Case Study: Directly achieves up to 80% more inquiries resolved and 20% higher customer satisfaction with Databricks

Driving customer retention with data and ML

Directly provides AI-driven customer support that blends automation with human help, but the company struggled with unreliable, hard-to-access data and an infrastructure that hindered scalable ML and analytics. Engineers faced slow, cyclical access to fresh customer data, limited compute, and sampling-based workflows that slowed model development and delayed insights for non-technical teams.

By adopting Databricks with Delta Lake on AWS, Directly built a unified, 24/7 real-time pipeline that democratized data access, supported BI via Looker, and accelerated distributed model training with MLflow. The change reduced data ingestion from hours to seconds, boosted issue resolution by up to 80% in some areas, improved customer satisfaction by 20% in those areas, and sped collaboration and innovation across teams.

Directly

Sinan Ozdemir

Director of Data Science