First decide which total the reader needs. Category A can be 75% of the selected region and 30% of all counts in the fixture. Both are useful answers, but they should have different measure names and visible denominators.
This walkthrough specifies an original four-row model and DAX measures for later native Power BI testing. No PBIX file, model, DAX query, or visual has been created or executed. The numbers below are independently calculated expectations.
Make the filter path explicit
Use three proposed tables. FactCounts contains:
| Region | Category | Count |
|---|---|---|
| West | A | 3 |
| West | B | 1 |
| East | A | 2 |
| East | B | 4 |
DimRegion has two unique Region values, West and East. DimCategory has two unique Category values, A and B. In this tiny fixture, the labels themselves serve as the relationship keys.
Specify active one-to-many relationships from DimRegion[Region] to FactCounts[Region] and from DimCategory[Category] to FactCounts[Category]. Each dimension is the one side; filtering flows in a single direction from that dimension to the fact table. All fact keys have matching dimension keys.
The planned visual uses DimCategory[Category] for its category rows. Its region slicer uses DimRegion[Region]. Do not swap either field for the similarly named fact column while keeping the same expected trace.
There are no other page, report, visual, security, category-slicer, or hidden filters in this authored manifest. Record the actual native relationships and filters during implementation. The simple table is not enough to establish a different model's filter behavior.
Start with the count measure
The authored base measure is:
Count =
SUM ( FactCounts[Count] )
With West selected and category A on the visual row, the current count is 3. West/B is 1, and the West total across categories is 4. Across both regions and categories, the total is 10.
The following expression does not answer either share question:
Same Context Ratio =
DIVIDE ( [Count], [Count] )
For a nonzero cell, numerator and denominator are the same value, so the result is 1. With percentage formatting, that becomes 100%. It confirms division by the current cell, not a correct total.
Keep the base count visible while testing. A percentage alone hides which input is wrong.
Remove the category filter while retaining the region
Define the region-preserving denominator:
Region Total =
CALCULATE (
[Count],
REMOVEFILTERS ( DimCategory[Category] )
)
Then define the share:
Share of Region =
DIVIDE ( [Count], [Region Total] )
Microsoft's CALCULATE documentation describes evaluating an expression in a modified filter context. REMOVEFILTERS can clear specified filters; it is a modifier, not a function returning a replacement table.
In this model, removing the category-column filter lets both categories contribute to the denominator while the West region filter remains. The expected West/A calculation is 3 / 4 = 0.75, displayed as 75%.
The label “Region Total” belongs to this exact fixture. Removing one column does not promise to clear every category-related restriction in a larger model. Additional dimension columns, relationships, or filters require a revised manifest and new checks.
Do not substitute ALLSELECTED and call it universally equivalent. This article asks for a particular column filter to be removed, and the native acceptance test should check that exact contract.
Remove report-context filters for the all-count denominator
Specify a separate denominator and measure:
All Counts Total =
CALCULATE (
[Count],
REMOVEFILTERS ()
)
Share of All Counts =
DIVIDE ( [Count], [All Counts Total] )
For the declared four-row fixture without security restrictions, the all-count denominator is 10. West/A is therefore 3 / 10 = 0.30, displayed as 30%.
The numerator still comes from the current West/A cell. Only the denominator's evaluation context is modified. Clearing denominator filters does not mean that the slicer stops filtering the displayed numerator.
These two measures make the distinction readable:
| West category row | Count | Region Total | Share of Region | All Counts Total | Share of All Counts |
|---|---|---|---|---|---|
| A | 3 | 4 | 75% | 10 | 30% |
| B | 1 | 4 | 25% | 10 | 10% |
The global-share percentages for the West rows sum to 40%, because West contributes four of the fixture's ten counts. They should not be forced to sum to 100% by renormalizing after calculation.
“All Counts” here describes the authored test model. It is not a promise to bypass row-level security or expose inaccessible rows. Security and different deployment contexts require their own supported native review.
Change the slicer and check the subtotal
Switch the proposed region selection to East:
| East category row | Count | Region Total | Share of Region | All Counts Total | Share of All Counts |
|---|---|---|---|---|---|
| A | 2 | 6 | 33.3333% | 10 | 20% |
| B | 4 | 6 | 66.6667% | 10 | 40% |
The regional denominator changes from 4 to 6. The all-count denominator stays at 10 under the stated filter manifest. East/A's regional ratio is exactly 2/6, or one-third; its percentage display is rounded.
Now inspect a subtotal with West selected and no category-row filter:
| Context | Count | Region Total | Share of Region | All Counts Total | Share of All Counts |
|---|---|---|---|---|---|
| West subtotal | 4 | 4 | 100% | 10 | 40% |
| East subtotal | 6 | 6 | 100% | 10 | 60% |
These subtotals are expected measure evaluations in the subtotal context, not an instruction to average displayed row percentages. Averaging 75% and 25% gives 50%, which is not West's regional subtotal share.
