DATEADD

The dates in the filter context moved forward or back by a number of days, months, quarters or years (with a calendar, weeks too).

Group: Time intelligence. Compatible with DAX.

Syntax

DATEADD ( <dates>, <number_of_intervals>, <interval> [, <extension>] [, <truncation>] )

Parameter	Type	Description
`dates`	Dates	A column of dates of the model, such as `Sales[Month]`, or the name of a calendar defined in the model.
`number_of_intervals`	Whole number	How many intervals to move: forward when positive, back when negative. A fraction is rounded, halves away from zero.
`interval`	Keyword	`DAY`, `MONTH`, `QUARTER` or `YEAR`; with a calendar that has weeks, `WEEK` too.
`extension`	Keyword optional	With a calendar only: what a shift into a longer period takes. `EXTENDING` (when left out) adds the rest of the longer period, `PRECISE` keeps to the same place, `ENDALIGNED` moves a period's end to the other period's end.
`truncation`	Keyword optional	With a calendar only: what a shift into a shorter period does with a place it lacks. `BLANKS` (when left out) leaves it out, `ANCHORED` takes the period's last date instead.

Returns

Table. The dates of the filter context moved by the intervals, as far as the column holds them.

Remarks

Given a column, it reads the dates visible in the current filter context; in a row context, such as ADDCOLUMNS, the row becomes a filter first, so each row sees its own dates.
A date moved by months keeps its day where it can, else takes the month's last day: 31 January plus a month is 28 February. When the selection covers the column's first to last date of a month, the whole target month comes back: all of March moved back is all of February.
Only dates the column holds come back: a date the move or the period lands on that the column lacks is left out, and when none is left the table is empty. A business day moved onto a weekend is left out (the second example).
The selection must be one unbroken run of the column's dates. One with gaps, such as January and March without February, is an error, except on a complete calendar column (one row for every day from the first to the last, no times), where each run moves on its own.
To compare with the same period a year before, SAMEPERIODLASTYEAR says the same as DATEADD ( dates, -1, YEAR ); PARALLELPERIOD moves whole months, quarters or years instead.

Examples

The month before and the year before

EVALUATE
ADDCOLUMNS (
    DATESBETWEEN ( Sales[Month], DATE ( 2025, 1, 1 ), DATE ( 2025, 3, 1 ) ),
    "Liters", [Liters sold],
    "Month before", CALCULATE ( [Liters sold], DATEADD ( Sales[Month], -1, MONTH ) ),
    "Year before", CALCULATE ( [Liters sold], DATEADD ( Sales[Month], -1, YEAR ) )
)
ORDER BY Sales[Month]

Sales[Month]	Liters	Month before	Year before
2025-01-01	44635	50415	42637
2025-02-01	44831	44635	41966
2025-03-01	43593	44831	43893

Business days moved back a month

14 to 18 September 2026, a Monday to a Friday. A month back, 15 and 16 August are a weekend that the column does not hold.

EVALUATE
CALCULATETABLE (
    DATEADD ( Prices[Date], -1, MONTH ),
    DATESBETWEEN ( Prices[Date], DATE ( 2026, 9, 14 ), DATE ( 2026, 9, 18 ) )
)
ORDER BY Prices[Date]

Prices[Date]
2026-08-14
2026-08-17
2026-08-18

A selection with a gap

January and March 2025 without February: Sales[Month] is no complete calendar, so the call is an error.

EVALUATE
CALCULATETABLE (
    DATEADD ( Sales[Month], -1, MONTH ),
    FILTER ( ALL ( Sales[Month] ), Sales[Month] = DATE ( 2025, 1, 1 ) || Sales[Month] = DATE ( 2025, 3, 1 ) )
)

Result: an error.

SAMEPERIODLASTYEAR, PARALLELPERIOD, DATESINPERIOD, EDATE, CALCULATE.

Last updated 3 November 2026 · OpenBI Studio 0.3.74