PDF Fighting Churn with Data 1st Edition Carl Gold download

Download as pdf or txt
Download as pdf or txt
You are on page 1of 62

Download the full version of the textbook now at textbookfull.

com

Fighting Churn with Data 1st Edition Carl


Gold

https://textbookfull.com/product/fighting-churn-
with-data-1st-edition-carl-gold/

Explore and download more textbook at https://textbookfull.com


Recommended digital products (PDF, EPUB, MOBI) that
you can download immediately if you are interested.

Data Analysis with Small Samples and Non-Normal Data:


Nonparametrics and Other Strategies Carl F. Siebert

https://textbookfull.com/product/data-analysis-with-small-samples-and-
non-normal-data-nonparametrics-and-other-strategies-carl-f-siebert/

textbookfull.com

Information Security Science Measuring the Vulnerability


to Data Compromises 1st Edition Carl Young

https://textbookfull.com/product/information-security-science-
measuring-the-vulnerability-to-data-compromises-1st-edition-carl-
young/
textbookfull.com

Outofshapeworthlessloser 1st Edition Gracie Gold

https://textbookfull.com/product/outofshapeworthlessloser-1st-edition-
gracie-gold/

textbookfull.com

Voices of Komagata Maru Imperial Surveillance and Workers


from Punjab in Bengal 1st Edition Suchetana Chattopadhyay

https://textbookfull.com/product/voices-of-komagata-maru-imperial-
surveillance-and-workers-from-punjab-in-bengal-1st-edition-suchetana-
chattopadhyay/
textbookfull.com
Chinese FDI in the EU and the US: Simple Rules for
Turbulent Times Tim Wenniges

https://textbookfull.com/product/chinese-fdi-in-the-eu-and-the-us-
simple-rules-for-turbulent-times-tim-wenniges/

textbookfull.com

Interventional Radiology for Medical Students 1st Edition


Hong Kuan Kok

https://textbookfull.com/product/interventional-radiology-for-medical-
students-1st-edition-hong-kuan-kok/

textbookfull.com

Marijuana Harvest How to Maximize Quality and Yield in


Your Cannabis Garden Ed Rosenthal

https://textbookfull.com/product/marijuana-harvest-how-to-maximize-
quality-and-yield-in-your-cannabis-garden-ed-rosenthal/

textbookfull.com

Data Management at Scale Best Practices for Enterprise


Architecture 1st Edition Piethein Strengholt

https://textbookfull.com/product/data-management-at-scale-best-
practices-for-enterprise-architecture-1st-edition-piethein-strengholt/

textbookfull.com

Elements of Optical Networking: Basics and Practice of


Glass Fiber Optical Data Communication 2nd Edition
Brückner
https://textbookfull.com/product/elements-of-optical-networking-
basics-and-practice-of-glass-fiber-optical-data-communication-2nd-
edition-bruckner/
textbookfull.com
Nanostructures for Drug Delivery. A volume in Micro and
Nano Technologies Ecaterina Andronescu And Alexandru Mihai
Grumezescu (Eds.)
https://textbookfull.com/product/nanostructures-for-drug-delivery-a-
volume-in-micro-and-nano-technologies-ecaterina-andronescu-and-
alexandru-mihai-grumezescu-eds/
textbookfull.com
The science and strategy of customer retention

Carl Gold
Foreword by Tien Tzuo

MANNING
Churn-reducing strategy Core concepts/customer metrics Book chapters
1. Product improvement • Cohorts of metrics based on product- 3, 5, and 7
• Make more of the best use events identify engaging and
features. disengaging product features.
• Make the best features easy • Identify optimal feature-use ratios.
to find.

2. Engagement marketing • Metric cohorts provide benchmarks 5 and 7


• Promote the best features. for healthy levels of product use.
• Targeted product insights. • Segment customers with metrics.

3. Pricing and packaging • Unit cost and unit value metrics 6 and 7
• Differentiate pricing to identify customers getting high/low
provide value without value on the product.
discounting.
• Correlations show relationships between
• Monetize valued groups the use of different features/content.
of product features.

4. Customer success and • Metric cohorts benchmark healthy use 5, 8, and 9


support levels.
• Help customer in need. • Forecast customer risk with regression
and machine learning.
• Identify failing customers
proactively. • Metric-driven customer conversations.
• Provide training at key • Account tenure cohort analysis measures
points in customer journey. risk along the customer journey.

5. Channel targeting • Category cohort churn analysis with 10


• Identify your best confidence intervals.
customer channels. • Identify the best/worst sales channels
• Find lookalikes. and demographic/firmographic indicators
of success.
Fighting Churn with Data
Fighting Churn
with Data
THE SCIENCE AND STRATEGY OF
CUSTOMER RETENTION

CARL GOLD
FOREWORD BY TIEN TZUO

MANNING
SHELTER ISLAND
For online information and ordering of this and other Manning books, please visit
www.manning.com. The publisher offers discounts on this book when ordered in quantity.
For more information, please contact
Special Sales Department
Manning Publications Co.
20 Baldwin Road
PO Box 761
Shelter Island, NY 11964
Email: [email protected]

©2020 by Manning Publications Co. All rights reserved.

No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in


any form or by means electronic, mechanical, photocopying, or otherwise, without prior written
permission of the publisher.

Many of the designations used by manufacturers and sellers to distinguish their products are
claimed as trademarks. Where those designations appear in the book, and Manning Publications
was aware of a trademark claim, the designations have been printed in initial caps or all caps.

Recognizing the importance of preserving what has been written, it is Manning’s policy to have
the books we publish printed on acid-free paper, and we exert our best efforts to that end.
Recognizing also our responsibility to conserve the resources of our planet, Manning books
are printed on paper that is at least 15 percent recycled and processed without the use of
elemental chlorine.

Development editor: Toni Arritola


Technical development editor: Mike Shepard
Review editor: Ivan Martinović
Manning Publications Co. Production editor: Deirdre S. Hiam
20 Baldwin Road Copy editor: Frances Buran/Tiffany Taylor/
PO Box 761 Keir Simpson
Shelter Island, NY 11964 Proofreader: Pamela Hunt
Technical proofreader: Al Krinker
Typesetter: Dennis Dalinnik
Cover designer: Marija Tudor

ISBN: 9781617296529
Printed in the United States of America
brief contents
PART 1 BUILDING YOUR ARSENAL ..............................................1
1 ■ The world of churn 3
2 ■ Measuring churn 39
3 ■ Measuring customers 80
4 ■ Observing renewal and churn 134

PART 2 WAGING THE WAR . ....................................................171


5 ■ Understanding churn and behavior with metrics 173
6 ■ Relationships between customer behaviors 217
7 ■ Segmenting customers with advanced metrics 258

PART 3 SPECIAL WEAPONS AND TACTICS . ................................315


8 ■ Forecasting churn 317
9 ■ Forecast accuracy and machine learning 367
10 ■ Churn demographics and firmographics 407
11 ■ Leading the fight against churn 452

v
contents
foreword xv
preface xvii
acknowledgments xix
about this book xxi
about the author xxv
about the cover illustration xxvi

PART 1 BUILDING YOUR ARSENAL ....................................1

1 The world of churn 3


1.1 Why you are reading this book
The typical churn scenario 5 ■
4
What this book is about 5
1.2 Fighting churn 7
Interventions that reduce churn 7 Why churn is hard to fight 9

Great customer metrics: Weapons in the fight against churn 12


1.3 Why this book is different 14
Practical and in-depth 14 ■
Simulated case study 16
1.4 Products with recurring user interactions 16
Paid consumer products 17 Business-to-business services 17

Ad-supported media and apps 19 Consumer feed subscriptions



19
Freemium business models 19 In-app purchase models 20

vii
viii CONTENTS

1.5 Nonsubscription churn scenarios 20


Inactivity as churn 20 Free trial conversion 20

Upsell/down sell 20 Other yes/no (binary) customer


predictions 21 Customer activity predictions 21


Use cases that are not like churn 21


1.6 Customer behavior data 22
Customer events in common product categories 22 ■
The most
important events 25
1.7 Case studies in fighting churn 26
Klipfolio 26 Broadly
■ 27 ■ Versature 28 ■ Social network
simulation 29
1.8 Case studies in great customer metrics 30
Utilization 30 ■ Success rates 32 ■ Unit cost 35

