Managing Telecom Churn with Proactive Retention Programs
Fractal helps leading telecom provider control churn with predictive churn models
Large integrated telecom services provider (this client has requested anonymity).
Control customer churn.
Proactive retention program backed by Fractal Analytics’ predictive churn models.
Better able to control churn and enhance profitability with high-value customer retention programs.
The Business Challenge
The telecom industry is fiercely competitive. In the cellular space, consumers have a vast array of choices, resulting in acquisition costs and customer churn rates that have increased at a rapid pace. Our client was experiencing high customer churn rates that were cutting into its profitability. A proactive approach to customer retention was needed to ensure retention of high-value customers.
While it is well known that retaining customers is far less expensive than acquiring new customers, in practice many retention programs deliver less than acceptable returns due to imprecise targeting. Frequently, a large proportion of customers targeted with retention programs are often ones who would not have churned in the first place.
Our challenge therefore was to build churn models that would help our client identify high-value customers with a high propensity to churn.
The first task our analysts faced was to generate a single file that would capture all aspects of a customer’s interaction with our client. A key requirement for identifying customers at risk of churn is to have a single consistent view of the individual customers, as opposed to segregated records scattered throughout different business silos. Our analysts needed to integrate billing information, payments information, customer demographics and service record information into a single customer-level file.
Before we could begin there were additional challenges which had to be factored into the data aggregation process. Our client was in the midst of reassigning their telephone numbering scheme which increased the complexity of collating the billing and usage data. Further, they were in the process of migrating to a new data warehouse, which meant the data model provided by us for the implementing the churn model needed to be scalable to fit the new requirements.
Utilizing our advanced data management tools ensured timely and accurate data file generation from an extremely complex, dynamic and large transaction dataset.
Our analytic scientists then applied advanced modeling techniques including neural networks, decision trees & logistic regression to construct a model capable of scoring each customer based on their probability of churning in the next two months. Our proprietary visualization tools provided quick insights into which factors were most predictive of churn. We also used our non-linear pattern detection tools to identify interaction variables that are typically overlooked by linear techniques.
By integrating various techniques, we developed a model that would more accurately score customers according to their propensity to churn. The model also had to meet very stringent stability criteria, due to recent regulatory changes. Our final model was optimized to continue to be effective in the changed telecom environment.
Combining the predictive churn model with other CRM metrics such as customer profitability, we helped our client generate more effective customer retention strategies based on customer value. For example, customers with a high propensity to churn and high profitability were given the highest priority for retention. These customers would be contacted first and offered the best incentives to maintain their phone subscriptions. Customers who displayed very low profitability and yet had a high propensity to churn were encouraged to increase their usage.
The intelligent application of predictive models helped our client control churn and enhance the returns on its retention programs.
Source: How can I reduce churn?