7 diagnostic queries to run before any AI-generated DAX goes live.

AI writes plausible DAX. These queries tell you if it's actually correct — before it lands in a report and causes silent damage.

• Q1
  Grain mismatch

  COUNTROWS vs DISTINCTCOUNT. The most common AI error — inflated numbers that look correct.

  FAULTY: COUNTROWS('SalesLine') FIXED: DISTINCTCOUNT([OrderID])
• Q2
  Filter context

  ALL vs REMOVEFILTERS. The query that tells you if your % will ever sum to 100.

  FAULTY: ALL('Product') FIXED: REMOVEFILTERS([Name])
• Q3
  Context transition inside iterators

  Why every row returns the same number — and how CALCULATE fixes it.

  Most confusing: CALCULATE( COUNTROWS )
• Q4-7
  Relationships, BLANKs, time intelligence, non-additive totals

  Each with a diagnostic query and corrected DAX. Run them once — catch the class of error for good.