2 Measuring churn
2.1
39
Definition of the churn rate 43
Calculating the churn rate and retention rate 44
The relationship between churn rate and retention rate 45
2.2 Subscription databases 46
2.3 Basic churn calculation: Net retention 48
Net retention calculation 48 ■ SQL net retention calculation 50
Interpreting net retention 53
2.4 Standard account-based churn 56
Standard churn rate definition 56 Outer joins for churn

calculation 57 Standard churn calculation with SQL 57


When to use the standard churn rate 60


2.5 Activity (event-based) churn for nonsubscription
products 61
Defining an active account and churn from events 61
Activity churn calculations with SQL 62
2.6 Advanced churn: Monthly recurring revenue (MRR)
churn 64
MRR churn definition and calculation 64 MRR churn ■

calculation with SQL 66 MRR churn vs. account churn vs. net

(retention) churn 68
2.7 Churn rate measurement conversion 70
Survivor analysis (advanced) 70 Churn rate conversions 72

Converting any churn measurement window in SQL 74 Picking the ■

churn measurement window 76 ■


Visit https://textbookfull.com
now to explore a rich
collection of eBooks, textbook
and enjoy exciting offers!
CONTENTS ix

3 Measuring customers
3.1 From events to metrics 83
80

3.2 Event data warehouse schema 85


3.3 Counting events in one time period 86
3.4 Details of metric period definitions 88
Weekly behavioral cycles 88 ■
Timestamps for metric
measurements 90
3.5 Making measurements at different points in time 91
Overlapping measurement windows 91 Timing metric ■

measurements 94 Saving metric measurements 95


Saving metrics for the simulation examples 98


3.6 Measuring totals and averages of event properties 98
3.7 Metric quality assurance 100
Testing how metrics change over time 100 Metric quality ■

assurance (QA) case studies 103 Checking how many accounts


receive metrics 106


3.8 Event QA 108
Checking how events change over time 108 ■
Checking events per
account 111
3.9 Selecting the measurement period for behavioral
measurements 114
3.10 Measuring account tenure 117
Account tenure definition 117 Recursive table expressions

for account tenure 119 Account tenure SQL program 121


3.11 Measuring MRR and other subscription metrics 124


Calculating MRR as a metric 125 Subscriptions for specific

amounts 127 Calculating subscription unit quantities as


metrics 128 Calculating the billing period as a metric 129


4 Observing renewal and churn


4.1 Introduction to datasets 136
134

4.2 How to observe customers 137


Observation lead time 137 Observing sequences of renewals and a

churn 139 Overview of creating a dataset from subscriptions 140


4.3 Identifying active periods from subscriptions 142


Active periods 142 Schema for storing active periods 143

Finding active periods that are ongoing 144 Finding active ■

periods ending in churn 146


x CONTENTS

4.4 Identifying active periods for nonsubscription


products 151
Active period definition 151 Process for forming datasets from

events 152 SQL for calculating active weeks 154


4.5 Picking observation dates 156


Balancing churn and nonchurn observations 156 Observation ■

date-picking algorithm 157 Observation date SQL program 158


4.6 Exporting a churn dataset 161


Dataset creation SQL program 162
4.7 Exporting the current customers for segmentation 166
Selecting active accounts and metrics 166 ■ Segmenting customers
by their metrics 168

PART 2 WAGING THE WAR ...........................................171

5 Understanding churn and behavior with metrics


5.1 Metric cohort analysis 176
The idea behind cohort analysis 176 Cohort analysis ■
173

with Python 179 Cohorts of product use 182 Cohorts of


■ ■

account tenure 185 Cohort analysis of billing period 187


Minimum cohort size 188 Significant and insignificant cohort


differences 190 Metric cohorts with a majority of zero customer


metrics 191 Causality: Are the metrics causing churn? 192


5.2 Summarizing customer behavior 193


Understanding the distribution of the metrics 194 Calculating dataset ■

summary statistics in Python 196 Screening rare metrics 198


Involving the business in data quality assurance 198


5.3 Scoring metrics 199
The idea behind metric scores 199 The metric score ■

algorithm 201 Calculating metric scores in Python 203


Cohort analysis with scored metrics 205 Cohort analysis of ■

monthly recurring revenue 207


5.4 Removing unwanted or invalid observations 209
Removing nonpaying customers from churn analysis 209
Removing observations based on metric thresholds in Python 210
Removing zero measurements from rare metric analyses 212
Disengaging behaviors: Metrics associated with increasing churn 213
5.5 Segmenting customers by using cohort analysis 214
Segmenting process 215 ■ Choosing segment criteria 215
CONTENTS xi

6 Relationships between customer behaviors


6.1 Correlation between behaviors 219
Correlation between pairs of metrics 219 Investigating ■
217

correlations with Python 223 Understanding correlations


between sets of metrics with correlation matrices 225 Case ■

study correlation matrices 227 Calculating correlation


matrices in Python 228


6.2 Averaging groups of behavioral metrics 230
Why you average correlated metric scores 230 Averaging ■

scores with a matrix of weights (loading matrix) 231 Case ■

study for loading matrices 233 Applying a loading matrix


in Python 234 Churn cohort analysis on metric group


average scores 238


6.3 Discovering groups of correlated metrics 239
Grouping metrics by clustering correlations 240 Clustering ■

correlations in Python 242 Loading matrix weights that make


the average of scores a score 248 Running the metric grouping


and grouped cohort analysis listings 250 Picking the correlation


threshold for clustering 251


6.4 Explaining correlated metric groups to
businesspeople 253

7 Segmenting customers with advanced metrics


7.1 Ratio metrics 261
258

When to use ratio metrics and why 261 How to calculate ■

ratio metrics 264 Ratio metric case study examples 270


Additional ratio metrics for the simulated social network 272


7.2 Percentage of total metrics 273
Calculating percentage of total metrics 273 Percentage of total■

metric case study with two metrics 277 Percentage of total metrics

case study with multiple metrics 279


7.3 Metrics that measure change 280
Measuring change in the level of activity 281 Scores for metrics ■

with extreme outliers (fat tails) 285 Measuring the time since the

last activity 291


7.4 Scaling metric time periods 294
Scaling longer metrics to shorter quoting periods 295 ■
Estimating
metrics for new accounts 299
xii CONTENTS

7.5 User metrics 304


Measuring active users 305 ■
Active user metrics 307
7.6 Which ratios to use 308
Why use ratios, and what else is there? 309 ■
Which ratios
to use? 310

PART 3 SPECIAL WEAPONS AND TACTICS .......................315

8 Forecasting churn
8.1
317
Forecasting churn with a model 318
Probability forecasts with a model 318 Engagement and ■

retention probability 319 Engagement and customer


behavior 321 An offset matches observed churn rates to


the S curve 323 The logistic regression probability


calculation 325
8.2 Reviewing data preparation 326
8.3 Fitting a churn model 329
Results of logistic regression 330 Logistic regression code 332

Explaining logistic regression results 335 Logistic regression case ■

study 338 Calibration and historical churn probabilities 338


8.4 Forecasting churn probabilities 341


Preparing the current customer dataset for forecasting 341
Preparing the current customer data for segmenting 346
Forecasting with a saved model 347 Forecasting case ■

studies 350 Forecast calibration and forecast drift 351


8.5 Pitfalls of churn forecasting 352


Correlated metrics 353 ■ Outliers 356
8.6 Customer lifetime value 360
The meaning(s) of CLV 360 From churn to expected customer

lifetime 363 CLV formulas 364


9 Forecast accuracy and machine learning


9.1 Measuring the accuracy of churn forecasts
Why you don’t use the standard accuracy measurement for
367
369

churn 369 Measuring churn forecast accuracy with the


AUC 372 Measuring churn forecast accuracy with the lift


■ 375
9.2 Historical accuracy simulation: Backtesting 380
What and why of backtesting 380 Backtesting code ■ 382
Backtesting considerations and pitfalls 384
CONTENTS xiii

