Simplify Data Schema Power BI is a useful feature that helps Copilot choose the right fields, measures, and tables for more accurate AI responses.
Simplify the Data Schema
This tool helps guide Copilot toward the most relevant fields, measures, and tables in your semantic model resulting in cleaner, more accurate responses.
However, many Power BI users misunderstand what this feature actually does.
In this article, we’ll break down:
- What Simplify the Data Schema is
- Where to find it
- What it does (and does NOT do)
- Real-world examples
- Best practices
- Common limitations to understand
What Is Simplify the Data Schema in Power BI?
Simplify the Data Schema is a feature inside Prep Data for AI that allows report creators to select which fields Copilot should prioritize when analyzing your Power BI model.
It helps reduce clutter inside complex semantic models by removing unnecessary columns, IDs, helper tables, and technical fields from Copilot’s primary decision-making process.
Where You Can Find It
You can access it inside:
- Power BI Desktop
- Power BI Service
- Microsoft Fabric experiences
- Chat with Your Data / Copilot pane
Navigate to:
Prep Data for AI → Simplify the Data Schema
Why This Feature Matters
Many enterprise Power BI models contain:
- Hundreds of columns
- Technical key fields
- Hidden parameter tables
- Complex bridge tables
- Duplicate business terms
- Multiple measures with similar names
Without optimization, Copilot may choose the wrong fields when answering natural language questions.
Example:
If a model contains:
- Customer Name
- Customer ID
- Customer Key
- Customer Number
Copilot may incorrectly use Customer ID instead of Customer Name in a visualization.
Simplifying the schema helps avoid these issues.
Important Clarification: This Is NOT a Security Feature
One of the biggest misconceptions is that deselecting fields means users can no longer access them.
That is false.
Simplify the Data Schema Does NOT Replace:
- Row-Level Security (RLS)
- Object-Level Security (OLS)
- Dataset permissions
- Workspace permissions
It is only a Copilot guidance layer.
Think of it as:
“Suggested fields for AI to use first.”
Not:
“Hide this data from users.”
What Happens When You Deselect a Field?
When a field is deselected:
- Copilot is less likely to use it in responses
- AI prioritizes selected fields first
- Responses may become cleaner and more business-friendly
However:
- Copilot may still use deselected fields in certain scenarios
- Full model access may still occur for some operations
- Relationships can still bring those fields into context
Hidden Fields Still Matter
Another surprise for many users:
Hidden columns and hidden tables may still appear in the AI schema.
Even if a table is hidden in the Fields pane, Copilot can still leverage it if relevant to the model.
That means hidden technical tables, parameters, and helper dimensions should still be reviewed.
Hierarchies Are Still Respected
If your semantic model contains hierarchies like:
- Year > Quarter > Month
- Country > Region > City
Copilot can still understand and use those hierarchies even if some individual columns are deselected.
Relationships Still Influence Answers
Power BI relationships remain foundational to how Copilot interprets data.
Even if a bridge table or supporting field is excluded from the simplified schema, Copilot may still reference it when needed.
Real Demo Example: Measure Behavior
In the transcript demonstration, a measure named:
Total Banks
was deselected from the simplified schema.
When the question was asked:
“What is Total Banks?”
Copilot did not use the measure.
Instead, it generated:
- Count Rows of Failed Bank Table
After re-enabling the measure and refreshing chat, Copilot correctly used:
Total Banks
This proves schema simplification directly influences how AI interprets user questions.
Why You Must Refresh Copilot After Changes
After modifying the AI schema:
- Apply changes
- Close the settings pane
- Close Copilot chat
- Reopen Copilot
This ensures Copilot reloads the newest schema settings.
Best Practices for Simplify the Data Schema
1. Remove Technical ID Fields
Deselect:
- Customer ID
- Product Key
- Employee Surrogate Key
2. Keep Business-Friendly Names
Prioritize:
- Customer Name
- Product Name
- Region
- Department
3. Reduce Duplicate Columns
Choose one preferred version of duplicate fields.
4. Keep Important Measures Enabled
Measures like:
- Total Sales
- Revenue
- Profit Margin
- Headcount
should remain available.
5. Test Frequently
Ask Copilot common business questions and refine the schema based on answers.
Key Limitations to Understand
Copilot may ignore the simplified schema for certain tasks:
- Creating report pages
- Searching data
- Generating DAX queries
For those scenarios, Copilot may inspect the full semantic model.
Final Thoughts
Simplify the Data Schema is one of the most valuable hidden features in Power BI Copilot.
Used correctly, it can:
- Improve AI answer quality
- Reduce confusion
- Encourage business-friendly visuals
- Guide Copilot toward trusted measures
- Make Chat with Your Data more useful
If you’re serious about AI adoption in Power BI, this should be part of every semantic model design process.
Need Help with Power BI Copilot or Microsoft Fabric?
Whether you’re migrating to Microsoft Fabric, optimizing semantic models, or training teams on Copilot, expert guidance can accelerate success.
