Improving the speed and competitiveness of pricing to win more

Overview

Steelcase built an AI-powered pricing engine on the Databricks Data + AI Platform, using a feature store of 70+ predictive features and random forest models served in real time to automate discount recommendations across 70,000+ annual pricing scenarios, boosting auto-approvals from 35% to nearly 50% and freeing over 1,500 sales hours annually.

The challenge

Steelcase relied on a manual, legacy pricing approval process: sales teams spent about 20 minutes on average per pricing submission across over 70,000 pricing scenarios annually, with intricate discount configurations and multiple stakeholder types, and prior automation attempts using predefined authorization thresholds concentrated activity around administrative checkpoints instead of strategic pricing decisions.

The solution

Steelcase's data science team built an AI-powered pricing engine on the Databricks Data + AI Platform on Azure, using Azure Data Factory and Delta Lake with a medallion architecture, a Databricks feature store caching 70+ predictive features, Databricks Notebooks and MLflow for collaborative model development, and random forest models served via serverless real-time inference into its internal Pricing and Contracts Management application, governed by Unity Catalog.

Reported business value

Steelcase boosted automatic pricing approvals from 35% to nearly 50% and reclaimed over 1,500 hours annually that salespeople can now spend working with customers rather than pricing.