Keep the subtotal context visible in the test ledger. Confirm the actual visual's subtotal settings and native values rather than reading a row-level result as proof of correct totals.
Preserve an undefined denominator
The proposed share measures use DIVIDE without an alternate result. Microsoft documents DIVIDE's default BLANK result for a zero denominator.
The four-row fixture's totals are positive. Use a separately identified zero-denominator test rather than altering these rows without recording a new model revision. The expectation for division by zero is BLANK, not an accepted 0% share.
A zero numerator over a positive denominator and an undefined ratio from a zero denominator are different cases. Do not convert both to the same 0% label for convenience. An explanatory review can say “Undefined: denominator is zero”; verify how the actual native visual presents BLANK.
Keep the ratios numeric. Apply percentage formatting once. The value 0.75 should display as 75%; multiplying by 100 before percentage formatting would produce 7,500%. Formatting and rounding should not change the underlying ratio used for subsequent reasoning.
Record model, context, and actual evidence
The native acceptance ledger needs model revision, table/key definitions, relationship direction and activity, measure text, visual fields, slicer selection, full filter manifest, current numerator, both denominators, raw ratios, displayed percentages, and subtotal context.
For West/A, record expected inputs 3, 4, and 10 alongside expected ratios 0.75 and 0.30. Actual values remain blank until the native model is tested. Repeat for East/A and the region subtotals, and keep the zero-denominator case separately identified.
A measure, relationship, data, visual-field, or filter change retires the old current expectations. Recompute the trace under the new manifest rather than reusing a screenshot from the earlier model. An optional preparation Reset restores the original four-row specification; it does not claim to clear native Power BI filters or service state.
Later production should inspect the native visual and the article explanation at the actual content width and narrow screen. The table should preserve numerator and denominator labels rather than presenting unexplained percentages.
Copy this prompt to prepare the model and tests
Create one complete, self-contained US-English HTML companion for “Power BI • choose your denominator”. Return the full HTML document, with inline CSS and any JavaScript needed for the local interactions. Use the original scenario, named records, numbers and expected cases below.
Design the page for this task: use clearly labeled input or source records, a readable result or worked example, and an adjacent explanation of the evidence. Use a restrained visual hierarchy, left-aligned text, readable tables or stacked records, and a responsive layout at 720px and 360px. Provide meaningful labeled controls for West; East; All regions; Category A; Category B; Subtotal; Apply revised counts; Separate zero fixture; Restore R1. Use keyboard-accessible native controls and visible focus. Keep the complete content readable without an external dependency.
The following brief supplies the substantive content and constraints. Native-format requests describe the material to represent or copy manually; this HTML must not pretend to create DOCX, XLSX, PDF, PPTX, H5P or another native file, run Python/Flask/Django/Snowflake/Power BI, connect a live service, send messages, or perform an authorized real-world action. Where the task requires a native source, actual observation or separate approval, retain that requirement and leave unsupplied evidence blank.
Prepare an original Power BI percent-of-total model specification, DAX measure text, and native acceptance ledger. Do not create a PBIX file, build a model, execute DAX, connect to a service, publish, or invent visual observations.
Fixture R1:
FactCounts rows:
West, A, 3
West, B, 1
East, A, 2
East, B, 4
DimRegion[Region]: unique West and East. DimCategory[Category]: unique A and B. Use those labels as the fixture keys. Specify active one-to-many, single-direction dimension-to-fact relationships on the matching columns. All fact keys match. Visual category rows use DimCategory[Category]. Region slicer uses DimRegion[Region]. No other filters or security restrictions are part of this fixture.
Authored measures:
Count = SUM(FactCounts[Count])
Region Total = CALCULATE([Count], REMOVEFILTERS(DimCategory[Category]))
Share of Region = DIVIDE([Count], [Region Total])
All Counts Total = CALCULATE([Count], REMOVEFILTERS())
Share of All Counts = DIVIDE([Count], [All Counts Total])
Keep default DIVIDE behavior; do not add alternate zero. Format raw ratios as percentages once, without multiplying by 100 first.
Expected checks, not actual outputs:
West/A: numerator 3, regional denominator 4, all-count denominator 10; ratios 0.75 and 0.30; displays 75% and 30%. West/B: 1/4 = 25%; 1/10 = 10%. East/A: 2/6 = one-third, displayed 33.3333%; 2/10 = 20%. East/B: 4/6, displayed 66.6667%; 4/10 = 40%. West subtotal with no category-row filter: 4/4 = 100%; 4/10 = 40%. East subtotal: 6/6 = 100%; 6/10 = 60%. Subtotal measures are evaluated in subtotal context, not averages of displayed percentages. DIVIDE([Count],[Count]) is a same-context counterexample yielding 1 for a nonzero cell, not a share measure. A separately identified zero-denominator test expects BLANK, not accepted 0%.
