JUNE 27-30
Edmunds
Customer Case Study
Edmunds is a leading car information and shopping network serving nearly 20 million visitors each month.
Industry
Automotive Retail
Vertical Use Case
- Improving online inventory management
Technical Use Case
- Data ingest, ETL
- Machine learning
Customer TestimonialThe Challenges
Edmunds is on a mission to make car shopping an easy experience for all. On a monthly basis, tens of millions of customers visit Edmunds.com to begin their car research journey. Core to providing a phenomenal user experience is powering their applications with data. As Edmunds built out these capabilities, they ran into a number of challenges:
- Unable to Maintain Accurate Inventory Status: 100’s of TBs of data collected from third-party providers made it difficult to accurately maintain vehicle inventory online — resulting in missing or inaccurate details which impacted the customer experience.
- Infrastructure Complexity: Maintaining infrastructure to support data processing and machine learning requirements (such as the latest ML frameworks) required significant DevOps effort to scale.
- Poor Collaboration Due to Team Silos: Data scientists were working in disjointed and unproductive ways.
- Model Deployment Manual and Inconsistent: Lack of automation and consistent processes to ensure efficient model management.
The Solution
With Databricks, Edmunds is now able to simplify access to their disparate data sources and ensure their inventory of vehicle listings on their website is accurate and up to date, improving overall customer satisfaction.
- Automated Infrastructure: Fully managed, serverless cloud infrastructure for speed, cost control and elasticity.
- Interactive Workspace: Shared workspaces significantly improved collaboration across teams with different skillsets. Data scientists and engineers can write and share code in their preferred languages (SQL, R, Python) while interactive dashboards enable less technical users to engage with the data from within the same platform.
- Streamlined ML Lifecycle: MLflow serves as the core framework and tool to make model deployment to production very easy and seamless.
- Optimized Data Lake: Delta Lake provides a centralized system that ensures reliability, transactionality, and performance of their data.
The Results
With the implementation of Databricks, Edmunds.com was able to democratize data access across their organization, allowing its data engineering, data science, and business analyst teams to work collaboratively on the data at scale. They were also able to ensure data reliability and performance of ETL pipelines and streamline the ML model lifecycle. Edmunds.com also achieved the following quantitative results:
- Improved operational efficiencies from data access to the productionization of ML models has resulted in millions of dollars in savings.
- Accelerated ad hoc data analysis by six-fold allowing them to answer data integrity questions faster.
- Improved reporting speed by reducing processing time by 60 percent, or an average of 3-5 hours per week for the engineering team.
- Improved vehicle data quality metrics across their website by 35 percent.
Within Edmunds, Databricks democratizes data, data engineering, and machine learning, and allows us to instill data-driven principles within the organization.