Client: MetricHarbor · AI/ML & Data · US analytics platform (100 staff)
Reducing Snowflake Spend for an Analytics Platform
TechSupport Team supported MetricHarbor’s Snowflake and dbt analytics platform. Over six months, the client’s Snowflake bill fell from $52,000 to $36,000 per month. Airflow pipeline failures fell from 38 to 8 per month.
Client and engagement facts
- Client
- MetricHarbor
- Industry
- AI/ML & Data
- Business context
- US analytics platform (100 staff)
- Delivery model
- Data platform support
- Application
- Analytics platform on Snowflake
- Supplied geography
- US
- Company size in source
- 100 staff
Related services: Application support
Application stack and tools
Application stack
- Snowflake
- dbt
- Airflow
- Python
- AWS
Tools used
- Snowflake usage records
- Airflow
- Datadog
- dbt
Why Support Was Needed
A dbt refactor dropped a filter, causing three dashboards to scan a 40 TB table every 15 minutes. The engineers restored the filter and added cost alerts by dashboard.
Technical Work
- Rebuilt the affected dbt model with the missing filter.
- Right-sized warehouses based on actual concurrency.
- Added per-dashboard cost alerting so waste becomes visible within hours.
Representative Incident
A Snowflake price change added cost in month five. The team adjusted warehouse sizing again within two weeks to partially offset it.
Results and how they were measured
| Metric | Baseline | Result | Units | Measurement window | Source and scope |
|---|---|---|---|---|---|
| Snowflake spend | $52k/mo | $36k/mo | 6 months | Client Snowflake | |
| Pipeline failures | 38/mo | 8/mo | /mo | 6 months | Client Airflow |
| Dashboard freshness | 6 h | 45 min | Baseline: h; Result: min | 6 months | Client Datadog |
Evidence and Limitations
The comparison uses six months of client Snowflake, Airflow and Datadog records. Usage was flat at the time of the original cost spike, and the provider changed pricing during the reporting window.
Cause of Excess Query Cost
A dropped WHERE clause in a dbt refactor caused three dashboards to scan the full table.
Scope a Similar Engagement
Share one query or dashboard that drives a disproportionate share of your platform cost. Request a support plan or explore application support.
Get a technical support plan
Tell us which application needs support, what is reaching your engineers and when you need coverage. TechSupport Team will review the fit and explain the next scoping step.
- We reply by email.
- We agree a call about the roles, your stack and the hours.
- If there is a fit, we propose engineers after an assessment.