Case Study: Via achieves operationally efficient omnichannel retail and R$3.9M savings with Databricks

Building an operationally efficient omnichannel business

Via (formerly Via Varejo) is a major Brazilian retailer serving nearly 100 million customers with electronics, appliances, furniture and credit services. The company struggled with decades of siloed data and a complex Hadoop mainframe that prevented collaboration and scalable ML, leading to poor demand forecasting, supply‑chain disruption and gaps in fraud prevention and customer insights.

By moving to the Databricks Lakehouse Platform on Azure and standardizing on Delta Lake, Feature Store, Databricks SQL and MLflow, Via unified its data and operationalized ML across use cases like CLV, demand forecasting, safety stock, fraud detection and churn. The platform improved governance and collaboration, sped time‑to‑market for models, increased productivity by 30% (R$3.9M savings) and cut data/compute costs by 25%, while boosting conversion, customer lifetime value and supply‑chain efficiency.

Via

Cezar Steinz

Manager of MLOps