FSICOST05: How do you track anomalies in your ongoing costs for AWS services?

Understanding your organization's costs and drivers is critical for managing your cost and usage effectively, and identifying cost-reduction opportunities. Accurate cost and usage monitoring allows you to make more informed decisions about where to allocate resources within your organization.

FSICOST05-BP01 Be aware of anomalies and periodically review your architecture

Anomalies can drive up cost. Set up AWS Cost Anomaly Detection to detect and alert on anomalous spend patterns in your deployed AWS services. Cost Anomaly Detection automatically determines thresholds each day by adjusting for organic growth and seasonal trends (like usage increases from Sunday to Monday or increased spend at the beginning of the month) through machine learning models. Financial systems usually integrate with several other third-party systems, and Cost Anomaly Detection can integrate with these systems as well.

Extend Cost Anomaly Detection with custom metrics such as token_in, token_out, and embedding_ops, and route alerts to product or data owners when cost spikes correspond to new prompt deployments, unexpected retrieval-augmented generation (RAG) expansion, or fine-tuning jobs running out of schedule. Combine these alerts with CloudWatch dashboards to correlate generative AI usage trends with cost anomalies in near real-time.