Amazon Q in Quick Sight for CUR data

Q&A
, ,
1

Hello - I use Quick Sight to ingest and display CUR, Compute Optimizer, and Trusted Advisor data. I’m beginning to work with Q for Quick Sight and am interested to know if there is AWS documentation (or other sources) for using Q in Quick Sight specifically for CUR data. I’m interested in developing topics and making them user facing so that teams can ask nature language questions about their data.

Thank you in advance for the feedback.

Matt

2

Hello Matthew,

For using Amazon Q with Quick Sight for CUR (Cost and Usage Report) data, here are the key resources and approaches:

Official Documentation:

  1. AWS Quick Sight Q documentation: Get started with Generative BI - Amazon QuickSight
  2. CUR data model best practices: Amazon QuickSight

Key Steps for CUR Data:

  1. Prepare your Topics:
  • Create topics focused on cost analysis
  • Include common cost metrics and dimensions
  • Add business context about AWS services and billing
  1. Potential Recommended Topic Areas for CUR:
  • Cost by service
  • Resource utilization
  • Spending trends
  • Reserved Instance coverage
  • Savings Plans utilization
  • Tag-based cost allocation
  1. PotentialSample Questions to Enable:
  • “What’s our monthly AWS spend?”
  • “Show me highest cost services”
  • “Compare costs across accounts”
  • “What’s the RI utilization trend?”

Note: Since Amazon Q for Quick Sight is relatively new, the documentation specific to CUR analysis is still evolving. I would recommended to:

  • Start with basic cost analysis topics
  • Gradually add complexity
  • Test and refine based on user feedback
  • Monitor Q’s performance with CUR-specific questions

Hope this gives some idea.

Cheers,
Deep

1 Like
3

Hi Deep - thank you for your reply. We’ve reviewed the existing AWS documentation, however, I was hopeful that someone within AWS had already worked this use case and would be able to share lessons learned. We’re working through it now, so we may have something to share with the community.

1 Like