Spike detection
Live on BigQuery job statistics.
Use case · Cost
Partition cost guardrails
Stop the missing partition filter before finance sees it.
The problem
One explore drops a filter. Bytes billed jump overnight. Nobody owns the query.
The outcome
Cost spike detection flags the outlier jobs; playbooks name the next step.
2×+
Unusual vs baseline
Owner
Job identity context
Playbook
What to fix
Capabilities
Top expensive jobs
See who burned what.
Playbooks
Partition and clustering next steps.
Calculator
Estimate waste at /calculator.
- ✓/vectordata/cost-spikes is live
- ✓/learn/partition-filter-mistakes
- ✓Flat pricing — monitoring won’t bill-shock you
- ✓Demo includes a cost spike tile
Related
Watchdog / Fix first
AI ranks stale tables, failed dbt runs, and cost spikes into one Fix-first queue.
Read →Executive KPI trust
Know the KPI tables are fresh before the Monday leadership meeting.
Read →Board deck confidence
Freeze the numbers that go in the board pack — with a freshness check first.
Read →Data contracts / SLOs
Freshness targets and owner expectations per table — on the roadmap.
Read →