DAX interview questionsTechnical interview questions
← Interview & exam prep
Technical interview questions

DAX interview questions

32 questions from beginner to advanced, each with the correct answer and a one-line explanation of why. Study them here, or take them as a timed test.

32 questionsBeginner 12 · Intermediate 11 · Advanced 9Answers + explanations
Take it as a timed test
Free PDF
All 389 questions & answers, in one file

DAX, Excel, Power BI, Fabric, certification practice and analyst interviews — every answer with the reason behind it. Print it, study it offline.

Instant download, plus the free Tuesday briefing. Unsubscribe in one click.

Beginner 12 questions

1Which function sums an expression evaluated row by row over a table?

  1. CALCULATE
  2. SUMX
  3. ADDCOLUMNS
  4. SUM
  5. TOTALYTD
Show answerHide answer

B · SUMX SUMX iterates the table and sums the per-row expression. SUM only aggregates a single column.

2DIVIDE(10, 0) returns…

  1. BLANK() by default
  2. An error
  3. Infinity
  4. 10
  5. 0
Show answerHide answer

A · BLANK() by default DIVIDE returns BLANK on divide-by-zero unless you pass an alternate result.

3When is a calculated column computed and stored?

  1. Never stored
  2. At query time per cell
  3. Only when a slicer changes
  4. At data refresh, stored in the model
  5. On the report canvas only
Show answerHide answer

D · At data refresh, stored in the model Calculated columns are materialized at refresh and stored; measures are computed at query time.

4What does CALCULATE primarily do?

  1. Removes the data model
  2. Iterates row by row
  3. Evaluates an expression in a modified filter context
  4. Creates a relationship
  5. Sorts a table
Show answerHide answer

C · Evaluates an expression in a modified filter context CALCULATE evaluates its expression after applying/overriding filters — and triggers context transition.

5Correct way to count the rows of a table?

  1. ROWS(Table)
  2. COUNTROWS(Table)
  3. LEN(Table)
  4. COUNT(Table)
  5. COUNTA(Table)
Show answerHide answer

B · COUNTROWS(Table) COUNTROWS counts rows. COUNT/COUNTA take a column, not a table.

6RELATED works across which relationship direction?

  1. Many-to-one, from the many side to the one
  2. No relationship needed
  3. Only between two fact tables
  4. One-to-many, from the one side
  5. Only with bidirectional filters
Show answerHide answer

A · Many-to-one, from the many side to the one RELATED pulls a value from the one side while iterating the many side (many-to-one).

7COUNT(column) counts…

  1. Only text values
  2. All cells including blanks
  3. Rows of a table
  4. Only numeric, non-blank values
  5. Distinct values
Show answerHide answer

D · Only numeric, non-blank values COUNT counts non-blank numeric values; COUNTA counts non-blank of any type.

8A measure dropped into a visual is evaluated in…

  1. The first row
  2. No context
  3. Alphabetical order
  4. Row context only
  5. The filter context created by the visual
Show answerHide answer

E · The filter context created by the visual Visuals create filter context (rows, columns, slicers, filters); measures respond to it.

9SELECTEDVALUE(column) returns…

  1. The row count
  2. The first alphabetical value
  3. A table of values
  4. The single value in context, or a fallback if none/many
  5. Always BLANK
Show answerHide answer

D · The single value in context, or a fallback if none/many SELECTEDVALUE returns the one visible value, or the optional default when zero or many values are in context.

10HASONEVALUE(column) is TRUE when…

  1. There are no blanks
  2. The table has one row total
  3. The column is sorted
  4. The column has any data
  5. Exactly one value is visible in the current filter context
Show answerHide answer

E · Exactly one value is visible in the current filter context It tests whether the column is filtered down to a single value — often paired with SELECTEDVALUE.

11Which returns the maximum of a column directly?

  1. MAX
  2. MAXX
  3. RANKX
  4. TOPN
  5. LARGE
Show answerHide answer

A · MAX MAX takes a column; MAXX evaluates an expression per row before taking the maximum.

