface

Introduction to marketing analytics

PART ONE    Overview – how can marketing analytics help you?

01    A brief statistics review

Measures of central tendency
Measures of dispersion
The normal distribution
Confidence intervals
Relations among two variables: covariance and

correlation
Probability and the sampling distribution
Conclusion
Checklist: You’ll be the smartest person in the room if

you …

02    Brief principles of consumer behaviour and marketing strategy

Introduction
Consumer behaviour as the basis for marketing strategy
Overview of consumer behaviour
Overview of marketing strategy
Conclusion

Checklist: You’ll be the smartest person in the room if
you …

03    What is an insight?

Introduction
Insights tend not to be used by executives
Is this an insight?
So, what is an insight?
Ultimately, an insight is about action-ability
Checklist: You’ll be the smartest person in the room if

you …

PART TWO    Dependent variable techniques

04        What drives demand? Modelling dependent variable
techniques

Introduction
Dependent equation type vs inter-relationship type

statistics
Deterministic vs probabilistic equations
Business case

Results applied to business case

Modelling elasticity
Technical notes
Highlight: Segmentation and elasticity modelling can

maximize revenue in a retail/medical clinic chain: field
test results

Abstract
The problem and some background
Description of the dataset
First: segmentation
Then: elasticity modelling
Last: test vs control
Discussion
Conclusion

Checklist: You’ll be the smartest person in the room if
you …

05    Who is most likely to buy and how do I target them?

Introduction
Conceptual notes
Business case

Results applied to the model

Lift charts
Using the model – collinearity overview
Variable diagnostics
Highlight: Using logistic regression for market basket

analysis

Abstract
What is a market basket?
Logistic regression
How to estimate/predict the market basket
Conclusion

Checklist: You’ll be the smartest person in the room if
you …

06    When are my customers most likely to buy?

Introduction
Conceptual overview of survival analysis
Business case

More about survival analysis
Model output and interpretation
Conclusion

Highlight: Lifetime value: how predictive analysis is
superior to descriptive analysis

Abstract
Descriptive analysis
Predictive analysis
An example

Checklist: You’ll be the smartest person in the room if
you …

07    Panel regression – how to use a cross-sectional time series

Introduction
What is panel regression?
Panel regression: details
Business case

Insights about marcom (direct mail, e-mail and SMS)
Insights about time period (quarters)
Insights about cross sections (counties)
Conclusion

Checklist: You’ll be the smartest person in the room if
you …

08        Systems of equations for modelling dependent variable
techniques

Introduction
What are simultaneous equations?
Why go to the trouble of using simultaneous equations?
Desirable properties of estimators
Business case

Conclusion

Checklist: You’ll be the smartest person in the room if
you …

PART THREE    Inter-relationship techniques

09        What does my (customer) market look like? Modelling inter-
relationship techniques

Introduction
Introduction to segmentation
What is segmentation? What is a segment?
Why segment? Strategic uses of segmentation
The four Ps of strategic marketing

Criteria for actionable segmentation
A priori or not?
Conceptual process
Checklist: You’ll be the smartest person in the room if

you …

10    Segmentation – tools and techniques

Overview
Metrics of successful segmentation
General analytic techniques
Business case

Analytics
Comments/details on individual segments
K-means compared to LCA

Highlight: Why go beyond RFM?

Abstract
What is RFM?
What is behavioural segmentation?
What does behavioural segmentation provide that RFM does not?
Conclusion

Segmentation techniques
Checklist: You’ll be the smartest person in the room if

you …

PART FOUR    More important topics for everyday marketing

11    Statistical testing – how do I know what works?

Everyone wants to test
Sample size equation: use the lift measure
A/B testing and full factorial differences
Business case
Checklist: You’ll be the smartest person in the room if

you …

12    Implementing Big Data and Big Data analytics

Introduction
What is Big Data?
Is Big Data important?
What does it mean for analytics? For strategy?
So what?
Surviving the Big Data panic
Big Data analytics
Big Data – exotic algorithms
Conclusion
Checklist: You’ll be the smartest person in the room if

you …

PART FIVE    Conclusion

13    The finale – what should you take away from this?

What things have I learned that I’d like to pass on to
you?

What other things should you take away from all this?

