
Consumer Goods & Retail - Prediction and Allocation Optimization for stationary Channels
Consumer Goods & Retail
A model for predicting and optimizing product allocation in stationary channels led to significant annual revenue growth by reducing miscalculations and returns.
Initial situation
A large retailer faced an unsatisfactory rate of miscalculations in the stationary sector. There was a significantly high return rate, while products at other locations were sold out much earlier than planned, leaving demand unmet.
Architecture
The architecture is built on Google Cloud, using technologies such as Google Dataflow, Google Vertex AI, Google Cloud Run, and OpenAI.
Generated benefits
By optimizing the product allocation, miscalculations and returns are reduced, leading to significant annual revenue growth.
Services accomplished by synvert
synvert developed a model for the drivers of demand. Various features were developed, and demand was mathematically modeled as a target value based on sales. For this purpose, a feature store with various logics was set up. A parallelization tool was used to efficiently handle the underlying big data challenges when generating a training set. For inference, synvert developed a fully automated, cloud-native software solution that integrated into the target systems.