Cost Explorer, Cost Anomaly Detection mismatch

Question

I am seeing a mismatch of the actual spent cost on multiple resources.

For an anomaly with the following details,

| Service | Start date | Last detected date | Duration | Total cost impact | Actual spend |
| --- | --- | --- | --- | --- | --- |
| Amazon Kinesis Analytics | June 11th | June 12th | 2 days | $90.64 | $95.51 |

Among the two Usage types during that anomaly period(APN2-KPU-Hour-Java, APN2-RunningApplicationStorage), APN2-KPU-Hour-Java was the potential root cause.

I applied the same date range and service to verify the root cause, but found a discrepancy on the actual spent cost.
While Cost Anomaly Detection showed $95.51, Cost Explorer showed $106.71 for the actual cost on AmazonKinesisAnalytics 6/11~6/12.
Narrowing down to APN2-KPU-Hour-Java gives $100.99, which is still bigger than the Cost Anomaly Detection data.

Further investigating other cases has given me that Cost Anomaly Detection cost is roughly 89.5% of Cost Explorer cost

I'm sure that data latency doesn't matter because Cost Anomaly Detection and Cost Explorer uses the same data(same latency). And I have verified that it wasn't a timezone issue.

Any help would be very much appreciated.

Garre Sandeep · Answer

Hello

check these steps to resolve the issue:
* Compare Filters: Ensure your date range, service selection, and other filters are identical in both CAD and CE.
* Breakdown by Usage Type: In CE, break down the cost for Amazon Kinesis Analytics during the anomaly period by Usage Type. See if the sum of individual usage types matches the total cost shown in CE.
* Check for Cost Adjustments: In CE, verify if there are any cost adjustments applied during the anomaly period. Look for refunds, credits, or other adjustments that might explain the difference.

For more details follow the links:

https://aws.amazon.com/aws-cost-management/aws-cost-anomaly-detection/faqs/

https://docs.aws.amazon.com/cost-management/latest/userguide/ce-what-is.html

https://docs.aws.amazon.com/cur/latest/userguide/what-is-cur.html

AWS-User-7098804 · Answer

Hi, Cost Anomaly Detection tries to report the two contributing factors in the root cause, but there may be additional causes, so the root cause impact may not equal the total.

From [FAQ](https://aws.amazon.com/aws-cost-management/aws-cost-anomaly-detection/faqs/ ):
“For the anomalies detected, we report up to two root causes. This is our best estimate of the largest contributing factors to the anomaly. Since we use machine learning models to select a maximum of two possible root causes, sometimes there are multiple, small contributors to the total impact, so the root cause explains only a small portion of it.”

Not sure if you used used manually applied filters in Cost Explorer, but worth mentioning that in the console, the anomaly details in the Detection history will have “View root cause” links that take you to Cost Explorer to visualize the anomaly. There will be one for the anomaly and separate ones if a root cause has be listed.

Cost Explorer, Cost Anomaly Detection mismatch

Add your answer

Relevant content