Glossary
Bibliography and further reading
Index
Backcover

List of Figures

FIGURE 1.1    Home sales prices

FIGURE 1.2    Standard deviation

FIGURE 4.1    Actual and predicted unit sales

FIGURE 5.1    Actual events and logistics

FIGURE 5.2    Lift chart

FIGURE 6.1    General survival curve

FIGURE 6.2    Survival analysis

FIGURE 9.1    Levels of consumer behaviour

FIGURE 10.1    CHAID output

FIGURE 10.2    Hierarchical clustering – dendogram

FIGURE 10.3    Significant variables

FIGURE 10.4    % of market vs % of revenue

FIGURE 11.1    Z-scores

FIGURE 12.1    Example of a neural network

FIGURE 12.2    Example of a support vector machine

List of Tables

TABLE 1.1    Variance

TABLE 1.2    Covariance and correlation

TABLE 4.1    Demand model data

TABLE 4.2    Ordinary regression

TABLE 4.3    Quarterly model

TABLE 4.4    Regression output

TABLE 4.5    Elasticity, inelasticity, and total revenue

TABLE 4.6    Average price and ad spend

TABLE 4.7    Serial correlation

TABLE 4.8    Serial correlation

TABLE 4.9    Elasticity modelling

TABLE 4.10    Elasticity modelling

TABLE 4.11    Further elasticity modelling

TABLE 4.12    Further elasticity modelling

TABLE 4.13    Further elasticity modelling

TABLE 5.1    Simplified dataset

TABLE 5.2    Co-efficient output

TABLE 5.3    Confusion matrix

TABLE 5.4    New variables

TABLE 5.5    Updated confusion matrix

TABLE 5.6    Variance

TABLE 5.7    Probability to purchase

TABLE 5.8    Associated probabilities

TABLE 6.1    Final desktop model, lifereg

TABLE 6.2    Time to event (in weeks)

TABLE 6.3    Three model comparison

TABLE 6.4    Comparison of customers from different behavioural
segments

TABLE 6.5    Results of survival modelling

TABLE 6.6    LTV calculations

TABLE 7.1    Panel data structure

TABLE 7.2    Coefficients on marcom: random effects

TABLE 7.3    Coefficients on marcom: fixed effects

TABLE 7.4    Quarterly seasonality

TABLE 7.5    Cross-sectional analysis

TABLE 8.1    Model results

TABLE 10.1    Segmenting algorithms compared

TABLE 10.2    Bayes Information Criterion

TABLE 10.3    Bayes Information Criterion: second model

TABLE 10.4    List of variables removed

TABLE 10.5    General view of six segments

TABLE 10.6    Details by segment

TABLE 10.7    Additional details by segment

TABLE 10.8    KPIs

TABLE 10.9    Customer totals

TABLE 11.1    Testing discounts against different audiences

TABLE 11.2    Testing discounts against different audiences in a 16
cell matrix

TABLE 11.3    Multiple sources model

Test banks, datasets and PowerPoint lecture slides relating to chapters are available online at:
www.koganpage.com/MarketingAnalytics2

http://www.koganpage.com/MarketingAnalytics2

FOREWORD TO THE FIRST EDITION

In Marketing Analytics Mike Grigsby provides a new way of thinking
about solving marketing and business problems, with a practical set
of solutions. This relevant guide is intended for practitioners across a
variety of fields, but is rigorous enough to satisfy the appetite of
scholars as well.

I can certainly appreciate Mike’s motivations for the book. This
book is his way of giving back to the analytics community by offering
advice and step-by-step guidance for ways to solve some of the most
common situations, opportunities, and problems in marketing. He
knows what works for entry, mid-level, and very experienced career
analytics professionals, because this is the kind of guide he would
have liked at these stages.

While Mike’s education includes a PhD in Marketing Science, he
also pulls from his vast experiences from his start as an Analyst,
through his journey to VP of Analytics, to walk the reader through
the types of questions and business challenges we face in the
analytics field on a regular basis. His authority on the subject matter
is obvious, and his enthusiasm is contagious, and best captured by
my favourite sentence of his book: ‘Now let’s look at some data and
run a model, because that’s where all the fun is.’

What this education and experience means for the rest of us is that
we have a well-informed author providing us with insight into the
realities of what is needed from the exciting work we do, and how
we can not only provide better decision making, but also move the
needle on important theoretical and methodological approaches in
Analytics.

More specifically, Marketing Analytics covers both inter-relational
and dependency-driven analytics and modelling to solve marketing
problems. In a light and conversational style (both engaging and
surprising) Mike argues that, ultimately, all markets rely on a strong
understanding of the ever-changing, difficult to predict, sometimes

fuzzy, and elusive minds and hearts of consumers. Anything we can
do to better arm ourselves as marketers to develop this
understanding is certainly time well spent. Consumers can and
should be the focal point of great strategy, operational standards of
excellence and processes, tactical decisions, product design, and so
much more, which is why it makes perfect sense to better
understand not just consumer behaviours, but also consumer
thoughts, opinions, and feelings, particularly related to your vertical,
competitors, and brand.

