Case Study: Overstock achieves scalable, ML-driven one-to-one engagement with mParticle

How Overstock uses machine learning to drive one-to-one engagement

Overstock, a major U.S. online retailer focused on home goods, needed to deliver one-to-one personalized experiences across millions of SKUs but faced technical challenges around collecting, organizing, and modeling large streams of customer data in real time. To meet requirements for speed, scalability, and low engineering overhead, Overstock selected mParticle as the central Customer Data Platform to orchestrate user-level data alongside Snowflake and Databricks for storage and ML.

Using mParticle’s CDP and AudienceSync as the core of a real-time stack (with Snowflake for feature engineering and Databricks for fast model development), Overstock automated ad bidding and personalization pipelines and fed model outputs back into activation channels. The mParticle-centered solution produced measurable gains: a 10% lift in advertising efficiency in the first eight months, model deployment time cut from ~3 months to 1 day, vendor launch cycles reduced from months to days, plus improved conversion rates, lower API dependencies, higher uptime, and faster experimentation.

Overstock

Craig Kelly

Group Product Manager