Guide · FinOps
How to Set Up AWS Cost Anomaly Detection
AWS Cost Anomaly Detection flags unusual spending. Here's how to set it up—monitors, alerts, and root cause analysis.
Anomaly detection supports Cost Optimization—pair it with hard limits, not alerts alone. Primary control: Cost Optimization
Thresholds miss weird spend; ML catches shape
AWS Cost Anomaly Detection learns your baseline and flags unusual patterns—by service, account, or tag—even when you're still under a Budget percent. One team under 50% of budget still got a 20% odd spike in one service/region; Budgets stayed quiet; Anomaly Detection pointed at a misconfigured Lambda before it compounded.
Wire a monitor that pages someone
Enable Cost Anomaly Detection in Cost Management. Create a monitor (all costs, or split by account/service/tag). Attach SNS, confirm the email (same trap as Budgets). Start with default sensitivity; tighten only after you see noise. Review the anomaly dashboard weekly—root cause columns beat guessing from the invoice.
Budgets + anomalies, not either/or
Budgets answer "did we cross $X?"; anomalies answer "does this look wrong?" Both belong in FinOps. APRF Cost Optimization expects early warning with a path to the offending service—not hope that the next invoice is fine.
Next: Cost Optimization
Open the related pillar specification for mandatory checks, artifacts, and pass conditions. Self-attest is optional.
Related
Frequently asked questions
- How do I set up AWS Cost Anomaly Detection?
- Go to Billing & Cost Management > Cost Anomaly Detection. Create a monitor (all costs or by service). Set up SNS alerts. Confirm the email subscription.
- What is the difference between AWS Budgets and Cost Anomaly Detection?
- Budgets alert on thresholds (e.g., 80% of expected spend). Anomaly Detection catches unusual patterns—e.g., one service spiking. Use both.
- Does Cost Anomaly Detection cost money?
- Cost Anomaly Detection is free. You pay for the underlying Cost Explorer data (if not already enabled).