9.3 The regression control parameter 385


Controlling the strength and number of regression weights 386
Regression with the control parameter 387
9.4 Picking the regression parameter by testing
(cross-validation) 389
Cross-validation 389 Cross-validation code
■ 390
Regression cross-validation case studies 393
9.5 Forecasting churn risk with machine learning 394
The XGBoost learning model 395 XGBoost cross-

validation 397 Comparison of XGBoost accuracy to


regression 400 Comparison of advanced and basic metrics



402
9.6 Segmenting customers with machine learning
forecasts 404

10 Churn demographics and firmographics


10.1 Demographic and firmographic datasets
407
409
Types of demographic and firmographic data 409 Account data■

model for the social network simulation 410 Demographic


dataset SQL 411


10.2 Churn cohorts with demographic and firmographic
categories 414
Churn rate cohorts for demographic categories 415 Churn rate ■

confidence intervals 416 Comparing demographic cohorts with


confidence intervals 417


10.3 Grouping demographic categories 425
Representing groups with a mapping dictionary 425 Cohort ■

analysis with grouped categories 426 Designing category


groups 428
10.4 Churn analysis for date- and numeric-based
demographics 430
10.5 Churn forecasting with demographic data 432
Converting text fields to dummy variables 432 Forecasting churn

with categorical dummy variables alone 435 Combining dummy


variables with numeric data 437 Forecasting churn with


demographic and metrics combined 441


10.6 Segmenting current customers with demographic
data 444
xiv CONTENTS

11 Leading the fight against churn


11.1 Planning your own fight against churn
452

Data processing and analysis checklist 455 ■


453
Communication to
the business checklist 457
11.2 Running the book listings on your own data 459
Loading your data into this book’s data schema 459 ■
Running
the listings on your own data 460
11.3 Porting this book’s listings to different environments 461
Porting the SQL listings 462 ■
Porting the Python listings 462
11.4 Learning more and keeping in touch 463
Author’s blog site and social media 463 Sources for churn

benchmark information 463 Other sources of information


about churn 464 Products that help with churn 464


index 467
foreword
This book is a rarity. Although it’s intended primarily for technically oriented people
with some familiarity with coding and data, it also happens to be lucid, compelling,
and occasionally even (gasp!) funny. The first chapter in particular should be manda-
tory reading for anyone who’s interested in running a successful subscription-based
business. Buy a copy for your boss.
It’s exciting to think about all the different companies that will benefit from the
sharp analysis in these pages. Data folks from all sectors of the global economy, from
streaming-media services to industrial manufacturers, will be paying close attention to
Carl’s book. Today, the whole world runs “as a service”: transportation, education,
media, health care, software, retail, manufacturing, you name it.
All these new digital services are generating vast amounts of data, resulting in a
huge signal-to-noise challenge, which is why this book is so important. I study this
topic for a living, and no one has written such a practical and authoritative guide to
effectively filtering through all that information to reduce churn and keep subscribers
happy. When it comes to running a subscription business, churn rates are a matter of
life and death!
Thousands of entrepreneurs are already deeply familiar with Carl Gold’s work. He
is the author of the Subscription Economy Index, a biannual benchmark study that
reflects the growth metrics of hundreds of subscription companies spread across a
variety of industries. As Zuora’s chief data scientist, Carl works with the most timely
and accurate dataset in the subscription economy. He’s a big part of why Zuora is not
only a successful software company but also a respected thought leader.

xv
xvi FOREWORD

If you’re reading this book, you will soon have the ability to make immediate and
material contributions to the success of your company. But as Carl discusses exten-
sively throughout the book, it’s not enough to do the analysis; you also need to be able
to communicate your results to the business at large.
So by all means, use this book to learn how to conduct the proper analysis, but also
use it to learn how to share, execute, and basically excel at your job. There are exam-
ples and case studies and tips and benchmarks galore. How lucky are we? We get to
work in the early days of the subscription economy, and we get to read the first land-
mark book on churn.

—Tien Tzuo, founder and CEO, Zuora


preface
Customer churn (cancelations) and engagement are life-and-death issues for every
company that offers an online product or service. Coinciding with the wide adoption
of data science and analytics, it is now standard to call in data professionals to help in
the effort to reduce churn. But understanding churn has many challenges and pitfalls
not common to other data applications, and until now, there has not been a book to
help a data professional (or student) get started in this area.
Over the past six years, I have worked on churn for dozens of products and ser-
vices, and served as the chief data scientist at a company called Zuora. Zuora provides
a platform for subscription companies to manage their products, operations, and
finances, and you will see some Zuora customers in case studies throughout the book.
During that time, I experimented with different ways to analyze churn and feed the
results back to people at companies that were fighting churn. The truth is that I made
a lot of mistakes in the early years, and I was inspired to write this book to save other
people from making the same mistakes that I made.
The book is written from the point of view of a data person: whoever is expected to
take the raw data and come up with useful findings to help in the fight against churn.
That person may have the title of data scientist, data analyst, or machine learning
engineer. Or they may be someone else who knows a bit about data and code and is
being asked to fill those shoes. The book uses Python and SQL, so it does assume that
the data person is a coder. Although I advocate spreadsheets for presentation and
sharing data (as I detail in the book), I do not recommend attempting the main ana-
lytic tasks of churn fighting in spreadsheets: many tasks must be performed in

xvii
xviii PREFACE

sequence, and some of these tasks are nontrivial. Also, there is a need to “rinse and
repeat” the process multiple times. That kind of workflow is well suited to short pro-
grams but difficult in spreadsheets and graphical tools.
Because the book is written for a data person, it does not go into details on the
churn-reducing actions that products and services can take. So this book does not
contain details on how to do things like run email and call campaigns, create churn-
save playbooks, and design pricing and packaging. Instead, this book is strategic in
that it teaches a data-driven approach to devising your battle plan against churn: pick-
ing which churn-reducing activities to pursue, which customers to target, and what
kinds of results to expect. That said, I will introduce various churn-reducing tactics at
a high level as is necessary to understand the context for using the data.
Visit https://textbookfull.com
now to explore a rich
collection of eBooks, textbook
and enjoy exciting offers!
acknowledgments
There are many people without whom it would not have been possible for me to cre-
ate this book for you.
Starting at the beginning, I thank Ben Rigby for bringing me to my first churn case
study and everyone who worked at Sparked (Chris Purvis, Chris Mielke, Cody Chapman,
Collin Wu, David Nevin, Jamie Doornboss, Jeff Nickerson, Jordan Snodgrass, Joseph
Pigato, Mark Nelson, Morag Scrimgoeur, Rabih Saliba, and Val Ornay) and all the cus-
tomers of Retention Radar. Next, I have Tien Tzuo and Marc Aronson to thank for
bringing me to Zuora, and thanks to Tom Krackeler, Karl Goldstein, and everyone
from Frontleaf (Amanda Olsen, Greg McGuire, Marcelo Rinesi, and Rachel English)
for welcoming me to their team. Continuing in chronological order, I also thank
everyone who worked on or with the Zuora Insights team (Azucena Araujo, Caleb
Saunders, Gail Jimenez, Jessica Hartog, Kevin Postlewaite, Kevin Suer, Matt Darrow,
Michael Lawson, Patrick Kelly, Pushkala Pattabhiraman, Shalaka Sindkar, and Steve
Lotito), the data scientist on my team who worked on churn (Dashiell Stander), and
all the Zuora Insights customers. All these people were part of the projects on which I
learned what I now know about churn; in that way, they made it possible for me to
write this book for you. And I want to thank everyone at Zuora who either helped pro-
mote or edit the book: Amy Konary, Gabe Weisert, Helena Zhang, Jayne Gonzalez,
Kasey Radley, Lauren Glish, Peishan Li, and Sierra Dowling.
Next comes my publisher, Manning, where I thank my first acquisitions editor,
Stephen Soehnlen, for bringing me on board; my main development editor, Toni
Arritola, and my temporary DE, Becky Whitney, for patiently teaching me how to write