12FORMAT(0.25, "0%") returns…

  1. The text "25%"
  2. BLANK()
  3. 0.25%
  4. An error
  5. The number 0.25
Show answerHide answer

A · The text "25%" FORMAT converts a value to text using a format string — the result is text, not a number.

Intermediate 11 questions

13Context transition occurs when…

  1. You sort a column
  2. Any SUMX runs
  3. You use RELATED
  4. A slicer changes
  5. CALCULATE (or a measure) is invoked inside a row context
Show answerHide answer

E · CALCULATE (or a measure) is invoked inside a row context CALCULATE turns the current row context into an equivalent filter context — context transition.

14Which is TRUE about ALL()?

  1. It adds filters to the context
  2. It only works inside FILTER
  3. It is a time-intelligence function
  4. It always keeps the current filter
  5. It can act as a table function AND a CALCULATE filter modifier
Show answerHide answer

E · It can act as a table function AND a CALCULATE filter modifier ALL both returns a table (all rows) and, as a CALCULATE modifier, removes filters.

15KEEPFILTERS inside CALCULATE…

  1. Removes every filter
  2. Keeps only visual-level filters
  3. Disables relationships
  4. Adds the new filter to the existing context instead of overwriting it
  5. Sorts the filters
Show answerHide answer

D · Adds the new filter to the existing context instead of overwriting it KEEPFILTERS intersects the new filter with the current one rather than replacing it.

16VALUES(column) can differ from DISTINCT(column) because VALUES…

  1. Returns a scalar
  2. Returns sorted values
  3. Returns only numbers
  4. Ignores the filter context
  5. May add a blank row for referential-integrity violations
Show answerHide answer

E · May add a blank row for referential-integrity violations VALUES can include the blank row that represents unmatched rows on the many side.

17A classic running-total filter is…

  1. ALLEXCEPT(Dates, Dates[Year])
  2. REMOVEFILTERS(Dates) alone
  3. FILTER(ALL(Dates), Dates[Date] <= MAX(Dates[Date]))
  4. VALUES(Dates)
  5. FILTER(Dates, Dates[Date] = MAX(Dates[Date]))
Show answerHide answer

C · FILTER(ALL(Dates), Dates[Date] <= MAX(Dates[Date])) You remove the date filter with ALL, then re-apply a less-than-or-equal-to-max-visible-date filter.

18EARLIER() is mainly used to…

  1. Do time intelligence
  2. Return an earlier measure version
  3. Sort ascending
  4. Get the previous date
  5. Reference an outer row context in nested iteration
Show answerHide answer

E · Reference an outer row context in nested iteration EARLIER accesses an outer row context (often in calculated columns); variables are the modern alternative.

19Which pair is functionally closest?

  1. VALUES and SUMX
  2. EARLIER and BLANK
  3. ALL and KEEPFILTERS
  4. FILTER and RELATED
  5. REMOVEFILTERS and ALL (as filter remover)
Show answerHide answer

E · REMOVEFILTERS and ALL (as filter remover) REMOVEFILTERS is the explicit modern name for ALL used to clear filters in CALCULATE.

20ALLEXCEPT(Sales, Sales[Region]) removes filters from…

  1. All columns of Sales except Sales[Region]
  2. Only Sales[Region]
  3. Every table in the model
  4. Nothing
  5. The Date table
Show answerHide answer

A · All columns of Sales except Sales[Region] ALLEXCEPT clears filters on all columns of the table except the ones you list.

21CALCULATETABLE differs from CALCULATE in that it…

  1. Is always faster
  2. Cannot change context
  3. Ignores filters
  4. Returns a table instead of a scalar
  5. Only works on dates
Show answerHide answer

D · Returns a table instead of a scalar CALCULATETABLE returns a table evaluated in a modified filter context; CALCULATE returns a scalar.

22FILTER() returns…

  1. A number of rows
  2. A blank
  3. A table of rows meeting the condition
  4. A scalar
  5. A relationship
Show answerHide answer

C · A table of rows meeting the condition FILTER is a table function — it returns the subset of rows that satisfy the predicate.

