Snowflake cost spikes: causes, checks, and alert policy.
Snowflake spend rises when warehouses stay warm, query patterns change, or analytics and AI workflows run at higher frequency.
What usually moves the Snowflake bill
Warehouses are resized, run longer, or miss auto-suspend settings.
Dashboards, dbt jobs, reverse ETL, or analyst queries scan more data.
Data sharing, replication, storage, or transfer patterns increase.
Experimentation workloads move from ad hoc to scheduled jobs.
Triage checklist
- Break spend down by warehouse, query tag, database, user, and workload.
- Compare warehouse runtime, queue time, and auto-suspend behaviour.
- Review scheduled jobs and dashboard refresh frequency.
- Look for large scans, new data sources, and storage growth.
Green, amber, red thresholds for Snowflake
Green
Daily Snowflake spend is within 10% of baseline and warehouse runtime is expected.
Amber
Daily Snowflake spend is 10-25% above baseline or warehouse runtime grows unexpectedly.
Red
Daily Snowflake spend is more than 25% above baseline or forecast exceeds data platform budget.
Turn this playbook into a daily signal.
StackSpend connects Snowflake to your cloud and AI cost view with daily Slack or email reporting, anomaly detection, and pace-to-forecast.