xix
xx ACKNOWLEDGMENTS

a Manning-style book; and my second AE, Michael Stephens, for getting the book
across the finish line. I also thank my technical and code editors—Mike Shepard,
Charles Feduke, and Al Krinkler—and everyone who commented on the liveBook
forum during the early access period. My thanks also go to Deirdre Hiam, my project
editor; Pamela Hunt, my proofreader; and Frances Buran, Tiffany Taylor, and Keir
Simpson, my copyeditors. I would also like to thank all the reviewers: Aditya Kaushik,
Al Krinker, Alex Saez, Amlan Chatterjee, Burhan Ul Haq, Emanuele Piccinelli,
George Thomas, Graham Wheeler, Jasmine Alkin, Julien Pohie, Kelum Senanayake,
Lalit Narayana Surampudi, Malgorzata Rodacka, Michael Jensen, Milorad Imbra,
Nahid Alam, Obiamaka Agbaneje, Prabhuti Prakash, Raushan Jha, Simone Sguazza,
Stefano Ongarello, Stijn Vanderlooy, Tiklu Ganguly, Vaughn DiMarco, and Vijay
Kodam. Your suggestions helped make this book better.
Special thanks go to the three companies that allowed me to present a selection of
their case study data to bring the material in the book to life: Matt Baker and everyone
at Broadly; Yan Kong and everyone at Klipfolio; and Jonathan Moody, Tyler Cooper,
and everyone at Versature.
Finally, I thank my wife, Anna, and children, Clive and Skylar, for their support and
patience during a challenging but fruitful time.
about this book
This book was written to enable anyone with a little background in coding and data to
make a game-changing analysis of customer churn for an online product or service.
And if you are experienced in programming and data analysis, the book contains tips
and tricks for churn and customer engagement that you won’t find anywhere else.

Who should read this book


The primary audience for this book is data scientists, data analysts, and machine
learning engineers. You will want this book when you are tasked with helping under-
stand and fight churn for an online product or service. Also, the book is absolutely
suitable for students of computer science and data science, or anyone who knows how
to code and wants to learn more about an important area of data science at a typical
modern company. Because the book begins with raw data and provides the necessary
background on every analytic task described, it reads as a complete hands-on course
in data science, taught on a consistent project: analyzing churn for a small company.
(A sample dataset is provided.)
That said, chapters 8 and 9 in part 3 of the book, on forecasting and machine
learning for churn, may entail a steep learning curve for someone who does not have
some experience on the subjects it covers. If you don’t have that background, I think
you can still learn everything you need to know in chapters 8 and 9, but you may have
to spend extra time to read some of the recommended online resources.
This book should also be read by noncoding business professionals. The book
includes a unique set of case study observations about churn at real companies. The

xxi
xxii ABOUT THIS BOOK

book explains the data typically available for analyzing churn, the practices used to
turn that data into actionable intelligence, and the most typical findings. One empha-
sis of the book is how to communicate data results to businesspeople; consequently, all
the important takeaways are explained in plain English (no jargon!). So if you care about
churn but aren’t a coder, you should skim the book for the takeaways (clearly labeled)
and skip the coding and math. Then share the book with one of your developers to
get help putting the concepts into action.

How this book is organized: A road map


The book is organized to take you step-by-step through a specific process: the process
a data person at an online company should go through when they harness raw data to
drive the fight against churn. As such, the book is best read in order, chapter by chap-
ter. That said, the material in the book is front-loaded in the following two senses:
■ In every chapter, the most important topics are taught first, and details about
less common scenarios come at the end of the chapter.
■ The most important lessons come in the earliest chapters, and the topics in
later chapters are more specialized.
So if you find yourself near the end of a chapter that doesn’t seem to be relevant to
your scenario, there usually is no harm in skipping to the next chapter. Also, if you are
pressed for time and need to master the basics, you can try to take one of these abbre-
viated reading paths:
■ To get the foundations, read chapters 1–3 plus section 4.5, which corresponds
to reading almost all of part 1 (skipping all but one section of chapter 4).
■ To get an advanced course without the most specialized subjects, read chap-
ters 1–7, which corresponds to reading parts 1 and 2.
More details on these abbreviated courses of reading and how to apply the learnings
are given in chapter 11.
The book is divided into three parts. Part 1 explains what churn is and how to mea-
sure it, what data companies typically have available to help them understand and
reduce churn, and how to prepare the data to make it useful:
■ Chapter 1 is a general introduction to the field and includes an introduction to
the case studies, highlighting the type of intelligence the book will help you
achieve for your own product and service.
■ Chapter 2 explains how to identify churned customers and measure churn in a
variety of ways. SQL code begins in this chapter.
■ Chapter 3 introduces the creation of customer metrics from the event data that
most online companies collect about their users.
■ Chapter 4 explains how to combine the churn data from chapter 2 with the
metrics from chapter 3 to create an analytic dataset for understanding and
fighting churn.
ABOUT THIS BOOK xxiii

Part 2, which contains the core techniques in the book, is devoted to understanding
how customer behavior relates to churn and retention and using that knowledge to
drive churn-reducing strategies:
■ Chapter 5 teaches a form of cohort analysis, which is the primary method for
understanding and explaining the relationship between behaviors and churn.
Chapter 5 also includes many case study examples, and the code is in Python.
■ Chapter 6 looks at how to deal with data that is big in an undesirable way: most
company datasets have closely related measurements of the same underlying
behavior. How you deal with this somewhat-redundant information is important.
■ Chapter 7 returns to the subject of metric creation and uses the information
from chapters 5 and 6 to design advanced metrics, which help explain complex
customer behaviors such as price sensitivity and efficiency.
Part 3 covers forecasting with regression and machine learning. When it comes to
reducing churn, forecasting is less important than having a good set of metrics, but it
can still be useful, and some special techniques are needed to get it right:
■ Chapter 8 teaches how to forecast customer churn probabilities with a regres-
sion and how to interpret the results of those forecasts, including calculating
customer lifetime value.
■ Chapter 9 is about machine learning and measuring and optimizing the accu-
racy of churn forecasts.
■ Chapter 10 covers analyzing demographic or firmographic data in the context
of churn and finding lookalikes for your best customers.
Most readers should start at the beginning and read parts 1 and 2. If, after learning
and applying those techniques, you need to make forecasts or find lookalike custom-
ers, continue to part 3. If you are already using advanced analytics, you may be able to
skip part 1 and start in part 2 and/or 3. For purposes of this book, being advanced in
analytics means that you already have a good set of customer metrics and can identify
and measure churned customers. Otherwise, start with part 1.

About the code


The book contains code listings in SQL and Python. Each listing represents one small
step in the process of preparing data, understanding why customers churn, and reduc-
ing churn:
■ All the code from the book is available in the author’s GitHub repository at
https://github.com/carl24k/fight-churn.
■ The GitHub repository also provides a Python wrapper program to run both SQL
and Python listings. That program is the recommended way to run the code.
■ The book contains examples you can run on a simulated set of customer data,
designed to look like the data that would be generated by users of a small
online service: a social network with 10,000 customers.
Random documents with unrelated
content Scribd suggests to you:
The Project Gutenberg eBook of Jason, Son of
Jason
This ebook is for the use of anyone anywhere in the United States
and most other parts of the world at no cost and with almost no
restrictions whatsoever. You may copy it, give it away or re-use it
under the terms of the Project Gutenberg License included with this
ebook or online at www.gutenberg.org. If you are not located in the
United States, you will have to check the laws of the country where
you are located before using this eBook.

