Our Work

Machine learning automation preventing customer churn

Customer Retention | CRM | Loyalty Management | Machine Learning

Our client, one of the biggest Brands in the travel industry, wanted to eliminate customer churn and boost engagement.


Keeping churn at single digits.

Over the last few years the company’s program had significantly grown, making it difficult to understand customer behavior and build a coherent customer retention strategy, keeping churn at single digits.   

The organisation had run various analyses on how big churn was and how it varied between customers with distinct characteristics such as tenure, balance, gender and product preferences but hadn’t done in depth analysis of what drives churn and how it could be prevented.


A predictive model that could be applied to any current or future customer dataset.


We created a predictive model that could be applied to any current or future customer dataset and predict the probability to churn in the coming months for each individual customer. The technique applied was a Random Forest statistical learning algorithm; a machine learning technique that considers a series of customer characteristics such as when a customer last transacted, frequency of transactions, customer value and so on.

The model was trained using existing customer transactional data and before it was deemed fit for predicting future churn, it was tested and validated against the recorded historical behaviour of customers that were randomly excluded when training the model.

The resulting model managed to achieve close to 90% accuracy in predicting high risk customers, and has since been put in production. It is continuously used to proactively signal churn and trigger the appropriate marketing / promotional activities to prevent it from happening.

Sounds interesting?

Get in touch now and we can schedule a free, one hour, consultation. No strings attached.