pbiplintRules / Performance
Reduce the number of objects within visuals
What it checks
Visuals with more entries in their field wells than the threshold, 6 by default, counting each column, measure, visual calculation, or sparkline in any of the visual's wells as one.
Each finding names the visual, as "Sales by product" on "Overview" when it has a title and tableEx (c897ed) on "Overview" when it has none, and its detail gives the count, as 7 fields bound, more than 6.
Example
The table shows two columns from Product and five measures from Sales.
visual.json{
"$schema": "https://developer.microsoft.com/json-schemas/fabric/item/report/definition/visualContainer/2.8.0/schema.json",
"name": "c897ed0802274ab55e2d",
"position": { "x": 575, "y": 518, "z": 11007, "height": 190, "width": 651, "tabOrder": 11007 },
"visual": {
"visualType": "tableEx",
"query": {
"queryState": {
"Values": {
"projections": [
{
"field": { "Column": { "Expression": { "SourceRef": { "Entity": "Product" } }, "Property": "Category" } },
"queryRef": "Product.Category"
},
{
"field": { "Column": { "Expression": { "SourceRef": { "Entity": "Product" } }, "Property": "Product Name" } },
"queryRef": "Product.Product Name"
},
{
"field": { "Measure": { "Expression": { "SourceRef": { "Entity": "Sales" } }, "Property": "Total Sales" } },
"queryRef": "Sales.Total Sales"
},
{
"field": { "Measure": { "Expression": { "SourceRef": { "Entity": "Sales" } }, "Property": "Total Quantity" } },
"queryRef": "Sales.Total Quantity"
},
{
"field": { "Measure": { "Expression": { "SourceRef": { "Entity": "Sales" } }, "Property": "Average Unit Price" } },
"queryRef": "Sales.Average Unit Price"
},
{
"field": { "Measure": { "Expression": { "SourceRef": { "Entity": "Sales" } }, "Property": "Total Cost" } },
"queryRef": "Sales.Total Cost"
},
{
"field": { "Measure": { "Expression": { "SourceRef": { "Entity": "Sales" } }, "Property": "Margin %" } },
"queryRef": "Sales.Margin %"
}
]
}
}
}
}
}
visual.json{
"$schema": "https://developer.microsoft.com/json-schemas/fabric/item/report/definition/visualContainer/2.8.0/schema.json",
"name": "c897ed0802274ab55e2d",
"position": { "x": 575, "y": 518, "z": 11007, "height": 190, "width": 651, "tabOrder": 11007 },
"visual": {
"visualType": "tableEx",
"query": {
"queryState": {
"Values": {
"projections": [
{
"field": { "Column": { "Expression": { "SourceRef": { "Entity": "Product" } }, "Property": "Category" } },
"queryRef": "Product.Category"
},
{
"field": { "Column": { "Expression": { "SourceRef": { "Entity": "Product" } }, "Property": "Product Name" } },
"queryRef": "Product.Product Name"
},
{
"field": { "Measure": { "Expression": { "SourceRef": { "Entity": "Sales" } }, "Property": "Total Sales" } },
"queryRef": "Sales.Total Sales"
},
{
"field": { "Measure": { "Expression": { "SourceRef": { "Entity": "Sales" } }, "Property": "Total Quantity" } },
"queryRef": "Sales.Total Quantity"
},
{
"field": { "Measure": { "Expression": { "SourceRef": { "Entity": "Sales" } }, "Property": "Average Unit Price" } },
"queryRef": "Sales.Average Unit Price"
},
{
"field": { "Measure": { "Expression": { "SourceRef": { "Entity": "Sales" } }, "Property": "Margin %" } },
"queryRef": "Sales.Margin %"
}
]
}
}
}
}
}
The fix removes Total Cost, which the table's Total Sales and Margin % already imply.
Why it matters
A visual sends one query for everything in its wells, so every field makes that query bigger. Each column is another level the rows are grouped by, which multiplies the rows the model has to produce, and each measure is another calculation for every one of those rows. One wide table can ask the model for more work than the rest of the page together, and a reader has to scan across all of it to find the number they came for.
How to fix it
Keep the fields the visual is about and move the rest to where a reader asks for them: secondary measures to a report page tooltip that shows them for the row under the pointer, and detail columns to a drillthrough page. In Power BI Desktop, remove a field by selecting the X beside it in the field well on the Visualizations pane. Where readers do want to choose among many measures or columns, a field parameter (Modeling, then New parameter, then Fields) puts the choice in a slicer, and the visual shows only the fields a reader has picked. In visual.json, each field is one entry in a well's projections array under visual.query.queryState, and deleting the entry removes the field.
When to ignore it
A table or matrix whose job is to show many columns side by side, such as the detail table on a drillthrough page or a table readers export to Excel, is doing what it is for. Six is also a strict budget: Microsoft's own guidance recommends limiting a visual to 10 to 20 fields. Check the visual with Performance Analyzer in Power BI Desktop before splitting it, and ignore the finding where it loads quickly and the width is what readers came for.
To ignore this rule on one visual, add { "name": "pbiplint.ignore", "value": "REDUCE_OBJECTS_WITHIN_VISUALS" } to the annotations array of its visual.json. Power BI Desktop keeps the annotation. To turn the rule off for a whole project, set "REDUCE_OBJECTS_WITHIN_VISUALS": "off" under rules in pbiplint.config.json.
Quirks
- pbiplint counts the fields bound to the visual's roles once, where PBI Inspector counts every projections array in the file and can count a field twice.
- Only the entries in the wells count, one for each well a field sits in, so a field in two wells counts twice, in PBI Inspector too. A field in the visual's filters that is not also in a well is not counted.
Related rules
REDUCE_VISUALS_ON_PAGEis the same budget one level up: splitting a wide visual in two clears this finding and adds to that rule's count.AVOID_SHOW_ITEMS_WITH_NO_DATAmakes each extra grouping field costlier, because the visual then lists groups with no data beside those with data, and every grouping field multiplies them.
Links
- Customize the Visualizations pane, and how many fields a visual should hold
- Use field parameters in Power BI reports
Ported from PBI Inspector's base rules by Nat Van Gulck.