About Cost Anomaly Detection

Based on the information provided, there are a few potential reasons why AWS Cost Anomaly Detection didn't detect the significant cost increase from April 2nd to April 5th:

1. Detection frequency: AWS Cost Anomaly Detection runs approximately three times a day to monitor for anomalies. However, there can be a delay of up to 24 hours in detecting anomalies after usage occurs. This delay is because the service uses data from Cost Explorer, which itself has a delay of up to 24 hours.

2. Historical data requirement: For a new service subscription, AWS Cost Anomaly Detection needs 10 days of historical service usage data before it can start detecting anomalies for that service. If this was a new S3 usage pattern, it's possible that there wasn't enough historical data to establish a baseline.

3. Threshold setting: While you mentioned setting a 10% threshold, it's important to note that AWS Cost Anomaly Detection uses machine learning models to determine what constitutes an anomaly. The 10% threshold you set might not directly correlate to how the service identifies unusual spend patterns.

4. Monitor type: The effectiveness of anomaly detection can depend on how you've set up your cost monitors. Ensure that you have the appropriate monitor type (AWS services, linked account, cost category, or cost allocation tag) that covers your S3 usage.

To address this issue and improve anomaly detection for future cases, consider the following solutions:

1. Wait for more data: Allow more time for the service to gather sufficient historical data to establish accurate baselines for your S3 usage patterns.

2. Review monitor settings: Check your cost monitor configuration to ensure it's set up to properly capture S3 usage. You might want to create a specific monitor for S3 or for the tag you're using.

3. Adjust alert preferences: Review and possibly adjust your alert preferences to ensure you're receiving notifications for the right types of anomalies.

4. Use multiple monitoring approaches: In addition to Cost Anomaly Detection, consider using AWS Budgets or setting up custom CloudWatch alarms for more immediate notifications of cost increases.

5. Analyze manually: Regularly review your costs in AWS Cost Explorer, especially after implementing new services or changing usage patterns, to catch any issues that automated systems might miss.

Remember that while AWS Cost Anomaly Detection is a powerful tool, it works best as part of a comprehensive cost management strategy that includes regular manual reviews and multiple monitoring approaches.
Sources
Detecting unusual spend with AWS Cost Anomaly Detection - AWS Cost Management
Getting started with AWS Cost Anomaly Detection - AWS Cost Management
Community | From Cloud Chaos to Cost Control
Faster anomaly resolution with enhanced root cause analysis in AWS Cost Anomaly Detection | AWS Cloud Financial Management