23RANKX(ALL(Product), [Sales]) needs ALL() mainly to…

  1. Rank against all products, not just the current one
  2. Enable time intelligence
  3. Sort the visual
  4. Speed up the query
  5. Remove blanks
Show answerHide answer

A · Rank against all products, not just the current one Without removing the row's own filter, every product would rank 1 — ALL provides the full ranking population.

Advanced 9 questions

24Which statement about a DAX variable (VAR) is TRUE?

  1. It re-evaluates each time it is referenced
  2. It is evaluated once, in the context where it is defined
  3. It changes filter context automatically
  4. It cannot hold a table
  5. It creates a stored column
Show answerHide answer

B · It is evaluated once, in the context where it is defined Variables are lazy but evaluated only once, fixed in the context where defined — a frequent gotcha.

25ALLSELECTED() returns…

  1. Only the clicked data point
  2. An empty table
  3. Only the current row
  4. All rows, ignoring every slicer
  5. The filter context from outside the current visual (respecting outer filters/slicers)
Show answerHide answer

E · The filter context from outside the current visual (respecting outer filters/slicers) ALLSELECTED restores the context as it was outside the visual, honoring slicers and page filters.

26TREATAS is used to…

  1. Cast a data type
  2. Apply one table's values as a filter on another column (virtual relationship)
  3. Treat blanks as zero
  4. Sort a column
  5. Rename a measure
Show answerHide answer

B · Apply one table's values as a filter on another column (virtual relationship) TREATAS pushes values onto target columns as a filter, emulating a relationship.

27A circular dependency is commonly caused by…

  1. A single measure
  2. Two calculated columns referencing each other
  3. Marking a date table
  4. Using DIVIDE
  5. Using SUMX
Show answerHide answer

B · Two calculated columns referencing each other Mutually-referencing calculated columns (or measure/column loops with CALCULATE) create circular dependencies.

28Time-intelligence functions like DATESYTD require a proper Date table mainly to…

  1. Improve visuals
  2. Enable row-level security
  3. Guarantee a contiguous, complete range of dates
  4. Allow bidirectional filtering
  5. Speed up refresh
Show answerHide answer

C · Guarantee a contiguous, complete range of dates They rely on an unbroken sequence of dates; gaps break YTD/MTD calculations.

29USERELATIONSHIP inside CALCULATE…

  1. Marks a date table
  2. Deletes a relationship
  3. Removes filters
  4. Activates an inactive relationship for that calculation
  5. Creates a new table
Show answerHide answer

D · Activates an inactive relationship for that calculation It temporarily uses an inactive relationship (e.g., a second date role) for the current expression.

30SUMMARIZECOLUMNS is generally preferred over SUMMARIZE because it…

  1. Always returns one row
  2. Cannot group
  3. Is older
  4. Handles filter context correctly and performs better
  5. Ignores measures
Show answerHide answer

D · Handles filter context correctly and performs better SUMMARIZECOLUMNS is the modern grouping function with better performance and correct filter handling.

31A measure total that 'doesn't add up' usually means the measure is…

  1. Using SUM
  2. Non-additive — recomputed at the total's context, not summed
  3. Broken
  4. Missing a relationship
  5. Formatted wrong
Show answerHide answer

B · Non-additive — recomputed at the total's context, not summed Ratios and DISTINCTCOUNT-style measures re-evaluate at the total level, so they rarely equal the sum of rows.

32To ignore a slicer on Product but respect all other filters, wrap the measure with…

  1. REMOVEFILTERS() on everything
  2. ALLSELECTED()
  3. VALUES(Product)
  4. CALCULATE([Sales], ALL(Product))
  5. ALL() with no arguments
Show answerHide answer

D · CALCULATE([Sales], ALL(Product)) ALL(Product) inside CALCULATE clears only the Product filter, leaving other context intact.

What changed lately? Latest DAX news, releases & questions →

Coming soon $19 one-time · early-access price
Prep Pro: full mock exams
  • Full-length, timed mock exams in the real PL-300, DP-600 and MO-201 format
  • Score breakdown by skill, with a study plan for your weak spots
  • New questions every month, written against the current exam outline

No payment now. We'll email you once, when it opens.

More technical interview questions