Inside Target’s LLM-Based System for Semantic Matching in Marketing Forecast Pipelines

Target has replaced its rule-based marketing forecasting workflows with a generative AI system that utilizes vector search and large language models to identify relevant historical campaigns. By automating the retrieval and ranking process, the retailer reports that the new system achieves a 75 percent top-1 coverage rate while significantly reducing the manual workload previously required to manage campaign predictions.