Definitions

Churn
Term used to describe customer attrition or loss

Churn rate
The number of participants who discontinue their use
of a service divided by the average number of total
participants during a period
churn rate is calculated by dividing the number of
customer cancellations within a time period by the
number of active customers at the start of that period.
Customer recovery-Churn Analysis
 Your customer churn rate has a direct impact
on your customer lifetime value and the ability
to grow your business. If your customer churn
rate is higher than 10%, then even if you bring
on 10% new business, you’re not going to be
able to grow.
 Measuring your customer retention rate can
help… so can increasing customer satisfaction
and building customer relationships, but what
can you really do to improve customer churn?
 Your churn rate can be represented in a
number of ways, including:
 Number of customers lost
 Percent of customers lost
 Value of recurring business lost
 Percent of recurring value lost
How to calculate your customer churn rate?

 Let’s say you are an IT Support company. You

have 100 clients. In the month of November,
you have eight new clients sign up for your IT
Services and have three ended their contracts.
 The best churn rate formula to use to calculate
customer churn in take the number of clients
lost, divided by the number of clients at the
start of the month.
 3 clients lost / 100 starting clients = 3% churn
rate
 How to reduce customer churn, boost
customer lifetime value and grow your
business.
 Set and meet customer expectations:
 Customers love it when you exceed their expectations.
It builds strong customer loyalty and goes a long way
to gaining their trust. This also contributes to
increased customer satisfaction and reduced churn.
 Nail the first impression:
 By creating a value-adding experience as a first
impression, you can help educate your customer on
the true value of your offering, which will aid in
creating a more ‘sticky’ relationship that is less likely
to churn.
 Always be adding value
 Improve competitive advantages
 Build ‘sticky’ customer loyalty
 Focus on becoming a dependency
 Listen to your customers
 Keep your eyes open for external
environment changes
 Track customer lifetime value of marketing
channels
 Reasons for churn

Easy to switch provider

Difficult to manage the customer data
Inadequate services
Quality of service
Plenty of attractive offers
Customer dissatisfaction
 Types of churn

Involuntary
Voluntary
 Voluntary churn

 Voluntary churn is characterized by customers actively

choosing to cancel their service. This can happen for any
number of reasons, including:

 Switching to a competitor
 Closing down a business venture
 Negative customer experiences
 This type of churn is preventable. A customer who’s thinking
about leaving you for a competitor might be convinced to
stay if they’re reminded of the value your service provides.
However, it is tricky to dissuade customers from voluntary
churn since the reasons are often complicated to solve.
 Involuntary churn
 Involuntary churn occurs when a customer’s account is canceled
when they didn’t intend it to be. Some examples of involuntary
churn include:

 Expired credit cards

 Reaching the limit of available funds
 Failed mechanical payment processing
 Fraud protection on recurring payments
 We’ve found that 20% to 40% of total churn comes from
involuntary churn.
 Involuntary churn is more easily prevented than voluntary churn
because it is almost always the fault of a mechanical failure in
your internal processes. In most instances, a proper system of
dunning emails and notifications can take care of delinquencies at
the company level.
 churn model 

 It is no secret that customer retention is a top priority

for many companies;
acquiring new customers can be several times more
expensive than retaining existing ones. Furthermore,
gaining an understanding of the reasons customers
churn and estimating the risk associated with
individual customers are both powerful components of
designing a data-driven retention strategy
 A churn model can be the tool that brings these
elements together and provides insights and outputs
that drive decision making across an organization.
Find the data you need to build a churn model

 Building a predictive churn model for your

business starts with categorizing everything
you know about your customers. Most
businesses are already tracking this data, so
you just have to know how to use it. Every
customer data point you have helps build a
more targeted churn model.
 Customer information

 The first step is building comprehensive

customer profiles. At their core, these profiles
should include the customer’s name and address
but can be expanded to include job title,
employment status, team size, and much more.
 With this data, you can easily spot patterns in
churned customers related to their
demographics and segment them into cohorts
for more granular analysis. Different customer
types will churn in significantly different ways
 Purchase Information

 Expand on your customer profiles by including

information about their purchase and billing
history. Knowing when a customer signed up,
when they canceled your service, their
payment history, and overall 
lifetime value (LTV) helps you build a clear
picture of how billing processes impact your
churn.
 Interaction Information
 One of the biggest contributors to voluntary churn is
the customer experience. Make sure you’re tracking
every interaction a customer has with your team as
well as your product. Including this information in
your customer profiles helps you see the impact of
your product and the customer experience on churn
rates.
 Tracking past interactions can also be valuable for
surfacing points along the customer journey where
churn is more likely to occur.
 Customer profiles are the basis for more in-depth
churn analysis. With this data, you can start looking
for patterns in how and why different types of
customers leave your service.
Look for patterns in how customers churn

 Comprehensive customer profiles help you see

what types of customers are canceling their
accounts. Now it’s time to figure out how and
why they’re churning. Ask yourself the
following questions to learn more about the 
pain points in your product and customer
experience that lead to a customer deciding to
churn.
 What is a churn model?

