HomeChef tamed Snowflake costs while raising cost awareness across their analytics team

SELECT provided immediate cost visibility and optimization insights that HomeChef's internal tools didn't provide. Setup took just 30 minutes. SELECT's deep integration with dbt and comprehensive cost breakdown enabled the team to quickly identify high-impact optimization opportunities across their 1,200+ dbt models while providing the visibility they previously built custom dashboards to achieve.

HomeChef migrated from EMR to Snowflake five years ago, initially using it to fuel Looker dashboards with daily dbt batch updates, resulting in a stable and cost-effective setup. Everything changed when HomeChef started using Snowflake to power more and more operational workloads. They went from running queries once a day to running queries every 15 minutes. When you do a lot of upserts on churn tables, Snowflake spend can move up very quickly. They had about an order of magnitude increase pretty much overnight.

Before SELECT, HomeChef built internal cost monitoring through incremental query history harvesting, warehouse segregation for dbt environment spend tracking, and custom Looker dashboards. The homegrown solution worked for routine monitoring but became problematic when new issues arose. Adding features required PRing new dimensions into tables and measures into dashboards, while ad-hoc analysis was slow due to Snowflake's complex metadata structure. The inertia and friction was high. As the feature set that they used inside Snowflake expanded, they were effectively learning a new esoteric dataset each time.

When costs spiked, HomeChef needed immediate visibility into what was driving spend. Within 30 minutes, they were able to get up and running with SELECT. SELECT's deep understanding of the dbt ecosystem and comprehensive cost visibility made the transition seamless, eliminating their need for homegrown solutions.

HomeChef achieved significant cost reductions by first addressing storage issues (saving 20%), then using SELECT to identify performance optimization opportunities worth an additional 40%. With 1,200+ dbt models, identifying optimization targets manually would have been impossible. SELECT's cost breakdown and performance insights enabled the team to focus their efforts where they'd have maximum impact. SELECT served as both their target identification tool and their monitoring panopticon for performance optimizations.

SELECT transformed HomeChef's approach from reactive firefighting to proactive monitoring. Devin now starts every workday with a quick SELECT review. He gets his coffee, looks at the pipes, then checks SELECT. It takes him two to three minutes to get a sense of what's going on in their Snowflake setup every day. This routine enables early detection of anomalies, new workloads, and cost spikes before they become major issues. He can see changes in usage over time and it lets them know if there's new use cases they haven't anticipated, which lets the business analytics team work with those end users directly.

Perhaps the most significant cultural change was analytics engineers beginning to consider optimization during development rather than after deployment. Through pairing sessions with SELECT insights, the broader team learned query optimization principles. Now analytics engineers proactively optimize performance before submitting pull requests. The AEs now will front-load 'hey, this is going to be an expensive one' and will say 'I got the runtime on this down from X to Y prior to putting in the PR because I knew this window function was going to kill me'. Equally important, SELECT helps guide when not to optimize, preventing wasted effort on low-impact models.