Chalice AI Increases Snowflake Compute by 21% for the Same Spend
Challenges
- 500% increase in Snowflake usage from 2022 to 2024 required new attention to cost efficiency
- Performance demands required optimizations not to negatively impact performance
- Small team required a solution that would not require excessive effort to operate
Results with Keebo
Chalice AI helps clients optimize their programmatic advertising campaigns around specific goals & targets (i.e. store sales, customer acquisition cost) through artificial intelligence and machine learning. These efforts require extensive data and compute resources. As Chalice AI grew, their Snowflake costs began to balloon. By implementing Keebo’s AI-powered warehouse optimizations, Chalice AI consumed 15,000 fewer Snowflake credits than previously and cut overall spend by 21%.
Our three considerations were: Does [Keebo] cost less than the money we’ll save? Does it impact query performance? And does it take a lot of manpower to maintain? And the answer to all three was exactly where we wanted them to be.
Joe Wilson,
CTO, Chalice AI
Rapid growth requiring increased cost efficiency
To optimize their clients’ programmatic advertising campaigns, Chalice AI compiles, aggregates, and transforms massive datasets from a wide variety of platforms. Many of these data are category-specific, with high value for one vertical and none for others. Snowflake is critical for managing such large, diverse datasets.
As Chalice AI grew, however, their data usage increased exponentially. From 2022 to 2024, their Snowflake usage grew by 500%. Chalice AI found the workload of manually optimizing their Snowflake warehouses not only overwhelming for their small team, but diverting resources from pursuing other opportunities to grow the business and better serve customers.
Pricing structure fostering aligned incentives
When presented with Keebo as an automated warehouse optimization solution, the Chalice AI team jumped at the opportunity. A key point in their decision making was the cost structure, that Keebo is priced as a percentage of Snowflake cost savings. “I looked at it as free money,” said Joe Wilson, Chalice AI CTO. “We pay Keebo X and save 2.5X. It’s a no-brainer—why wouldn’t we do that?”
No negative impact on performance
In addition to the pricing structure, two other factors made Keebo a no-brainer for Chalice. First, their optimizations had no negative impact on performance. Second, it required few internal resources to manage, which meant their savings was a true net positive.
