In a retail reference architecture, a solid central inventory system is crucial for improving and enriching the customer experience. However, legacy inventory systems commonly used by major retailers present real challenges due to their inflexible data structures and high transaction volumes. MongoDB can help address these challenges with its flexible schema, horizontal scalability, and high performance capabilities. A key aspect of this approach is creating a unified view of inventory that is accessible across stores and applications in real-time. This involves designing a schema for the stores collection to associate locations with inventory and creating an inventory collection to track actual inventory counts of each item and its variants. The use of indices optimizes queries and reduces data duplication, ensuring a high degree of accuracy and availability. By leveraging MongoDB's geoNear command, retailers can query stores by proximity to a location, enabling features like in-store pick-up at a particular location or detecting the need for intra-day replenishment if there is a run on an item.