Today’s dynamic and complex marketplaces have rendered traditional forecasting models less reliable. Established models are built primarily on syndicated data and cannot account for real-time consumer perceptions, needs, and behavior. The forces of digital transformation further complicate the situation by giving customers unprecedented levels of information and market access.
To optimize media campaigns for nutrient-based products/juices, PepsiCo decided to try something new.
See how the company implemented a unique solution to predict Influenza like-illness in a cluster of regions in the US via internet search data.
From the discovery of frequently used symptom and treatment keywords to building machine learning models to predict Influenza-like illness counts, search behavior provided weekly updates at a region level. The model predicted influenza outbreaks with 98% accuracy when compared to Center of Disease Control (CDC) data.