Uses Predictive Analytics to Anticipate and Reduce Churn

The custom application produced with Data Science Studio predicts customer churn with 77 percent accuracy

ParisJan. 4, 2016Data Science Studio (DSS), developed by Dataiku, was adopted by, an e-commerce website in Europe, to develop and analyze its internal data to prevent churn and maximize the customer retention rates of its nearly 20 million members across Europe.

Anticipating and Reducing Attrition Rate (Churn)

In an ultra-competitive environment, optimizing retention rates is a major issue for commercial sites. soon realized the limitations of a generic marketing strategy to address its retention rate based on static rules common to all customers. To improve customer loyalty, it is fundamental to disseminate the right message to the right customer. Simple in appearance, this solution presupposes the establishment of preliminary identification rules for potential churners (customers who may not renew their purchases) and determines the value of each individual client. For, the challenge of customer loyalty rests largely on the value of this data, enabling it to support its customers in a more personalized manner.

Dataiku technology allows to detect customers who no longer make purchases on the site, depending on the frequency of individual purchases, and refine the precision targeting of marketing campaigns.

Importing Data into a Predictive Model

With Data Science Studio, established an effective tool that is complete and easy to use, allowing it to internalize the development of a predictive analytics application. The solution, developed with DSS, automates the detection of customers who have a high probability of no longer purchasing products on the site. With Dataiku technology, the marketing and business intelligence teams at mastered their anticipatory attrition project in full. uses DSS for:

  • Automating the integration and enhancement of a wide variety of data sources (customer data, order data and delivery, web logs, etc.).
  • Creating more than 690 features derived from this data and according to specific variables (clicks on sales, orders, litigation, customer, etc.).
  • Testing several machine-learning algorithms to achieve the best predictive model.

Damien Garzilli, the strategy and business intelligence manager at, commented: "From the data import to the development of a predictive algorithm, DSS’s ease of use allowed us to gain autonomy throughout the entire process. Today, we are able to predict the future actions of our customers and act accordingly. It is also a tool that will allow us to speed up the production of other big-data use cases as our needs emerge."

A Successful Collaboration and New Perspectives

Since the commissioning of the application developed with DSS, can now detect potential churners with 77 percent accuracy, which, in turn, initiates targeted marketing actions.

The success of this first project on DSS allowed the emergence of many other innovative data-driven initiatives with DSS and opened new perspectives for

Garzilli stated, "The objective of churn prevention is to send the right message to the right person. We are more than satisfied with the results we get through DSS. Today, Dataiku allows us to talk about the data … the past to project ourselves into the future."

Companies in this article