Cost Advisor: cut Snowflake costs with automated, read-only analysis
A free Snowflake Native App that finds where your spend is going and what to do about it. It reads your ACCOUNT_USAGE data inside your own account, computes cost and right-sizing recommendations, and hands you the exact SQL to act on them. It never modifies your resources.
What it analyzes
Warehouses
Right-sizing recommendations, auto-suspend tuning, and idle time detection for every warehouse in your account.
Storage
Unused tables sorted by storage cost, clustering opportunities, and time-travel retention review.
Workloads
Query patterns by day and hour, query anti-patterns, and serverless credit consumption per service.
Governance
Month-to-date spend per usage group or team, plus resource monitor coverage and thresholds.
Insights
A cross-domain, prioritized list of your biggest savings opportunities, each with copy-paste SQL to act on it.
Cortex AI cost
What Snowflake Cortex actually costs, per model and per function, and who drove it. Six checks flag the ways AI spend runs away: an oversized model on a labelling task, one statement fanning across a table, capacity bought and never drawn.
Security and architecture
Everything runs inside your Snowflake account and nothing leaves it. No external access integration, no API integration, no network rule, no secret.
Scroll the diagram sideways.
Privileges we ask for, and why
CREATE COMPUTE POOLRun the container service that serves the dashboard.
CREATE WAREHOUSEProvision an X-Small warehouse to query ACCOUNT_USAGE.
BIND SERVICE ENDPOINTExpose the web UI as a Snowflake-authenticated endpoint.
IMPORTED PRIVILEGES ON SNOWFLAKE DBRead the ACCOUNT_USAGE metadata views. Also conveys Cortex access for the optional account brief.
EXECUTE TASKRun the daily email digest on a schedule. Requested at install, but nothing is scheduled unless you turn alerts on.
Recommendations arrive as SQL you run yourself. The app issues no DDL or DML against your objects and has no code path that could. It creates only its own: a compute pool, an X-Small warehouse, cache tables and five procedures.
How the image is built and checked
Built from python:3.11-slim, runs as non-root uid 10001. Every build runs all 6 gates below. 4 of them block the release.
Secret scanning across full git history
Container CVE scan, fails the build on CRITICAL
Malware scan of the built image
Python static analysis
Dependency CVEs
Static type checking
Send us anything this page does not answer.
Getting started
Four steps from install to your first savings recommendation.
- 1
Install from the Snowflake Marketplace
Get the app from the Snowflake Marketplace listing. Installation is free.
- 2
Grant the four privileges
When you open the app, Snowsight prompts for the privileges it needs to run:
- Create compute pool: runs the optimizer service
- Create warehouse: an X-Small warehouse to query ACCOUNT_USAGE
- Bind service endpoint: exposes the web UI
- Imported privileges on the SNOWFLAKE database: read-only ACCOUNT_USAGE access
- 3
Grant read access to account usage
Run this one statement as ACCOUNTADMIN so the app can read your usage data:
GRANT IMPORTED PRIVILEGES ON DATABASE SNOWFLAKE TO APPLICATION TABBY_COST_OPTIMIZER; - 4
Start the service and open the dashboard
On the launcher console, click Start optimizer. First launch takes 1-2 minutes while the container service provisions. Then click Open Cost Advisor to load the dashboard.
Running cost
The app itself is free. It runs one X-Small compute pool node and an X-Small warehouse in your account, both with auto-suspend. Use Stop service on the console when you are not analyzing.
Uninstall
Remove the app and everything it created with one statement:
DROP APPLICATION TABBY_COST_OPTIMIZER;Inside the dashboard
12 pages across 6 sections, from a spend overview down to per-table detail.
Overview
- Home: KPI cards, daily spend trend, top opportunities, anomalies
Optimization
- Warehouses: Right-sizing recommendations per warehouse
- Storage: Storage category breakdown and top tables
- Clustering: Tables with degraded clustering
- Unused tables: Unused tables sorted by storage cost
Workloads
- Query patterns: Day-of-week by hour heatmap and top queries
- Serverless: Credit consumption per serverless service
- dbt models: Compute cost attributed to models, tests and seeds
Governance
- Usage groups: Month-to-date spend per usage group or team
- Resource monitors: Monitor coverage and thresholds
Intelligence
- Cortex AI: Cortex spend by model, function and owner, with the anti-patterns flagged
Configuration
- Settings: Pricing config, thresholds, refresh, connection info
Pricing
Free
No license fee, no tiers, no trial clock. You pay only the Snowflake compute the app uses in your own account.
Support and legal
- support@tabbystack.io
- Terms of Service
- Privacy Policy
- Built by Tabby Stack, a certified Snowflake and dbt partner
See where your Snowflake spend is going
Install free from the Snowflake Marketplace. Want hands-on help instead? Our team runs full cost optimization engagements.