Title: Jason, Son of Jason

Author: J. U. Giesy

Release date: March 19, 2022 [eBook #67655]

Language: English

Original publication: United States: New Publications, Inc

Credits: Greg Weeks, Mary Meehan and the Online Distributed


Proofreading Team at http://www.pgdp.net

*** START OF THE PROJECT GUTENBERG EBOOK JASON, SON OF


JASON ***
JASON, SON OF JASON
By J. U. Gíesy
CHAPTER I
THE GATEWAY OF LIFE
It was midnight when the night superintendent called and told me
No. 27 had died. I rose. The thing was no surprise. I had known it
was going to happen. No. 27 had told me so himself. None the less,
I went to his room. Routine in the mental hospital had nothing to do
with that strange secret held in common between myself and the
man—that strange state of affairs which had enabled him to
predicate his own death so accurately.
And yet as I mounted the stairs to the room where his body now lay
as a worn-out husk I had none of the feeling which so customarily
assails the average mortal in such an hour. To me it was not as
though he had died. To my mind in those moments it was no more
than the casting aside by the activating spirit of that instrument
which for its own ends it had used. The body then was a husk
indeed—an emaciated, worn-out thing which, because of our mutual
secret, I knew had been kept alive by the sheer force of the spiritual
tenant, now removed.
I stood looking down upon it, with very much the same sensations
one might have in viewing the tool once plied by the hand of a
friend. It was nothing more than that really. Jason Croft had used it
while he had need of its manipulation, and when his need was
accomplished he had simply laid it down.
Jason Croft. Dead? I felt an impulse to smile in most improper
fashion. Not at all. The man was not only not dead, but I knew—as
positively as I knew I was presently going to leave the room where
his dead shell lay on a hospital bed and return to my own quarters—
exactly where he had gone.
The statement sounds a bit as though I were better qualified as an
inmate than the superintendent of an institution for the care of the
insane. And I don't suppose it will help any for me to add that I had
seen Jason Croft die before—or that he had informed me on the
former occasion, though in less specific fashion, of his approaching
end.
That was after he had told me a most remarkable tale, which, in
spite of its almost incredible nature, I found myself strongly inclined
to believe. It had concerned Croft's adventures on another planet—
Palos—one of the spheres in the universe of the Dog Star Sirius, to
which he had traveled first by astral projection, but on which he had
found means to establish an actual existence in the flesh.
"Unbelievable—can a man be dead and yet live again?" you will say.
Well, yes, but—Croft's earth body died just as he had told me it
would, and was buried, and time passed, and this patient No. 27
was committed to the institution of which I was the head; and when
I went to examine and inspect him, he asked me to dismiss the
attendants, and then he spoke to me in the voice of Jason Croft.
More than that, he took up the story of his adventures where he had
left off in the previous instance, admitted freely that he had reversed
the experiment by which he had gained material existence on Palos,
and, driven by the necessity of gaining knowledge for use in his new
estate, had deliberately returned to earth. Unbelievable, you will say
again. And again I answer:
"Yes—but wait."
Croft was a physician, even as am I. He was a scientific man. In
addition he was a student of the occult—the science of the mind, the
spirit, and its control of the physical forces of life.
He was an earth-born man. The home in which I first met him
contained the greatest private collection of works on the subject I
have ever seen. In dying he left them to me—I have them all about
me. They are mine. According to his statements and his notations on
margins, he had gone so far in his investigations that he could
project the astral consciousness anywhere at will. And when I say
anywhere, I mean it in the literal sense.
Many men have mastered the astral control on the earthly plane.
Croft had carried it to an ultimate degree. He shook off the envelope
of the earth atmosphere, led thereto, as he frankly confessed in our
conversations, by the attraction of a feminine spirit, though he did
not know it at the time, and recognized it only when he first viewed
Naia—Princess of Tamarizia—on a distant star.
I had dabbled in the occult to some extent myself. Hence when he
spoke of the doctrine of twin souls he had no further need to
explain. He alleged that since a child the Dog Star had called him
subtly through the years in a way he could not explain. Once having
come into her presence, however, he knew that it was Naia—the
feminine counterpart of his nature—whose existence on the other
planet had called across the void to him. Or so he claimed. And
certainly his portrayal of the events on Palos were characterized by a
detail that made the atmosphere of his alleged other existence most
vividly plain.
To an accomplishment of his marrying her, Croft declared that he
had done a weirdly wonderful thing. Discovering a Palosian dying of
a mental rather than a physical ailment, he had waited until his
death occurred, then appropriated the still physically viable body to
himself, as he most comprehensively explained, describing his act in
a scientific way that counseled belief while staggering the mind.
Over that body he obtained absolute control, exactly as he had
gained the same ability with his own. For a time thereafter he led a
sort of dual existence, sometimes on Palos, sometimes on earth,
until he had fully shaped his plans. Then, and then only, did he
voluntarily forsake the mundane life to enter that other and fuller
existence he felt that Naia of Aphur could make complete.
I questioned him closely. I was faced by a most amazing thing. I
took up first the question of time required in passing from earth to
Palos. He smiled and replied that outside the mental atmosphere a
man's time ceased to exist; that it was man's measure of a portion
of eternity, and nothing more, and that he could not use what was
non-existent, hence reached Palos as quickly in the astral condition
as I could span the gulf between that member of the Dog Star's
Pack and earth in thought. All other points I raised he met. Even so
it was a good deal of a shock to find my new patient speaking to me
with Croft's evident understanding, looking at me out of what
seemed oddly like Croft's eyes.
But in the end I was convinced. The man knew too much. He was
too utterly conversant with Croft's accomplishments, his aims and
ambitions and hopes, to be anyone but Croft himself. And, too, he
naïvely explained that it was a poor rule that would not work two
ways, and that he had therefore repeated his experiment in gaining
a Palosian body when he felt the pressing need of a return to earth.
This night, earlier in the evening, he had bidden me goodbye—told
me he was going back to Naia, the woman he had dared so much to
win, his mate who ere long was to bear him, Jason Croft of Earth, a
child. And now—well, now as before, it would seem he had kept his
word. Jason Croft was dead again.
Is it any wonder that I felt that strange, almost amused desire to
smile? Dead! Why, Croft, in so far as I knew him, could practically
laugh at death—he was a man who had actually demonstrated, if
one believed his narrative, of course, the truth of the saying that the
spirit is the life. He was a man, who, because of the needs of his
spirit, had deliberately switched his existence from one to the other
of two spheres.
I gave what directions were needed for the disposal of No. 27's
body, returned to my bed, and stretched myself out. But I didn't
sleep all that morning. I buried myself in thought.
Both the narratives to which I had listened—first from the man I
knew to be Jason Croft really, secondly from the pitiable wreck he
had employed on his return, that worn-out husk which had just died
—had produced on me a somewhat odd effect. So clearly had he
portrayed the events and emotions which had swayed him in his
almost undreamed courtship of the Aphurian princess that I had
come to accept the characters he mentioned as actually existent
persons, acquaintances almost, just as, in spite of all established
precedent, I still regarded Croft himself as alive.
Naia of Aphur—many a time as I listened to his account of their
association I had thrilled to the picture of that supple girl with her
crown of golden hair, her crimson lips, her violet-purple eyes. So real
she had come to seem that I had felt I would know her had I seen
her with my physical rather than my mental vision. So real indeed
was her mental picture that when he told me she was about to
become a mother I had cried out, on impulse, that I wished as a
medical man I might attend her—would be glad to see the light in
her eyes when they first beheld his, Jason's, child.
And Croft had replied, "Man, I could love you for that," and he
flashed me an understanding smile.
So now that he was gone back to her—I lay on my bed unsleeping,
and let all he had told me unroll in a sort of mental panorama,
dealing wholy with the Palosian world.
Tamarizia! It was into the empire Croft blundered blindly when he
went to Palos first—a series of principalities surrounding the shore of
a vast inland sea, with the exception of a central state—the seat of
the imperial capital, embracing the island of Hiranur located in the
sea itself, and Nodhur to the west and south. From the central sea a
narrow strait led into an outer ocean to the west.
This was known as the Gateway. To the north was Cathur, a rugged,
mountainous state, the seat of national learning, in its university at
the capital city of Scira, and east of Cathur was Mazhur, known as
the Lost State at the time of Croft's first arrival, because it had been
wrested from the empire some fifty years before, in a war with
Zollaria, a hostile nation to the north.
Croft, after gaining physical life on Palos, succeeded in winning it
back, and in gaining thereby the consent of Naia's father, Prince
Lakkon, and her uncle, Jadgor, King of Aphur, to their marriage. It
was at this point his narrative had ended first.
East of Mazhur, still hugging the sea and extending into the
hinterland of the continent was Bithur. And Milidhur joined Bithur to
the south. West of Milidhur, completing the circle, was Aphur—the
name meaning literally "the land to the west" or "toward the sun."
Aphur was the southern pillar of the Gateway, ending at the western
strait. Nodhur lay south of Aphur, gaining access to the sea by the
navigable river Na, on whose yellow flood moved a steady stream of
commerce driven by sail and oar until Croft revolutionized
transportation by producing alcohol-driven motors. And—if I were to
believe his second account—since then he had actually electrified the
nation, harnessing mountain streams to generate the force.
Except for the waterways, traffic prior to Croft's innovations was by
conveyances drawn by the gnuppa—a creature half deer, half horse,
in appearance—or by means of caravans of the enormous beast
called sarpelca, resembling some huge Silurian lizard, twice the size
of an elephant, with a pointed tail, scale-armored back, camel-like
neck, and the head of a marine serpent tentacle-fringed about the
mouth.
They were driven by reins affixed to these fleshy appendages, and
streamed across the Palosian deserts, bearing huge merchandise
cargoes upon their massive backs.

