Snowflake Cost Optimisation Strategies

Snowflake's consumption-based pricing is flexible, but without governance it can grow rapidly. We regularly see organisations where Snowflake spend grows 20–40% quarter-on-quarter without corresponding value increases.

Common cost drivers

• Over-sized warehouses — running XL warehouses for queries that need a Small
• No auto-suspend — warehouses running idle for hours between queries
• Inefficient queries — full table scans, missing clustering, poor join strategies
• Duplicate processing — multiple teams running identical transformations
• No resource monitors — no alerts or limits when spend exceeds expectations

Optimisation strategies

1. Right-size warehouses — match warehouse size to actual workload requirements. Most analytical queries run efficiently on Small or Medium.
2. Implement auto-suspend and auto-resume — set aggressive suspend timeouts (60–120 seconds for most workloads)
3. Query optimisation — add clustering keys on large tables, use materialised views for repeated aggregations, avoid SELECT *
4. Resource monitors — set credit quotas per warehouse with alerts at 75% and hard stops at 100%
5. Warehouse governance — assign warehouses by workload type (reporting, ETL, ad-hoc) with appropriate sizing for each
6. Storage optimisation — time-travel retention policies, transient tables for staging, regular COPY history cleanup

What we typically achieve

In our experience, a structured optimisation engagement can reduce Snowflake costs by 30–50% without impacting performance. In some cases we have achieved greater reductions by consolidating redundant processing and implementing proper warehouse governance.

The ongoing approach

Cost optimisation is not a one-time activity. We recommend implementing monitoring dashboards, monthly cost reviews, and automated alerting to maintain control as usage evolves.