Using data science to predict subscriber churn and boost customer lifetime value

This client is one of Europe’s leading media and entertainment companies, offering satellite TV, broadband, mobile and streaming services to millions of customers across the UK and beyond. Known for premium content and customer experience, they continuously hone their competitive edge with more personalized, more engaging customer interactions.

This client set out to reinforce their market leadership with a bold ambition: to “own the living room.”

Competing with direct-to-consumer giants like Netflix and Disney+ brought both urgency and opportunity. The client's senior leadership embraced the challenge by backing innovation, enabling agile collaboration across teams and driving experimentation from the top down.

Identifying the right use cases and proving impact required both strategic alignment and technical validation. Valtech stepped in to help the client prove the business value of their ambitions and bring them to life. 

Valtech helped implement a predictive model that identifies customers most at risk of churning and tailors discount offers accordingly.

The model balances personalization with profitability, adjusting the depth of offers to retain high-risk customers without overspending on those likely to stay. The solution has delivered long-term business impact and continues to evolve in partnership with the client's internal teams.

We helped deploy a machine learning–powered recommender that personalizes in-broadcast content promotions on set-top boxes.

The goal is to increase incremental viewing and, ultimately, customer lifetime value by showing the right promotion to the right viewer at the right moment.

The system segments customers into micro-audiences and delivers ranked promotional content to boost engagement. The result was a 113% uplift in targeted groups, with the model now optimized for incremental revenue and integrated into the client's core broadcast infrastructure. 

The client worked with us to improve how they handled at-risk customers during online cancellation flows.

They had an existing propensity model for cancellation journeys, but we saw room for improvement. We recommend what we call a Large Action Model, which uses the same transformer architecture as large language models.

Instead of relying on static logic, the Large Action Model anticipates what a customer is likely to do next. This allows it to determine the likelihood of a customer churning in specific scenarios. When that likeliness meets a specified threshold, the model can serve that customer tailored retention offers.

This Large Action Model outperformed the existing propensity model by 20%, delivering real-time decision support during high-risk moments.

• Personalized retention modeling delivers measurable, ongoing cost savings
• The recommender system increases content engagement and revenue uplift
• Action-based modeling has improved retention outcomes during digital cancellation journeys
• AI solutions moved beyond POC into enterprise-scale production environments 

Valtech has been a partner to this client for nearly a decade. Because we have spent years delivering real results, the client knew they could trust us to deliver scalable, tested AI systems. Valtech brought the technical depth, the retail media experience and the proven delivery chops.

The Large Action Model is still active and learning, with ongoing improvements as new data flows in. We are looking to expand personalization to more channels and tap into agentic AI to simplify customer support and sharpen product recommendations. Valtech is continuing to find smart, practical ways to put intelligence to work across the client’s entire business.