After a review of seminal work on consumer behaviour, and an
overview of general statistics and statistical techniques, Marketing
Analytics dives into realistic business scenarios with the clever use of
corporate dialogue between Scott, our fictitious analyst, and his boss.
As our protagonist progresses through his career, we see an
improvement in his toolkit of analytical techniques. He moves from
an entry level analyst in a cubicle to a senior leader of analytics with
staff. The problems become more challenging, and the process for
choosing the analytics to apply to the situations presented is an
uncanny reflection of reality – at least based on my experiences.

What I appreciate absolutely most about this work though is the
full spectrum of problem solving, not just analytics in a vacuum.
Mike walks us from the initial moment when a problem is identified,
through communication of that problem, framing by the Analytics
team, technique selection and execution (from the straightforward
to somewhat advanced), communication of results, and usefulness to
the company. This rare and certainly more complete picture
warrants a title such as Problem Solving using Marketing Analytics
in lieu of the shorter title Mike chose.

Marketing Analytics will have you rethinking your methods,
developing more innovative ways to progress your marketing
analytics techniques, and adjusting your communication practices.
Finally, a book we all can use!

Dr Beverly Wright, VP, Analytics, BKV Consulting

FOREWORD TO THE SECOND EDITION

Mike Grigsby has done the seemingly impossible: created a guide to
marketing analytics that is technically sound and clearly applicable
to real-world business cases, while also being a thoroughly enjoyable
(and even entertaining!) read.

The first edition of Marketing Analytics was so well received by
educators and practitioners because it delivered simple,
straightforward analytic prescriptions to those seeking an actionable
primer. In a market that remains largely saturated with jargon-laced
statistical tomes geared more towards academicians than to
marketers, the second edition of Marketing Analytics continues to
distinguish itself as the user-friendly text that earns the coveted
‘quick reach’ spot on analysts’ office bookshelves. Mike serves up
new chapters on panel regression and big data analytics, with these
(and all other covered techniques) framed and contextualized by a
thoughtful new overview chapter that addresses a basic – but often
elusive – question: ‘What is an Insight?’

The book is logically structured, with a clear progression from
foundational principles to workhorse techniques in predictive
modelling (dependent variable applications) and segmentation
(inter-relationship solutions), concluding with treatments of some of
the most important topics in the field, including the pivotal role of
consumer behaviour, the logic of testing and inference and the rise
of big data solutions. In short, this edition covers the lion’s share of
methods used by successful marketing analysts, and it should be
required reading for students and marketers alike.

A number of admirable traits define the discussion of the subject
matter. First, the description of core topics is informative and
practical. Each method contains a clear description of when and why
it is used, with key diagnostics and test statistics presented in
practical, lay terms. In addition, the treatment of each topic is
illustrative, following the career trajectory of our protagonist, Scott,

as he confronts and addresses increasingly complex business
problems over the course of his career. (The dialogue-laden
narratives are refreshing, making the material more accessible to –
and more fun for – the book’s diverse audience.) Further illustrations
to each technique are provided by ‘highlight’ sections in which Mike
presents case vignettes of how he actually applied these techniques
to answer business questions over the course of his career. Finally,
the book is to be commended for adopting a consumer-centric point
of view on applied analytics. Mike correctly challenges his readers to
don the mental mantle of the customer, tailoring analytic choices to
focus on how and why customers make decisions and how we as
marketers can impact those decisions.

I am fortunate to have Dr Grigsby join our Consumer Insights
practice at Brierley & Partners. I can attest that Mike practises what
he preaches in his book, and he is a patient mentor to his analysts
and a trusted advisor to many of our largest clients.

As the field of marketing analytics continues to explode, aspiring
practitioners and veteran analysts would do well to ensure that their
solutions are grounded in the customer-centric approaches
delineated in this work. As new touch points and information
sources continue to crowd and compete for attention in the modern
marketing ecosystem, analytics would be wise to heed Mike’s astute
observation: ‘Generally speaking, new data sources do not require
new analytic techniques.’ What is required is an action-oriented
approach to using analytics to meaningfully impact customer choice
– a framework cleanly served up in this second edition of Marketing
Analytics.

Don Smith, PhD, Chief Analytics Officer, Brierley & Partners

Preface

We’ll start by trying to get a few things straight. I did not set out to
write a (typical) textbook. I’ll mention some textbooks down the line
that might be helpful in some areas, but this is too slim for an
academic tome. Leaf through it and you’ll not find any mathematical
proofs, nor are there pages upon pages of equations. This is meant to
be a gentle overview – more conceptual than statistical – for the
marketing analyst who just needs to know how to get on with their
job. That is, it’s for those who are, or hope to be, practitioners. This is
written with practitioners in mind.

