POWER BI TOTALS VALIDATION CHECKLIST

Business question:
Meaning and label of the total:
Source row grain:
Entity grain at which the rule must run:
Report grouping:
Date, population, and security filters:
Missing-ID and zero-denominator policies:

[ ] Determine whether the measure is additive, a ratio, a distinct count,
    a snapshot, or a rule evaluated per entity.
[ ] Match population, grain, relationships, and filters with an independent check.
[ ] Recompute ratios from combined numerator and denominator.
[ ] Account for IDs appearing in multiple row groups.
[ ] Use an iterator only at the grain required by the business rule.
[ ] Check rounding, Top N, hierarchy levels, and existing custom totals.
[ ] Test all rows, one category, a selected customer, and no matching rows.
[ ] Test a zero denominator and missing IDs in a separate copy.
[ ] Label a sum of category memberships differently from unique customers.
[ ] Record the expected result before modifying a measure.

Four-row fixture expected values:
All lines: revenue 650; profit 365; margin 56.15%; unique customers 3;
customer-category memberships 4; category qualified revenue 400.
Hardware only: revenue 250; profit 65; margin 26.00%; customers 2;
category qualified revenue blank.
Services only: revenue 400; profit 300; margin 75.00%; customers 2;
category qualified revenue 400.
Customer A only: revenue 500; profit 300; margin 60.00%; customers 1;
category qualified revenue 300.
No matching rows: these base measures return blank.
Zero revenue denominator: Margin % is blank.
If one B ID becomes blank, DISTINCTCOUNT still counts that blank;
COUNT(DISTINCT customer_id) in SQL excludes NULL. Align the policy first.

Observed result:
Explanation of any discrepancy:
Definition agreed with the reader: