|
 |  |
 |
 Overview Success Stories
|  |
Predicting Churn Given the tough competition in the Dutch mobile telecom sector, one of the fastest growing providers turned to Chordiant Decision Management. The ChallengeAfter years of recruiting, the mobile telecom market in The Netherlands had become sated and the competing providers realized they had to change their strategy. Not only did their current recruitment strategies encourage people to churn (and sign up again), but competitive offers lured hard-won customers away. Our client decided to start managing churn pro-actively, for which they needed to develop a predictive model that would identify likely churners in a stage as early as three months in advance. The Solution Because the customers that are still under contract are not allowed to leave, the model would immediately spot that the lion's share of the churners is amongst the customers that are either near the end of their contract or even contract-free. Therefore, the population was divided between this latter group and the customers under contract. A scoring model was developed to identify those people most likely to churn and (almost) free to do so. How It Was Done 
From the telecom provider's customer database a
sample was taken containing some 550 variables
including personal details (age, gender, postal
code, etc.), contract details (activation date, days
free to go, length of contract, etc.) and usage
details (number of calls, sms, incoming, outgoing,
etc., for varying periods). Six months of historical
data and three blind months were used to create
a model that would predict churn three months
in advance.
The OutcomeFigure 1 shows the selection percentage against the percentage of churners identified by the model. At a selection of 50% of the customers near the end of contract or contract-free the Chordiant model can identify more than 75% of the churners three months in advance. This allows our client to spend their retention budget pro-actively and achieve better focus on the customers actually at risk of leaving. Furthermore, the model helps to reduce the risk of 'waking up' loyal customers. |  | |
 |
| © 2000-2008, Chordiant Software, Inc. All rights reserved. Legal/Disclaimer | Privacy Policy | Site Map | |  |
| Solutions | Retail Banking | Card Services | Lending | Insurance | Telecommunications | News & Events | About Chordiant | |
