Title: Large-Scale Forecasting for a CPG Manufacturer: A Practitioner's Insight
Date and time: 01 July 2025 (Tuesday), 10 - 11 a.m.
Venue: IEOR Seminar Room
Speaker: Dr. Anu Thomas, Unilever
Abstract: Forecasting the volume to ship to a customer (Sellin) is more challenging compared to unconstrained (customer to end consumer) sellout forecasting. Sellin forecasts are influenced by numerous internal and external factors such as production/supply constraints, life cycle management, inventory policies, special packaging/bundling, trade negotiations, and market competition. Typically, we observe that sellin forecast accuracy is at least 10% lower than sellout accuracy. At Unilever, sellin forecasts are generated for over 3 million product, customer, and channel combinations every week using an in-house solution. Continuous innovation is enhancing the solution to leverage customer and competition data, improving sellin forecast accuracy and responsiveness to sales disruptions. This integrated forecast is utilized by various stakeholders in supply chain, customer development, and promotional planning to ensure smooth operations. In addition to meeting KPI targets, we also prioritize cycle-on-cycle stability, computational efficiency, explainability and forecast diagnostics.
Bio: Dr. Anu Thomas is a seasoned machine learning and mathematical modeling practitioner with over a decade of experience in consulting across diverse domains. He holds a Master's degree from IIT Madras and a PhD jointly awarded by IIT Bombay and Monash University, reflecting a strong academic foundation in advanced analytics and applied mathematics.
Currently serving as the Global Data Science Leader at Unilever, Dr. Thomas leads the global data science strategy for Integrated Demand Forecasting. He is responsible for the development and oversight of Unilever's global forecasting engine, a critical system deployed across 40 countries that generates weekly demand forecasts for over 3 million time series. His work not only drives operational efficiency but also enables data-driven decision-making at scale within one of the world's largest consumer goods companies.