Indeed, it was a wonderful world into which Croft had projected


himself. Babylonian in seeming he had described it to me at first.
North of Tamarizia was Zollaria, inhabited by a far more warlike race.
Its despotic government had long cast a covetous eye on the Central
Sea, through which, and the rivers emptying into its expanse, most
of the profitable trade lanes were reached. Tamarizia, controlling the
western Gateway, had remained master even after the fall of Mazhur,
collecting toll from the Zollarian craft on her rivers despite the
foothold gained on her northern coast.
East of Tamarizia, beyond Bithur and Milidhur, lay Mazzeria, peopled
by a race little above the aborigine in their social life. Tatar-like, the
Mazzerians shaved their heads of all save a single tuft of hair, with a
most remarkable effect, since the race was blue of complexion and
the prevailing color of their hair was red.
Mazzeria, at the time of Croft's incursion into the planet's affairs,
was the acknowledged ally of Zollaria, although at peace with
Tamarizia. In earlier times, however, numbers of them had been
taken captive in border wars and brought to both nations as slaves.
These, in so far as Tamarizia was concerned, had later been freed
and given citizenship of a degree constituting in their ranks the
lowest or serving caste.
Each state was governed by a king, by hereditary succession, in
conjunction with a national assembly consisting of a delegate elected
by each ten thousand or deckerton of civil population. The occupant
of the imperial throne was elected for a period of ten years by vote
of the several states.
On Croft's advent, Scythys—a dotard—had been king of Cathur, with
his son Kyphallos, the crown prince, a profligate of the worst type,
sunk under the charms of Kalamita, a Zollarian adventuress of great
beauty, with whom he had plotted the surrender of Cathur to her
nation in return for the Tamarizian throne with Kalamita by his side.
Jadgor of Aphur, scenting the danger, had sought to bind the
northern prince to Tamarizian fealty through a marriage with Naia,
his sister's child. To win Naia and overthrow Zollaria's scheme had
been Jason's task. The introduction of both the motor and firearms
enabled him to overthrow the flower of Zollaria's hosts on a couple
of bloody fields. Victory gained and Zollaria forced to cede Mazhur
after fifty years of occupation, Croft prevailed upon the nation to
accept a democratic form of government, it being at the end of
Emperor Tamhys's term. This was accomplished without too much
difficulty.
As to the Tamarizians themselves, they were a white and well-
formed race. Their women held equal place with men. They believed
in the spirit and a future life. They had made no small progress in
the sciences and arts. They worked metal, gold being as common as
iron on Palos.
They tempered copper also and used it in innumerable ways. They
wove fabrics of great beauty, one being a blend of vegetable fiber
and spun gold. They cut and polished jewels. They had a system of
judiciaries and courts and a medical and surgical knowledge of sorts.
They were a fairly moral and naturally modest people. Their clothing
was worn for protection and ornamentation, rather than for any
other purpose. It was donned and doffed as the occasion required,
without comment being aroused. In women it consisted, rich and
poor, of a single garment falling to the knee or just below it,
cinctured about the body and caught over one shoulder by a jeweled
or metal boss, leaving the other shoulder, arm, and upper chest
exposed. To this was added sandals of leather, metal, or wood, held
to the foot by a toe and instep band and lacings running well up the
calves.
Men of wealth and soldiers generally wore metal casings, jointed to
the sandal to permit of motion and extending upward to the knees.
Men of caste wore also a soft shirt or chemise beneath a metal
cuirass or embroidered tunic. Save on formal occasions the serving
classes wore a narrow cincture about the loins.
Agriculture was highly developed, and they had advanced far in
architecture, painting and sculpture. They lavished much time and
expense in beautifying their homes. They had well-constructed
caravan roads. As Croft had pointed out, he found them an
intelligent race waiting, ready to be trained to a wider craft.
And among them, in Naia of Aphur, he believed he had found his
twin soul. And he had set about winning her in a fashion such as no
other man, I frankly believe, would have dared.
He had won her according to his belief and returned to earth, for the
last time, ere he should return and make her his bride. He had told
me about it, and he had cast off his earthly body, severing the last
tie that held him from his life in Palos. He had died.
He had gone back and found his plans disarranged through the
actions of Zud, the high priest of Zitra, the capital city of Hiranur,
where he had left Naia waiting his return in the Temple of Ga, the
Eternal Mother—the Eternal Woman, in the Zitran pyramid. Zud,
moved by Croft's works and by a story told him by Abbu, a priest
who knew Jason's story, had proclaimed him Mouthpiece of Zitu,
thereby raising an insurmountable barrier, as it seemed, between
him and Naia, since celibacy was one of the tenets of the Tamarizian
priests. And yet Croft had won to her, overcoming all obstacles, even
winning a second war, with all Mazzeria egged on, her armies
officered by Zollarians in disguise this time, ere he gained the goal of
his desire.
These things had been told me inside the last few weeks by No. 27
—the man who had been committed to the institution for a
dissociation of personality, at which he quietly laughed after he had
obtained my ear; because he wished to gain contact with me, who
knew his former story, and win my aid toward the fulfillment of his
mission.
Only he wasn't dead, and I knew it as I lay there with the names of
men and women of the Palosian world buzzing in my head. He had
gone back to them, now that his work was ended—to Naia, his
golden-haired, purple-eyed mate—to Lakkon, her father; to Jadgor,
her uncle, and Robur his son, governor now of Aphur in the palace
where his father, president of the Tamarizian republic, had been
king; to Robur, who, like a second Jonathan, had ever been Croft's
loyal assistant and friend, and Gaya his sweet and matronly wife; to
Magur, high priest of Himyra, the ruling red city of Aphur, by whom
Croft and Naia were bethrothed to Zud himself, to whom he had
taught the truth of astral control. And I found myself portraying
them as Croft had described them, predicating their thoughts and
feelings, as I might have done those of any man or woman I knew
on earth.
Actually I was projecting my intellect, if not my consciousness, to
Palos. The thought came to me. In spirit, if not in perception, I was
there for the moment with my friend. In spirit at least I was bridging
with little effort billions of actual miles. Thought and spirit and soul.
They are strange things. Croft, if I was any judge, had gone back to
Naia—and there was I lying, picturing the scene, where she waited
for his coming in their home high in the western mountains of Aphur,
given to them by Lakkon, a wedding gift, after the war with Mazzeria
was won. Croft had gone back to Palos, and here was I picturing the
thing in my spirit, certainly as plainly as any earth scene I had ever
known.
His body would be lying there, covered with soft fabrics, waiting for
its tenant on a couch of wine-red wood such as the Tamarizians
used—or perhaps of molded copper. And Naia—the woman who had
given him her life, would be watching, watching for the first stir of
his returning.
Only—I smiled—Croft had told me he could gain Palos as quickly in
the consciousness as I could project myself there in my mind—so, by
now, that stirring of her strong man's limbs, beneath the eyes of the
fair watcher, had occurred, and once more those two were together.
I smiled again.
The picture of that reunion appealed. There was nothing else to it at
the instant. For even in my wildest imaginings I did not in the least
suspect what its nearness, its clearness, the vividness of its seeming,
might portend.
No, even though I myself had delved more or less deeply into occult
lore, with a resulting knowledge of the subject that had brought
about the sympathetic understanding of all Croft had told me from
first to last, I had little or no conception that night of the inward
meaning of the distinctness with which I could conjure up the scene
of his return to Naia, or to where the ability might lead. Rather, I felt
merely that through his narrative of her wooing he had built up
within my mental cells a picture of the fair girl now his bride, so
clear, so positive in seeming, that to me she appeared no more than
a charming personality—a feminine acquaintance, such as one might
on occasion meet. She was no more removed, so far as my feeling
of familiarity with her was concerned, than had her residence been
not on Palos, but simply across the street. It is so easy to bridge
distance in the mind.
I slept after a time, as one will, drifting from continued thought upon
one subject into slumber. And I woke with the thought of Croft's
weird homecoming still in mind. It stayed with me more or less, too,
in the succeeding days.
Naia of Aphur! Oddly I dwelt upon her. Jason himself had told me
that she knew me—had actually seen me—that he had brought her
to earth more than once in the astral body—had pointed me out to
her as the one earth man who knew and believed his story—that she
looked upon me as a friend.
The thing seemed some way to establish a sort of personal bond,
just as the secret Croft and I had kept between us made me feel
toward him as I have never felt toward any other man.
Jason Croft and Naia of Aphur—the interplanetary lovers. It was
certainly odd. I knew her, even though I had never seen her; save
through the instrumentality of his description of her, and the
resultant picture printed on my mind. Yet I could close my eyes at
will and see her, slender, golden-haired, with her lips of flaming
scarlet, and her violet-purple eyes.
And I knew her home. I could lift it into my conscious perception as
a familiar scene. I could imagine her moving about it, young,
vibrant, happy, alone or with Croft by her side. I could fancy her
bathing in the sun-warmed waters of the private bath in the garden
—the gleam of her form against the clear yellow stone of which it
was constructed—until she seemed the little silver fish Croft had
called her, disporting in a bowl of gold, behind the white, screening,
vine-clad walls. Or I could dream of her walking about the grounds,
with the giant Canor—the huge, doglike creature she called Hupor,
who was at once her pet, her companion, and guard. Distant? Why,
she seemed no more distant to me in the days after Croft had gone
back to be with her when her child would be born than some fair
maid of earth waiting for the coming of her lover across a dividing
wall in an adjacent yard.
And yet so blind is the objective mind, that even then I did not
suspect I had established a sympathetic chain of interest between
the atmosphere of her existence and myself, capable of stretching
out to a most peculiar climax in the end. Then, one night something
over a month after No. 27 had died and been laid away, I dreamed.
I don't say I thought of it as a dream at the time. Then it was all too
seriously, too grippingly, real to seem other than the actual thing. It
was only after it was over that I thought of it as a dream—perhaps
because, despite the occurrence and all Croft had told me, I was still
not fully convinced.
Later—well, that's the story. I'll let it unfold itself.
I went to bed that night and fell asleep. How long I slept I do not
know. But a voice disturbed my slumbers after a time. At least it
disturbed the restful unconsciousness of my spirit. To this day I am
not sure whether or not my body moved.
"Murray—Murray." I heard it, dimly at first, but insistent. It kept
repeating itself over and over. Beyond doubt someone was
demanding my attention. I sought to rouse.
"Murray—in the name of Zitu—and Azil—"
I stiffened my attention. It was nothing short of startling to hear
those words spoken.
Zitu was God in the Tamarizian language, as I knew, and Azil was
the Angel of Life—as Ga was the Virgin Mother. Ga and Azil—the
mother and the life-bringer—they were the ones to whom the
Tamarizian women most frequently prayed. I gave over my endeavor
to waken my sleeping body and lay straining the ears of my spirit to
the voice.
It came again. Whoever the speaker was, he seemed to know he
had stirred my conscious perception.
"Murray—I need your advice—your council. Naia needs you. It's life
and death, Murray. You told me you would gladly render her
assistance as a physician. Murray—will you come?"
My spirit staggered. It was most amazing. For now I knew that the
speaker was Jason Croft.
I knew that he was appealing to me in the name of Zitu and Azil—in
the name of motherhood—that he was calling on me as a brother
physician, by the oath of my profession—in the name of all that was
highest and holiest in life.
I knew that Naia's hour was upon her—and I knew it as clearly as if
the thing were taking place somewhere within a neighboring home
on earth. I lay and let the knowledge beat in upon me. I recalled in
a flash all he had told me concerning medical knowledge on Palos. If
some complication in the birth of their child impended, there would
be none on that far planet to whom he could turn for aid. He knew
more than all the physicians of Palos put together, but—
"Murray!" the voice repeated. "Murray, in the name of God!"

