Variables
Profit Margin =
VAR TotalSales = SUM(Sales[Amount])
VAR TotalCost = SUM(Sales[Cost])
RETURN DIVIDE(TotalSales - TotalCost, TotalSales)Variables are evaluated once, where they're defined, and reused. That makes measures both faster and easier to read.
The subtle part: a variable captures the filter context at its point of definition. If you define a variable then wrap something in CALCULATE, the variable does not re-evaluate in the new context — which is sometimes exactly what you want and sometimes a bug.
DIVIDE, not slash
DIVIDE(numerator, denominator, [alternate]) returns blank (or your alternate) on divide-by-zero instead of an error. Always use it rather than /.
COALESCE(expr, 0) substitutes for blanks. Blank is not zero in DAX — blanks are excluded from visuals, which is often desirable, but breaks arithmetic when you didn't intend it.
Ranking
Product Rank =
RANKX(
ALL(Product[Name]),
[Total Sales],
,
DESC,
Dense
)RANKX needs an explicit table to rank over — ALL(Product[Name]) ranks across all products regardless of the visual's filter on that column. Omitting ALL ranks within the current context, which usually returns 1 for every row and confuses people.
TOPN(10, Product, [Total Sales], DESC) returns a table of the top 10, useful inside CALCULATE.
Iterators and comparison
SUMX, AVERAGEX, MAXX, COUNTX iterate a table with row context. MAXX(Sales, Sales[Qty] * Sales[Price]) finds the largest line value — impossible with plain MAX.
CALCULATETABLE is CALCULATE for tables. VALUES returns distinct values of a column in the current context; DISTINCT is similar but excludes the blank row added by invalid relationships. HASONEVALUE guards against showing a measure when multiple values are selected.