Predictive spend

Our client is a global leader in hospitality managing a portfolio of 40 brands, in luxury, mid-range and economic segments.

Within 5000 hotels in 100 countries, 2/3 are franchised. Hence, the group do not have visibility on hotel spend by supplier and category.

The objective is to accurately predict spend to identify hotels & categories to be tackled in priority by both sales & procurement teams to increase capture rate for the central buying office and support hotels in their most impactful costs.


  • Develop a machine learning model to predict spend by hotel and category by category
  • Assess the contribution of additional data to the overall model accuracy, such as : historical data from similar hotel, brand, location, hotel characteristics, restaurants, room and F&B revenue, number for rooms, date since last refurbishment, etc.
  • Work with procurement team (buyers, controllers and developers) on how to embedded this model into procurement processes


  • OPEX spend predicted with 85% accuracy at hotel/family/quarter level
  • Prototype a smart agent app to provide access to relevant data and predicted spend while the team is visiting hotels.