Introduction to marketing analytics

Who is the intended audience for this book?
This is not meant to be an academic tome filled with mathematical
minutia and cluttered with statistical mumbo-jumbo. There will
need to be an equation now and then, but if your interest is
econometric rigour, you’re in the wrong place. A couple of good
books for that are Econometric Analysis by William H Greene (1993)
and Econometric Models, Techniques and Applications by Michael
Intriligator, Ronald G Bodkin and Cheng Hsiao (1996). So, this book is
not aimed at the statistician, although there will be a fair amount of
verbiage about statistics.

This is not meant to be a replacement for a programming manual,
even though there will be SAS code sprinkled in now and then. If
you’re all about BI (business intelligence), which means mostly
reporting and visualizing data, this is not for you.

This will not be a marketing strategy guide, but be aware that as
mathematics is the handmaiden of science, marketing analytics is
the handmaiden of marketing strategy. There is no point to analytics
unless it has a strategic payoff. It’s not what is interesting to the
analyst, but what is impactful to the business that is the focus of
marketing science.

So, at whom is this book aimed? Not necessarily at the professional
econometrician/statistician, but there ought to be some satisfaction
here for them. Primarily, the aim is at the practitioner (or those who
will be). The intended audience is the business analyst that has to
pull a targeted list, the campaign manager that needs to know which
promotion worked best, the marketer that must DE-market some
segment of her customers to gain efficiency, the marketing
researcher that needs to design and implement a satisfaction survey,
the pricing analyst that has to set optimal prices between products
and brands, etc.

What is marketing science?
As alluded to above, marketing science is the analytic arm of
marketing. Marketing science (interchangeable with marketing
analytics) seeks to quantify causality. Marketing science is not an
oxymoron (like military intelligence, happily married or jumbo
shrimp) but is a necessary (although not sufficient) part of marketing
strategy. It is more than simply designing campaign test cells. Its
overall purpose is to decrease the chance of marketers making a
wrong decision. It cannot replace managerial judgement, but it can
offer boundaries and guard rails to inform strategic decisions. It
encompasses areas from marketing research all the way to database
marketing.

Why is marketing science important?
Marketing science quantifies the causality of consumer behaviour. If
you don’t know already, consumer behaviour is the centre-point, the
hub, the pivot around which all marketing hinges. Any ‘marketing’
that is not about consumer behaviour (understanding it,
incentivizing it, changing it, etc) is probably heading down the
wrong road.

Marketing science gives input/information to the organization. This
information is necessary for the very survival of the firm. Much like
an organism requires information from its environment in to
change, adapt and evolve, an organization needs to know how its
operating environment changes. To not collect and act and evolve
based on this information would be death. To survive, for both the
organization and the organism, insights (from data) are required.
Yes, this is reasoning by analogy but you see what I mean.

Marketing science teases out strategy. Unless you know what
causes what, you will not know which lever to pull. Marketing
science tells you, for instance, that this segment is sensitive to price,
this cohort prefers this marcom (marketing communication) vehicle,
this group is under competitive pressure, this population is not loyal,

and so on. Knowing which lever to pull (by different consumer
groups) allows optimization of your portfolio.

What kind of people in what jobs use
marketing science?
Most people in marketing science (also called decision science,
analytics, customer relationship management – or CRM,
direct/database marketing, insights, research, etc) have a
quantitative bent. Their education is typically some combination
involving statistics, econometrics/economics, mathematics,
programming/computer science, business/marketing/marketing
research, strategy, intelligence, operations, etc. Their experience
certainly touches any and all parts of the above. The ideal analytic
person has a strong quantitative orientation as well as a feel for
consumer behaviour and the strategies that affect it. As in all
marketing, consumer behaviour is the focal point of marketing
science.

Marketing science is usually practised in firms that have a CRM or
direct/database marketing component, or firms that do marketing
research and need to undertake analytics on the survey responses.
Forecasting is a part of marketing science, as well as design of
experiments (DOE), web analytics and even choice behaviour
(conjoint). In short, any quantitative analysis applied to
economic/marketing data will have a marketing science application.
So while the subjects of analysis are fairly broad, the number of
(typical) analytic techniques tends to be fairly narrow. See Consumer
Insight by Stone, Bond and Foss (2004) to get a view of this in action.

Why do I think I have something to say
about marketing science?
Fair question. My whole career has been involved in marketing
analytics. For more than 25 years I’ve done direct marketing, CRM,

database marketing, marketing research, decision sciences,
forecasting, segmentation, design of experiments and all the rest.
While my BBA and MBA are in finance, my PhD is in marketing
science. I’ve published a few trade and academic articles, …