The COVID-19 pandemic has accelerated the shift towards e-commerce and digital transformation in retail, with sales forecasting and revenue prediction becoming increasingly important to retailers as they try to stay ahead of evolving consumer behaviors. Retailers need to consider location aspects when analyzing sales data, especially when it comes to decisions around site selection. To address this challenge, CARTO's Analytics Toolbox for BigQuery can be used to train a spatial model that predicts annual store revenues across a territory, enabling retailers to optimize their revenue by avoiding over or under-stocking key product lines. The toolbox provides a complete framework of analysis capabilities to perform spatial analytics in SQL, and its recent Spatial Extension for BigQuery enables users to have a fully cloud-native GIS and Spatial Data Science experience while keeping the benefits around privacy, compliance, scalability, and lower costs. By leveraging spatial indexes and advanced configuration options, retailers can build predictive models that take into account location-aware context, such as population density and competitor stores, to predict revenue potential for different locations, ultimately informing optimal site selection strategies.