Case Study: RB (Reckitt Benckiser) achieves ML‑powered demand forecasting at scale — supporting 500,000 stores — with Databricks

Achieving demand forecasting at scale

Reckitt Benckiser (RB), a multinational consumer packaged goods company serving retailers across 60+ countries, faced major challenges forecasting demand for millions of fragmented "neighborhood" grocery stores. With over 16,000 field reps and more than 250 disjointed data pipelines processing ~2TB daily, RB’s legacy Hadoop setup was costly, slow to scale, and made it difficult for finance, sales, and operations to access external data sources and deliver timely insights.

RB moved to Azure Databricks’ Unified Data Analytics Platform—using managed clusters, collaborative notebooks, and Delta Lake—to simplify operations, speed development, and compress data. The change cut storage from ~80TB to 2TB (≈98% compression), halved 24x7 pipeline runtimes (24h → 13h), reduced DevOps overhead, and expanded RB’s support capacity from ~45,000 to nearly 500,000 stores (10x), unlocking faster, ML-powered demand forecasting and cost savings.

RB

Atif Ahmed

Director of Advanced Analytics