There was a desperate urge—a desperate plaint about it. I reached a


decision. I had never married. There was no one dependent upon
me. With a strange thrill I realized the fact. If I failed to return from
this strangest of calls to which a medical man was ever bidden, if the
body of me were not to be revived, I would be little missed.
So what did it matter? A man—or most men—surely could die but
once; and how better than in performing the duty of a physician, in
an endeavor to save other life? I recall now that such thoughts
flitted swiftly through my brain, and left me ready to dare the
venture suggested by Croft's voice, if thereby I might render an
intimate service to him and Naia of Aphur, in spirit if not in the flesh.
"Murray!"
Again the agony of a strong man's appeal for all he held dearest in
existence.
I think the lips of my sleeping material being must have moved at
last. Be that as it may, I know I answered:
"Yes."
And I know Croft sensed my acquiescence, for his response was
beating into my consciousness in a flash.
"Then—fix your mind on our home in the western mountains,
visualize it, Murray, as I have described it to you. Will your conscious
presence within it. I shall be waiting for you. Call up the scene and
demand that our will be granted. Think of nothing else."
Save for the directions for reaching to him, the thing was as real as
a telephone message, and the assurance that the husband of your
patient would be waiting your arrival at his house. But there was
about Croft's promise to await my coming a definite note of
conviction in my ability to encompass our mutual purpose that aided
me most materially in what followed, as I now confess.
He was so seemingly sure that I would not fail them—that what
assistance I could render would be granted—that for the time being
it overthrew all doubt of success. Too, I had grown so accustomed to
thinking of Naia of Aphur as a woman—a palpitant creature of
radiant flesh and blood—that the very reality of her seeming robbed
somewhat of its weirdness, its eery quality, the fact that I was about
to respond in the astral body to an urgent medical call. Consciously
then I sought to follow Croft's directions.
I fastened my thought on his Aphurian home.
I strove to exclude everything else from my mind. I brought up the
picture of it as a thing at the end of a distant vista, down which I
must pass to attain it, and—all at once that picture moved!
I say it moved, because that is how it at first appeared. At all events,
it seemed to come toward me with amazing swiftness.
For an instant my comprehension faltered, and then I knew. I knew
I had gained my purpose—that I was astrally out of my body, even
though I had not known the instant when I had left it; that I was
speeding with incredible rapidity toward the scene into which I had
wished to be projected; that darkness was all about me, like an
impenetrable wall; that I was like one in an infinite, an interminable
tunnel, with the lighted picture I had conjured up at the end.
Then that too faded, dissolved, lost its comprehensive quality, and
gave place to more finite detail, and—I was in a room. But it was not
strange. I knew it—recognized it instantly, thanks to Croft's previous
words.
Its walls were hung with purple hangings shot through with threads
of gold. There was a shallow pool of water in its center edged round
with white and golden tiles. Beside it on a pedestal of wine-red wood
there stood a figure—the form of a man straining upward as if for
flight, with outstretched arms and uplifted wings, translucent—
formed of a substance not unlike alabaster—the shape of Azil.
That too I recognized in a flash, and I seemed to catch my breath.
At last I was on Palos! This was Azil, the Angel of Life, before me—
poised by the mirror pool in the chamber of Naia of Aphur—ablaze
now with the light of many incandescent bulbs in copper sconces
against the walls. All this I saw, and became conscious that, as well
as light, the chamber was now full of life.
Naia of Aphur! She lay before me on a copper-moulded couch—and I
turned my eyes upon her, her body beneath coverings of silklike
fabric.
A woman, of whom two were in attendance, wearing the blue
garment embroidered with a scarlet heart above the left breast—the
badge of the nursing craft, as Jason had told me—spoke to Naia in
soothing accents the words of which I could not understand.
"Murray!"
Whirling, I beheld Jason Croft. Rather, I seemed to see two Jason
Crofts, instead of one. One sat in a chair of the same wine-red wood
of which the pedestal supporting Azil was formed, in the posture of a
man in more than mortal slumber. One floated toward me, ghostlike
—a shimmering, shifting, vaporlike semblance of the other as to
physical shape.
And it was this second Croft that seemed to speak.
I say seemed, because as I recall the episode now I know that
communication was in reality by thought transference, although it
appeared then to reach the understanding in the form of spoken
words. It came over me instantly that Jason had purposely assumed
the astral condition to welcome me on my arrival here.

