Ireckonu research reveals AI can pinpoint moment customer losses in hospitality

AI can identify and intervene

New research led by Ireckonu's head of data solutions and customer success has uncovered a breakthrough in hospitality churn management.

Dr. Rik van Leeuwen's analysis found the exact moment a guest is about to disengage and stop them from leaving. 

This finding comes from a multi-week study conducted in collaboration with a major North American hotel chain, and it marks a major step forward in data-driven guest retention.

The study concludes that when a guest reaches a 75% churn risk, sending a 20% discount email can significantly increase the likelihood of rebooking. The result: AI not only identifies guests at risk of leaving, but it also optimizes when and how to intervene—maximizing the return on retention efforts.

This new framework, which combines the BG/NBD model for churn probability with reinforcement learning for proactive engagement, is tailored specifically to the hospitality sector. 

The BG/NBD model predicts customer behaviour by estimating how likely they are to make repeat purchases over time in non-subscription settings.

Dr. van Leeuwen’s approach emphasises interpretability and adaptability, making it both transparent and actionable for hotel operations. 

His PhD research, titled Data-Driven Strategies in Hospitality, explores how predictive and prescriptive AI models can be adapted to real business contexts, prioritizing transparency and applicability. 

He hopes to educate the industry on the importance of white-box approaches in building trust and usability within hotel operations.

“This is not just about predicting customer behavior—it’s about turning that prediction into timely, effective action,” he said.

"Knowing who is at risk is no longer enough. The real value lies in knowing when to act and how to respond. That’s where AI can truly transform hospitality strategy”. 

He completed his PhD while working at Ireckonu, with the company’s full support which is something Ireckonu’s strongly commits to with applied research in hospitality technology.

The implications of the study go beyond a single use case. Ireckonu is actively working to integrate these insights into its broader middleware and customer data platform offerings, helping hotel groups operationalize AI models that are proven to work in real-world settings.

“Rik’s research brings scientific validation to something hotels have long struggled with: guest loyalty,” said Jan Jaap van Roon, CEO of Ireckonu. 

“This isn’t theory - it’s tested, actionable insight. And it’s a perfect example of the kind of innovation we champion at Ireckonu.”

The study also opens the door to future enhancements, such as adjusting dynamic pricing levels based on individual churn risk, incorporating qualitative feedback like sentiment analysis, and expanding the model to industries where customer relationships rely on frequent, non-contractual interactions.