Currently, the Cost Anomaly feature is designed to capture spikes in spend but does not monitor for dips. Similarly, the Alerts feature works on defined thresholds, which is more suitable for catching increases than decreases.

It would be valuable to have a feature that can detect sudden drops in cost as well, since unexpected decreases can also indicate issues such as:

Disabled services or resources that should be running

Failures in data pipelines or workloads

Gaps in logging/monitoring or data ingestion

Sudden deletion in data

Adding anomaly detection for dips would provide a more complete view of cost changes and help ensure early detection of potential problems.