I had been too much occupied with my surroundings until then to


give thought to my own possible appearance. But as I put out a
hand in answer to his single word of greeting, I found it no more
than a thin diaphanous cloud. I was even as he was—a nebulous
something. Still, that was to be expected. I put it aside and
considered the man before me. The features of his astral presence
were actually haggard, marked by a suffering plainly mental, yet
akin in its way to the lines that contorted Naia of Aphur's face in her
present mortal woe.
"Croft, in God's name what is the trouble?" I asked as once more a
low sound of smothered anguish came from the couch behind me.
Nor do I think I overshot the mark in declaring what followed to
have been the most remarkable medical consultation mortal man
might know. He lost no time in explaining the situation. It wasn't his
way.
He gave me at once an exact and scientific understanding of her
condition, ending his narration simply:
"Murray, you know how I love her. I faced the thing as long as I
could have alone. And then—knowing all that depended on me—I
became unnerved, and called for you. There was no one else—and
you'd said you'd be glad to attend her. Can you blame me, my
friend, now that you see her?"
I shook my head in negation, turning it for an instant toward the
glorious woman shape on the copper bed. "Can she see me? Does
she know I am here? Can I speak with her?" I questioned.
"She will sense your presence at least," Croft said. "I shall revivify
my body and draw the chair in which it is sitting close beside the
couch. You will sit there, Murray, and I shall tell her you are present,
watching, nerving me to my task, before I set to work. She knows I
called you, Murray, and now you must help us both. Your brain must
use my hands to save her. Come—what do you advise me to do,
Murray?"
I told him as soon as he had brought his almost panting response to
an end. His exposition of the problem we faced had made it
dreadfully plain.
He heard me out and then nodded with set lips.
"I—I'll do it, Murray," he said. "I—I felt it was the thing, but—without
counsel—simply on my own judgment, I could not do it. And—you
must coach me. I'll work in a purely subjective condition. That way,
even in the body, I'll be able to sense the guiding impulse of your
brain. God, man, how I need you! Come!"
The form beside me vanished. The body in the chair flung up its
head and rose. It pushed the chair it had occupied quite to the side
of the copper couch, and bent to speak to the woman who lay upon
it.
I followed. I sank into the seat provided. Croft straightened. Naia
turned her head directly toward me.
I looked for the first time into her violet-purple eyes.
They were clear, steadfast, flawless as a perfect amethyst, though
darkened by the ordeal through which she was passing—the eyes of
a true woman, high-spirited, brave, loyal, and pure. They strained
toward me. And suddenly she threw out a perfectly rounded arm, a
slender hand, as one who asks for succor. Her lips parted, and once
more she smiled, a smile so wistfully yearning that my whole heart
answered its appeal.
This was Naia of Aphur—wife of my friend Jason Croft. In that
instant I felt she was worth all that he had dared to win her. This
was Naia, the woman who months ago had told him that in the
silence of the night she had heard the beating of the wings of Azil,
the bringer of new life, because of which I was here now beside her
in that holiest of moments in a medical man's existence, when with
hand and brain he waits to welcome a new life's birth.
Her lips moved. Distinctly I heard her speak:
"Dr. Murray—good friend of my beloved, who tells me of your
presence in response to his appeal for your assistance to us—I bid
you welcome to our home. Thrice welcome are you, upon whose
coming depends, as he tells me also, our future happiness together,
as well as the life of our child."
She addressed me most surprisingly in English, until I bethought me
that Croft had doubtless taught her the tongue, exactly as he had
taught her so much else; to fly the first airplane in Palos, the control
of the astral body itself. Her words moved me oddly. I rose to
answer:
"I am more than happy to be here, Princess Naia, and to bid you be
of good cheer, remembering that even now Azil stands close by the
gateway of life, in charge of a newborn soul."
And then I sank back, confused. I had spoken wholly on impulse,
voicing the inmost emotions of my heart, forgetting my nebulous
condition entirely for the instant, in the spell of what seemed so real.
With a feeling akin to acute annoyance at my inability to speak thus
to her directly I resumed my chair.
But even so, it seemed that I had reached her—that in some way
akin to that in which Croft had assured me he would be able to
follow my mental direction while working, she had sensed my
meaning and intention. Women are intuitive by nature, more
susceptible to the waves of a personal or thought vibration. Her lips
moved again as I ceased speaking.
"Azil," she whispered. "But—that new soul is so long in passing, my
friend."
I turned to Croft.
"Come," I hurled my thought force toward him. "Let us spare her
more bodily anguish than must be endured. Let us make an end."
Of what followed I shall say no word. Suffice it to state that Jason
Croft labored, grim of lips and pallid of feature; that I sat in that
weirdest position of assistance capable of conception; that the lights
burned on in that room where the pale form of Azil spread his wings
on the pedestal of wine-red wood; that the eyes of Naia of Aphur
widened until they were two dark pools no more than fringed by the
purple iris; that the two female attendants waited, intent on naught
save the catching, the rendering of obedience to each of Croft's
intense though low-pitched words.
And then suddenly the man turned to me a face transfigured past
anything I had ever pictured—a thread of sound—a wailing, trailing
vibration—the first note of waking vocal strings, pulsed through the
room—and Jason Croft the physician, the father, was kneeling beside
that couch of copper, no longer the iron-nerved worker, the laborer
for unborn life, but the husband, the lover, clasping the slender body
of Naia of Aphur in his arms, and shaken by a strong man's sobs. I
turned away my eyes.
And then his voice boomed out, strangely exalted and triumphant:
"Murray—we win—win, man—thanks to you and—God!"
I turned back. Croft spoke to one of the attendants. She crossed to a
curtained doorway and lifted the purple drapings. There stole into
the room a girl of Mazzeria—a graceful creature, for all the odd blue
color of her skin. Twin braids of ruddy hair fell from her head to her
waist. Her figure held all the untrammeled litheness of a panther as
she advanced. Across her outstretched arms she bore a pure white
cloth.
Upon it, the child of Jason Croft and Naia of Aphur was placed.
She wrapped the fabric about it, cradling it against her breast. She
turned to Naia, smiling, sinking down beside her on her supple
rounded thighs.
And then—for one brief instant I saw the light of the Madonna flame
in those wonderful eyes—the light with which Naia the mother
looked first on Jason's—son.
Croft addressed me.
"Maia," he said softly. "I've described her to you before if you
remember, Murray. She asked that you might be permitted to attend
the—the little one."
His voice broke. His face was weary, overstrained, worn. I
understood. The graceful girl was Naia's personal attendant—the
Mazzerian woman, who had aided her mistress in saving Croft's life
at a time when he was taken captive during the Mazzerian war. I
nodded my comprehension. He bent again as though by irresistible
attraction above the couch where the blue girl still was kneeling, and
Naia seemed waiting his undivided attention. Once more I turned my
head. It was the holy moment—the hour of realization between man
and woman.
Through the half-drawn curtains of a window, light stole into the
room. It shamed the incandescents in their sconces. A finger of
golden glory touched the tips of the upflung wings of Azil. With a
start, I realized that the night of anguish was ended—that new life
had come into the house of Jason—with the dawn.

You might also like