The Customer

automotiveMastermind®, the leading provider of predictive analytics and marketing automation solutions for dealerships and manufacturers, revolutionizes the way to find, engage, and win customers by successfully leveraging Big Data Behavioral Analytics. Their proprietary technology analyzes years of customer, service, and market data to better understand and predict exactly which customers are ready to buy, the reasons why, and the key messages most likely to close the sale. Mastermind work with over 1,300 dealer partners across the United States.

Challenges

Mastermind provides a SaaS product used by car dealerships to predict buying patterns and habits in order to better target future purchases. Event data is brought together into a predictive behavior score that indicates when a consumer is ready to buy.  

Since their SaaS analytics engine processes 1,000’s of customer and vehicle data inputs, the service needs to be robust and respond in a timely manner. Thousands of sales associates are querying the system daily for real-time analytics to close deals. There can’t be any performance bottlenecks, and the system needs to be stable and reliable while they continuously roll out new features and improve on the service.

To maintain their market leadership in predictive auto sales, Mastermind was looking for a solution to continuously deliver new features fast and reliably. Clusters can go down during development and rebuilding them takes up valuable time. To mitigate this risk, and deliver new functionality quickly, the team realized they need tools that incorporated end to end workflows to rapidly deploy new features as well as recover from a cluster breakdown during development.

In order to provide a 360-degree view on sales leads to their customer base, their SaaS needs to be responsive and scalable. Building a complex service that scales in real-time means that when problems arise they must be understood and quickly resolved before they affect their customers.

Without the right tools, debugging distributed systems and drilling down to identify the root cause can be challenging. Traditional logging and tracing tools don’t work with applications running in a dynamic cluster like Kubernetes. To pin point and solve problems, the team at Mastermind chose Prometheus — the de facto standard for monitoring and alerting applications running in Kubernetes.  

But after implementing an in-house installation of Prometheus themselves, they realized that it didn’t scale well and was time consuming to maintain. In the end, they needed Prometheus’ capability to deliver powerful insights, but they didn’t want the maintenance overhead.

“Weaveworks has made deployments incredibly easy. We’re now deploying anywhere from 30-40 services a day instantaneously. That’s something that would have taken us weeks to do in the past.”  - Deavon McCaffery, Principal Engineer, automotiveMastermind

After adopting Weave Cloud and GitOps, automotiveMastermind scaled their deployments, and went from releases on an 8-10 week cycle to more reliable deployments 10-100 times a day.