Explain:
The category-only modifier retains the region filter in this exact model. The no-argument modifier gives denominator 10 under this declared fixture. The numerator stays in its current visual context. Do not assert ALLSELECTED equivalence, automatic handling of additional dimension restrictions, security bypass, or universal totals in a different model.
Produce table/key/relationship specifications, measure definitions, numerator/denominator traces, subtotal and zero-denominator cases, and expected-versus-actual native test ledger. Record actual Desktop/version/model/visual/filter observations later; leave them blank now. Changes to data, relationships, measures, visual fields or filters retire prior current evidence. Preparation Reset restores R1 expectations; it does not prove native filter reset.
Native Power BI model and DAX/visual tests remain authoritative. Keep the required native records authoritative; do not claim NOS executes DAX, loads PBIX, shares a service report, or grants report access.
COMPLETE AUTHORED SCENARIO AND EXPECTED CASES
The following is the supplied reading/teaching/calculation record to represent in the companion. Preserve its particular content, examples, source references and unknowns. Wording about a future native implementation refers to that native task, not permission to omit this HTML companion.
# Power BI Percent of Total: Name the Denominator Before Writing DAX
Your category A row shows 100%, even though category B also has counts. The numerator is probably being divided by another measure evaluated in the same filtered cell. A percentage format cannot change that denominator.
First decide which total the reader needs. Category A can be 75% of the selected region and 30% of all counts in the fixture. Both are useful answers, but they should have different measure names and visible denominators.
This walkthrough specifies an original four-row model and DAX measures for later native Power BI testing. No PBIX file, model, DAX query, or visual has been created or executed. The numbers below are independently calculated expectations.
## Make the filter path explicit
Use three proposed tables. FactCounts contains:
| Region | Category | Count |
| --- | --- | --- |
| West | A | 3 |
| West | B | 1 |
| East | A | 2 |
| East | B | 4 |
DimRegion has two unique Region values, West and East. DimCategory has two unique Category values, A and B. In this tiny fixture, the labels themselves serve as the relationship keys.
Specify active one-to-many relationships from DimRegion[Region] to FactCounts[Region] and from DimCategory[Category] to FactCounts[Category]. Each dimension is the one side; filtering flows in a single direction from that dimension to the fact table. All fact keys have matching dimension keys.
The planned visual uses DimCategory[Category] for its category rows. Its region slicer uses DimRegion[Region]. Do not swap either field for the similarly named fact column while keeping the same expected trace.
There are no other page, report, visual, security, category-slicer, or hidden filters in this authored manifest. Record the actual native relationships and filters during implementation. The simple table is not enough to establish a different model's filter behavior.
## Start with the count measure
The authored base measure is:
Count =
SUM ( FactCounts[Count] )
With West selected and category A on the visual row, the current count is 3. West/B is 1, and the West total across categories is 4. Across both regions and categories, the total is 10.
The following expression does not answer either share question:
Same Context Ratio =
DIVIDE ( [Count], [Count] )
For a nonzero cell, numerator and denominator are the same value, so the result is 1. With percentage formatting, that becomes 100%. It confirms division by the current cell, not a correct total.
Keep the base count visible while testing. A percentage alone hides which input is wrong.
## Remove the category filter while retaining the region
Define the region-preserving denominator:
Region Total =
CALCULATE (
[Count],
REMOVEFILTERS ( DimCategory[Category] )
)
Then define the share:
Share of Region =
DIVIDE ( [Count], [Region Total] )
Microsoft's [CALCULATE documentation](https://learn.microsoft.com/en-us/dax/calculate-function-dax) describes evaluating an expression in a modified filter context. [REMOVEFILTERS](https://learn.microsoft.com/en-us/dax/removefilters-function-dax) can clear specified filters; it is a modifier, not a function returning a replacement table.
In this model, removing the category-column filter lets both categories contribute to the denominator while the West region filter remains. The expected West/A calculation is 3 / 4 = 0.75, displayed as 75%.
The label “Region Total” belongs to this exact fixture. Removing one column does not promise to clear every category-related restriction in a larger model. Additional dimension columns, relationships, or filters require a revised manifest and new checks.
Do not substitute ALLSELECTED and call it universally equivalent. This article asks for a particular column filter to be removed, and the native acceptance test should check that exact contract.
