Contact
12 programs
Explore the site

FLOOR.PRECISE in Excel: Apply a Mathematical Lower Multiple

Math and trigonometry Intermediate Microsoft Excel

Use Excel FLOOR.PRECISE for signed lower bounds, complete-group calculations and offset bins with explicit zero-step behavior.

Verification status: See the evidence and test scope described in this guide. How formula examples are checked

Quick answer: FLOOR.PRECISE rounds to the mathematical lower multiple, using the absolute size of significance. =FLOOR.PRECISE(-13.4,4) is expected to return −16, regardless of whether the step is written as 4 or −4.

Availability and version history

  • Introduced: Added in Excel 2010, according to Microsoft’s new-function history; the first numeric build is not established.
  • Availability: official help lists Excel for Microsoft 365, Excel for Microsoft 365 for Mac, Excel 2024, Excel 2021, Excel 2021 for Mac, Excel 2019, Excel 2019 for Mac, Excel 2016, Excel Web App.
  • Checked: October 14, 2026; official FLOOR.PRECISE reference. Examples are independently calculated expectations, not native Excel test results.

Syntax and default step

=FLOOR.PRECISE(number, [significance])

Only the number is required. An omitted significance means a step of one. For a nonzero step, the result is the greatest permitted multiple no larger than the input. Microsoft documents a zero result when the number or significance is zero. Official FLOOR.PRECISE reference

There is no third mode argument. PRECISE identifies this function’s direction convention; it does not mean arbitrary numerical precision. In particular, rounding down for a negative number moves away from zero, not toward it.

Example 1: compare positive, negative and integral inputs

=FLOOR.PRECISE(13.4)
=FLOOR.PRECISE(-13.4)
=FLOOR.PRECISE(-13)

Expected results: 13, −14, −13. The integer −13 is already on the unit grid, while the lower neighboring integer for −13.4 is −14.

A common mistake is treating floor as removal of digits after the decimal point. For positive numbers those operations can agree, masking their different behavior for negatives.

Example 2: changing the step sign does not change direction

=FLOOR.PRECISE(13.4,4)
=FLOOR.PRECISE(13.4,-4)
=FLOOR.PRECISE(-13.4,4)
=FLOOR.PRECISE(-13.4,-4)

Expected results: 12, 12, −16, −16. The neighboring four-unit multiples around 13.4 are 12 and 16; around −13.4 they are −16 and −12. FLOOR.PRECISE picks the lower member in each pair.

Use a clear positive step in a workbook parameter even though a negative one has the same documented spacing. A readable parameter is easier to audit than relying on a sign that does not affect the result.

Example 3: zero is distinct from an omitted step

=FLOOR.PRECISE(13.4)
=FLOOR.PRECISE(13.4,0)
=FLOOR.PRECISE(0,4)

Expected results: 13, 0, 0. The function’s documented zero case is not a division-by-zero error. If a zero step is invalid for your model, create an explicit validation rule rather than depending on an error to reveal it.

Advanced example: allocate complete eight-piece kits

Enter available quantities 17, 24, 25 in A2:A4. In B2 enter and fill down:

=FLOOR.PRECISE(A2,8)

Expected pieces in complete kits: 16, 24, 24. In C2 use =B2/8; expected kit counts are 2, 3, 3. In D2 use =A2-B2; leftovers are 1, 0, 1.

Total complete-kit allocation is 64 from 66 pieces, leaving 2. These sample quantities happen to give the same total whether pooled or processed separately. That is not a general guarantee.

For separate quantities of 7 and 7, each makes zero eight-piece kits. Pooling makes one kit from 14 pieces, leaving six. Choose the grouping constraint first, then apply the rounding formula at that level.

Advanced example: define lower boundaries on an offset grid

Suppose bands begin at 2 and repeat every 5 units. In E2 enter the input and use:

=2+FLOOR.PRECISE(E2-2,5)

For inputs 3, 7, 8, −1, expected lower boundaries are 2, 7, 7, −3. For −1, subtracting the origin gives −3; flooring that to a five-unit multiple gives −5; adding two yields −3.

The convention defines intervals such as [−3,2), [2,7) and [7,12): include the left boundary and exclude the right. At input 7, the lower boundary is 7, not 2. State this convention when using the results to group observations.

Advanced example: round to quarter-unit lower bounds

=FLOOR.PRECISE(10.6,0.25)
=FLOOR.PRECISE(-10.6,0.25)

Expected results: 10.5 and −10.75. A nonzero-step check can confirm that each result lies on the permitted grid and is no greater than its input. With exact arithmetic, the next grid point above the result would exceed an off-grid input.

Finite-precision inputs near a boundary deserve testing in the target workbook. Changing a display format to show fewer decimals does not change the underlying number being rounded.

ROUNDDOWN uses a count of decimal digits and moves toward zero, so its negative result differs. INT supplies a simple integer floor; FLOOR.PRECISE adds an explicit multiple.

Keep numeric source values, step and rounded result visible in separate columns. This guide covers ordinary numeric inputs and documented zero behavior; a complete native error-code or coercion grid has not been established. The cited Applies To list is reproduced as documented rather than inferring identical support on every platform.

Verification: Fictional inputs and independently calculated outputs; no native Excel execution. See the editorial testing standards.