Zoho Sheet RANDARRAY Function: Shape, Type and Random-Array Controls

The official RANDARRAY reference was retrieved October 15, 2026. Zoho’s February 2020 announcement includes RANDARRAY among its new functions. This supplies an announcement date rather than a numbered first build. The last functioning version is unknown: current documentation is not a last-working-build test. The web entry does not establish mobile or offline-client parity.

The worked results and constraints below were independently checked. No native Zoho session was executed. Examples follow the help’s semicolon formula notation even where its syntax heading uses commas; verify the actual editor’s accepted separators. All datasets are fictional and fully specified rather than reconstructed from hidden embedded cells.

RANDARRAY([row]; [column]; [min]; [max]; [type])

Every argument is optional. row defaults to 1, column to 1, min to 0, max to 1, and type to FALSE. TRUE requests whole numbers; FALSE requests decimal numbers.

Although the column description accidentally says “sequential array,” the function is documented as a random array. Its random values should not be expected to form a SEQUENCE-style progression.

Zoho states that results change when the function recalculates. There is no documented seed argument. Fractional dimensions, reversed limits, blanks, equal limits and invalid type coercions are outside the controls in this guide.

Enter =RANDARRAY(3;2;1;10) in a clear output area. It requests three rows and two columns, or six decimal-mode values between the stated limits. Omitting type selects FALSE; it does not request integers merely because min and max are integers.

The reference does not explicitly state endpoint inclusion for decimal mode. This guide therefore uses the conservative check 1≤value≤10, without importing RAND’s separately documented exclusive-upper rule into RANDARRAY.

The array should be accommodated as a whole. Keep a clear intended footprint and inspect how the current editor places the output before referring to specific returned cells. No exact spill-collision error, historical array-entry keystroke or native placement behavior was tested here.

=RANDARRAY(2;3;1;50;TRUE) requests two rows and three columns, again six values, with whole-number type. Each result should be an integer within the specified range.

Reversing row and column does not preserve the layout even when the total count remains six. A downstream formula expecting three rows of paired measurements should not be fed two rows of three values without deliberate reshaping.

The output is not a unique list. Random draws may repeat, and the function has no sampling-without-replacement argument. Do not assume six returned integers represent six distinct record identifiers.

The source also shows =RANDARRAY(1;3;-1;1). It requests one row of three decimal-mode values spanning negative and positive limits. The signs are not guaranteed to include both negative and positive values in a particular draw.

=RANDARRAY() uses the documented defaults and therefore requests a one-by-one decimal array between zero and one. Supplying =RANDARRAY(4;1;0;1;FALSE) instead explicitly requests four rows and one column.

These shapes are parameter expectations, not captured screenshots. A published sample grid should never be treated as the exact values the next evaluation must return.

Suppose a model expects a three-by-two array with values between 1 and 10. To test the model without random variability, manually enter this fixed table in E2:F4:

These six values are synthetic controls, not observed RANDARRAY output. Row totals are 5.75, 16 and 10. Column totals are 10 and 21.75, and the grand total is 31.75.

If the row-total formulas fail on this fixed table, the error is in the surrounding layout or arithmetic rather than in a new random draw. The same fixed fixture can be reused after editing a downstream formula.

For six whole-number values within 1–50, the total must lie between 6 and 300. A reported total of 400 is incompatible with those six values and limits. The bounds do not predict the actual total or guarantee how often a particular total will occur.

For the three-by-two 1–10 request, conservative total bounds are 6 through 60. An integer-looking displayed value in decimal mode is not proof of the wrong type: formatting may hide decimals, and a particular value can look integral. Inspect stored values before changing the formula.

When comparing two downstream calculations, use the same recorded input grid rather than recalculating random data separately for each. Verify the retained grid contains values rather than live RANDARRAY formulas. This guide does not assume a particular menu command or claim such a snapshot was created in Zoho.

The official reference does not certify cryptographic use, measured randomness quality or independence of all returned elements. It should not be relied on for authentication secrets or security-sensitive randomness.

The shared help table lists generic name, type, reference and lookup errors, but it does not establish a complete shape/bounds/error matrix for RANDARRAY. Exact invalid-dimension and occupied-output-area behavior remains untested.

If the layout is wrong, check row and column order. If integers were intended, supply TRUE explicitly. If results change, check recalculation. If a fixed regression target is needed, use the deterministic table rather than asserting a particular random result.