 A churn model is a mathematical

representation of how churn impacts your
business. Churn calculations are built on
existing data (the number of customers who
left your service during a given time period). A
predictive churn model extrapolates on this
data to show future potential churn rates.
 How to build a churn model manually

 With your customer profiles created and

analyzed, it’s time to talk through how to
create an actual churn model. As this is a
mathematical process, you’ll need a strong
understanding of statistical concepts and data
science to move forward. You’ll also need a
significant amount of data to perform these
calculations.
 Gather and review your data

 ou’ve spent all this time building up a data set

—every bit of customer information you have
is a valuable data point in the upcoming churn
calculations. Make sure you review all of your
data for accuracy and validity before moving
on to the math.
 Set up a regression formula

 Mathematical modeling for churn is built on a

statistical process called logistic regression.
This process determines the relationships
between points in your data set based on a
formula and limits the outcome to between 0
and 1. You’ll take all the customer information,
purchase history, SaaS metrics, and prior churn
data and turn it into a statistical prediction of
when certain types of customers might churn
in the future.
 Come up with a retention plan

 once you’ve modeled churn through the logistic

regression formula, you’ll be able to more
clearly analyze retention and see the probability
of certain customer segments churning.
 To help maximize retention, use this information
to formulate a plan, based on these findings, that
targets each of your cohorts directly. The
probability of certain customers churning your
service earlier than others will make it easy to
prioritize your actions.
 Implement and track your results

 With a plan in place, it’s time to implement

your retention strategy. As you do so, keep
track of how it impacts your churn rate over
the next few months. Gather enough data to
see the real impact of your efforts before
making additional changes to your plan. This
data might look something like this:
 Test retention strategies

 Your churn model will provide probabilities for

a number of different cohorts in your customer
base. Make sure you’re always testing out new
strategies and recording the impact on these
customer segments. Each subsequent test can
help you create a better model for the future.
 There’s a lot more that goes into creating a
mathematical model of your churn than you
think. Analyzing this information takes time and
resources from across your team.
Key Steps Taken To Identify Parameters of Churn

 Exploratory analysis on satisfaction data - Installation,

Fault, Communication etc. to determine the impact of
different components of satisfaction data on the overall
churn.

 Built a predictive churn model leveraging SLA data

(billing, delivery, and assurance), customer satisfaction
data and complaints data to identify key drivers affecting
churn

 Separate models using techniques such as

regression analysis were built for each business units and
each service line, all rolling up to one overall churn model
 Data Transformation

Data identification
Data collection
Validation & Cleaning
Transpose to right granularity
Addition of derived variables
Preparation of Model Set
Conduct Modelling

SunTel Mobile Telecommunications Slide 6/19

 Propensity to Churn

Changes in spending or payment patterns

Recurrent or multiple problems or queries

SunTel Mobile Telecommunications Slide 7/19

 Customer profitability
calculation

completed calls
call forwarding
failed calls
customer care calls
voice calls
directory info
data calls

SunTel Mobile Telecommunications Slide 8/19

 Churn reduction

Identify customer’s value

Plan a profitable marketing strategy
Point out clients most likely to defect
Develop win-back policy for worthwhile
customers

SunTel Mobile Telecommunications Slide 9/19

 Data Set & Variables (1/2)

Race / Origin
Current Age
Occupation
Geographic Locale or Major City
Number of households at address
Length of service
Lifetime average minutes usage

SunTel Mobile Telecommunications Slide 10/19

 Data Set & Variables (2/2)

Average monthly calls (lifetime)

Average monthly minutes (completed voice)
Recent change in monthly minutes
Days since last retention call
Range of monthly recurring charges
Handset price

SunTel Mobile Telecommunications Slide 11/19

 Decision Trees or Neural
Networks?
Decision Trees
Decision trees
customer classification
easier to use

Neural Networks
Neural networks
hidden pattern
recognition
less costly
more advanced

SunTel Mobile Telecommunications Slide 12/19

 Premium customers

Characteristics
Long duration / high cost calls
Demand high quality of service
Churn due to network incapability / high cost
Strategy
Free talking time
Gift mobile devices

SunTel Mobile Telecommunications Slide 13/19

 Classic customers
Characteristics
Excessive SMS usage
Short duration of calls
Unstable user group
Churn due to competitors’ better offers
Strategy
Special prepaid cards
SMS offers
Discounts in new handsets

SunTel Mobile Telecommunications Slide 14/19

 Strategy

Keep both groups

Classic customers represent clients’ majority
Premium customers have
high customer value

Design different market models

for each group

SunTel Mobile Telecommunications Slide 15/19

 Past, present, future

SunTel Mobile Telecommunications Slide 16/19

 Latency effect

Need for immediate action

Undesired rise in churn rates
More dissatisfied customers

SunTel Mobile Telecommunications Slide 17/19