Lorem ipsum dolor sit . Proin gravida nibh vel vealiquete sollicitudin, lorem quis bibendum auctonisilin sequat. Nam nec tellus a odio tincidunt auctor ornare.

Stay Connected & Follow us

What are you looking for?

Simply enter your keyword and we will help you find what you need.
PBT GroupPredictive Churn model to support long range detection of customers at risk
Predictive Churn model to support long range detection of customers at risk​

The Company

A multinational mobile telecommunication company, operating in African and Middle East countries, with in excess of 200 million subscribers across its operations.

The Business Challenge

The company needs to be more pro-active in its dealing with potential churners. At the moment it can only be re-active and contact subscribers when they have already reached the first stage of dormancy.

The Solution

As the churn model is built on a wide range of historical data, it is robust to data noise interference and stable over time. The accuracy of its prediction as well as the health of the model are monitored at all times.

The churn model predictive scores are loaded into a visualisation dashboard in a way that allows the tracking of the global potential revenue loss should the warning provided by the predictive churn scores not be fully leveraged through effective marketing campaigns.


The Value Proposition

The Predictive Churn model allows for the detection of subscribers displaying a high churn risk and a likelihood to disengage with the network within a prediction window of up to 4 months warning. Business has then got enough time to approach the vulnerable subscribers with value propositions that will entice them to stay with the network.


Two versions of the churn solutions exist – one on the company’s environment using SPSS, SAS, SQL and Oracle OBIEE and the other on Google Cloud platform using R , PostgreSQL and Power BI.

Related Cases

No posts were found.