Churn Prevention
Don’t loose your customers: stop churn before it happens!
USE CASE
Your customers could leave you, unless you learn to understand when a give customer is developing a tendency to dissatisfaction.
When you get aware of this dangerous tendency you can take specific actions to invert it.
This is churn prevention.
Basically you want to:
- understand on time when there’s a growing probability that a given customer will leave you
- understand which actions can invert the process
SOLUTION
Our solution for churn prevention models is based both on classical descriptive analysis and overall on artificial intelligence. The former is used to explore the statistical nature of the environment and the latter to design an intelligent algorithm with predictive capabilities.
Our exclusive and performant methodology for churn prevention modelling, is able to work both on cases where the churning event is well defined and on cases where the churning event does not require a defined action of the exiting customer. A well defined action is for example a letter, or an official call, or action on a web form. On the other hand, sometimes the customer is not asked to do anything, he is simply stopping using our services. Our exclusive approach can manage also this latter situation, which is otherwise quite difficult to manage.
From a machine learning point of view, we can learn both from a labeled training set and from an unlabelled training set. In both cases, artificial intelligence will then provide a classification of the whole customer base assigning a risk level to every customer. Also, the algorithm will highlight the main cause for leaving, thus suggesting which actions have the best chance to save the customer.
DEPLOYMENT
Our churn prevention models can run on normal and big data. We have more than 100 certified connector to the main data bases and file types, thus we can collect data from every kind of source.
The model can run both in the cloud or on premise. Our product has a client/server architecture, made up by a computing server and an interactive web graphical interface. Also, the system can create periodic automatic reports and trigger alerts by email.
