The Spotlight knowledge-based data explanation system, which helps users grapple with the vast amounts of marketing data generated by supermarket scanners, is described. The system extracts information from databases, providing formatted reports that highlight and explain changes in a product's market share caused by promotional programs, shift among product segments (such as sales shifting from caffeinated to decaffeinated coffee), and changes in distribution and price. Spotlight also reports on competing products and product segments. Looking for common and unique behavior patterns. Spotlight embodies an explicit model of explanatory options and preferences, distinguishes pattern observation from expansion creation, and clearly separates both from its modular presentation capabilities.< >