Best Practice

Big data and real time marketing

Make it personal.
Limit churn and enhance customers satisfaction through real-time personalization.


Today’s business context is more and more fragmented and competitive: it becomes increasingly important for companies to retain their own customers. Since recovered long-term customers can be worth much more than newly recruited clients, being able to anticipate customers’ abandon in order to retain them on time, reducing costs and risks, is a must have. Therefore, in many sectors, from fintech to telecommunication, from retail to e-commerce, one of the main challenge is reducing customers’ attrition.


Reply is able to recognize clients’ actions and discern between satisfaction and dissatisfaction factors from analysis and exploration of an enormous amount of data identifying customers’ tastes and behaviors inside organizations, both historical and real-time data . By means of a big data platform and machine learning ensemble algorithms, it is possible to predict customers’ loss and monitor their thoughts and actions about purchased services/products.

These models enable marketing departments and agencies to:

  • Find as soon as possible which customers are about to abandon and know them in depth;

  • Put into practice personalized actions in order to reduce or avoid their migration, improving their journey;

  • Increase the capability to react and anticipate possible desertions.

It is possible to analyze and classify customers’ actions to detect main attrition reasons in order to prescribe personalized real-time marketing campaigns. Customers’ behavior is used to build alerting systems, driving fast and appropriate company reactions.

A churn framework

Reply designed and implemented an hybrid customer management and churn framework able to:

Analyse and combine diversified multiple data sources in order to generate a 360 customer view, integrating historical batch data and real-time streaming informations to build a dynamic golden record

Extract precious informations in terms of topics discussed by customers, customer habits and service usage to personalize their journey

Identify churn causes and intercept dissatisfaction reasons in order to prescribe retention/prevention marketing actions tailored on one-to-one customer behaviour

Significantly decrease the churn prediction forecast horizon, aiming at timely intercepting customers desertion

Set up an alerting system based on customers’ real-time actions in order to address proper marketing actions

Implementation and design process