Problem
The goal is not to understand what exists in a semantic model, but how report consumers actually use it.
As a semantic model owner, I understand what my model contains and why those assets were originally included. What is often missing is visibility into how report consumers actually interact with those assets after deployment.
Requirements gathering and stakeholder feedback help guide model design, but usage patterns evolve over time and may differ from original expectations. Existing Power BI and Fabric usage metrics help measure report adoption and engagement, but they provide limited insight into how semantic models themselves are being used.
As semantic models mature, additional columns, measures, hierarchies, filters, and report elements are often added to support new business requirements. Over time, it becomes increasingly difficult to determine which assets continue to provide value and which may no longer be relevant.
Proposed Solution
Provide semantic model usage insights based on actual consumer interactions and query behavior.
Examples could include:
- Most frequently used measures
- Most frequently used columns and dimensions
- Most common filter and slicer combinations
- Most common drill paths
- Rarely used fields
- Unused fields over a configurable time period
- Candidate columns, measures, or report elements for review
- Suggestions for model optimization based on observed usage patterns
- Recommendations for improving report usability and performance
Why This Matters
These insights would help organizations:
- Reduce model bloat
- Simplify report experiences
- Improve maintainability
- Focus development efforts on high-value features
- Increase confidence when cleaning up technical debt
- Potentially reduce storage, refresh, and processing costs
Additional Consideration
Microsoft currently analyzes query behavior for features such as Automatic Aggregations. If similar usage insights are available, surfacing actionable recommendations to semantic model owners could help organizations continuously improve the quality, efficiency, and maintainability of their reporting solutions.