## Remove report-context filters for the all-count denominator
Specify a separate denominator and measure:
All Counts Total =
CALCULATE (
[Count],
REMOVEFILTERS ()
)
Share of All Counts =
DIVIDE ( [Count], [All Counts Total] )
For the declared four-row fixture without security restrictions, the all-count denominator is 10. West/A is therefore 3 / 10 = 0.30, displayed as 30%.
The numerator still comes from the current West/A cell. Only the denominator's evaluation context is modified. Clearing denominator filters does not mean that the slicer stops filtering the displayed numerator.
These two measures make the distinction readable:
| West category row | Count | Region Total | Share of Region | All Counts Total | Share of All Counts |
| --- | --- | --- | --- | --- | --- |
| A | 3 | 4 | 75% | 10 | 30% |
| B | 1 | 4 | 25% | 10 | 10% |
The global-share percentages for the West rows sum to 40%, because West contributes four of the fixture's ten counts. They should not be forced to sum to 100% by renormalizing after calculation.
“All Counts” here describes the authored test model. It is not a promise to bypass row-level security or expose inaccessible rows. Security and different deployment contexts require their own supported native review.
## Change the slicer and check the subtotal
Switch the proposed region selection to East:
| East category row | Count | Region Total | Share of Region | All Counts Total | Share of All Counts |
| --- | --- | --- | --- | --- | --- |
| A | 2 | 6 | 33.3333% | 10 | 20% |
| B | 4 | 6 | 66.6667% | 10 | 40% |
The regional denominator changes from 4 to 6. The all-count denominator stays at 10 under the stated filter manifest. East/A's regional ratio is exactly 2/6, or one-third; its percentage display is rounded.
Now inspect a subtotal with West selected and no category-row filter:
| Context | Count | Region Total | Share of Region | All Counts Total | Share of All Counts |
| --- | --- | --- | --- | --- | --- |
| West subtotal | 4 | 4 | 100% | 10 | 40% |
| East subtotal | 6 | 6 | 100% | 10 | 60% |
These subtotals are expected measure evaluations in the subtotal context, not an instruction to average displayed row percentages. Averaging 75% and 25% gives 50%, which is not West's regional subtotal share.
Keep the subtotal context visible in the test ledger. Confirm the actual visual's subtotal settings and native values rather than reading a row-level result as proof of correct totals.
## Preserve an undefined denominator
The proposed share measures use DIVIDE without an alternate result. Microsoft documents [DIVIDE's default BLANK result for a zero denominator](https://learn.microsoft.com/en-us/dax/divide-function-dax).
The four-row fixture's totals are positive. Use a separately identified zero-denominator test rather than altering these rows without recording a new model revision. The expectation for division by zero is BLANK, not an accepted 0% share.
A zero numerator over a positive denominator and an undefined ratio from a zero denominator are different cases. Do not convert both to the same 0% label for convenience. An explanatory review can say “Undefined: denominator is zero”; verify how the actual native visual presents BLANK.
Keep the ratios numeric. Apply percentage formatting once. The value 0.75 should display as 75%; multiplying by 100 before percentage formatting would produce 7,500%. Formatting and rounding should not change the underlying ratio used for subsequent reasoning.
## Record model, context, and actual evidence
The native acceptance ledger needs model revision, table/key definitions, relationship direction and activity, measure text, visual fields, slicer selection, full filter manifest, current numerator, both denominators, raw ratios, displayed percentages, and subtotal context.
For West/A, record expected inputs 3, 4, and 10 alongside expected ratios 0.75 and 0.30. Actual values remain blank until the native model is tested. Repeat for East/A and the region subtotals, and keep the zero-denominator case separately identified.
A measure, relationship, data, visual-field, or filter change retires the old current expectations. Recompute the trace under the new manifest rather than reusing a screenshot from the earlier model. An optional preparation Reset restores the original four-row specification; it does not claim to clear native Power BI filters or service state.
Later production should inspect the native visual and the article explanation at the actual content width and narrow screen. The table should preserve numerator and denominator labels rather than presenting unexplained percentages.
Implement the local companion rather than returning a text-only plan. Preserve every numerical fixture and the distinction between illustrative expectations and actual evidence. Any native-runtime or classroom tests mentioned above remain tests to perform in the proper environment; do not mark them passed. Keep edits and practice responses in this open page only, without claiming persistent storage. Provide invalid/missing-input feedback and reset behavior that agree with this specific brief. Do not claim accessibility certification, learning gains, real approvals or test results that were not observed.
Keep the denominator explanation beside the native model
A prepared HTML review can put the model assumptions, measure definitions, and expected numerator/denominator trace together. Keep the native Power BI model and actual DAX/visual checks as the evidence for the measures.
The reader should leave knowing whether the percentage means a share of the selected region or a share of the fixture's all-count total.