Zoho Sheet RAND Function: Bounds, Recalculation and Stable Test Fixtures

The official RAND reference was retrieved October 15, 2026. The checked references do not establish an introduction release, so the first functioning version remains unknown. 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.

RAND()

There are no arguments. Zoho documents that the value changes when the function recalculates. The reference does not enumerate every event that triggers recalculation or offer a seed argument. Do not promise a repeatable draw sequence from this syntax.

If A2 contains =RAND(), the same formula can produce a different value later. A screen capture or copied displayed number records one moment. It does not guarantee that a live sheet will retain that value after subsequent calculations.

The source does not certify cryptographic randomness, independent draws or a measured uniformity guarantee. Avoid using these spreadsheet samples to create passwords, authentication tokens or security decisions.

Put Unit_draw in A1, Scaled_draw in B1 and In_bounds in C1. Enter =RAND() in A2, =10+5*A2 in B2 and =AND(A2>=0;A2<1) in C2. Fill the three formulas down through row 6.

Each A cell is independently expected to satisfy 0≤A<1. Each B cell must satisfy 10≤B<15 under ordinary arithmetic. C’s target is TRUE for a valid draw, regardless of its particular digits.

Both derived columns refer to the stored draw in A. Writing a second RAND call in B would request a separate random evaluation rather than transform the same A value. Reusing one draw makes the relationship between columns auditable.

To test the surrounding arithmetic without relying on random draws, use a separate table with the fixed values 0, 0.25, 0.5, 0.75 and 0.999 in E2:E6. These are synthetic controls, not observed RAND output.

Enter =10+5*E2 in F2 and fill down. Exact targets are 10, 11.25, 12.5, 13.75 and 14.995. The lower endpoint can be reached when the input is zero; the upper endpoint 15 would require the excluded input one.

This fixture tests the scaling formula, not the quality of a random-number generator. It is useful for detecting an incorrect multiplier or offset before switching the input column to live RAND calls.

The formula =-2*A2 maps RAND’s domain to -2<result≤0. Multiplication by a negative number reverses the inequalities. Zero remains possible, while exactly -2 would require a draw of one and is excluded by the documented domain.

For the fixed controls 0, 0.25, 0.5, 0.75 and 0.999, expected values are 0, -0.5, -1, -1.5 and -1.998. Simply copying the positive-scale endpoint labels would get this case wrong.

Zoho’s published RAND()*-2 value is one illustrative draw. It must not be copied into a test as the exact output every future recalculation should reproduce.

Using a stored A draw, =A2<0.3 marks whether it lies below 0.3. For the fixed controls 0.1, 0.29, 0.3 and 0.9, expected flags are TRUE, TRUE, FALSE and FALSE.

This deterministic test establishes the strict comparison at the boundary. It does not guarantee that exactly 30% of a small random sample will be TRUE. Any probability interpretation also depends on assumptions about the random generator that the brief reference does not independently certify.

Keep the threshold in a labeled parameter cell when building scenarios. Recalculating the random inputs and changing the threshold simultaneously makes it difficult to determine which change caused a result to move.

If a downstream model must be compared on identical inputs, retain a separate recorded set of numeric draw values before editing its formulas. Verify that the recorded cells contain values rather than live RAND formulas. This guide does not assume a particular menu command or claim that a snapshot was created in Zoho.

A deterministic control dataset is often a better regression fixture than a random one. Keep the fixed table alongside the live scenario and label which results are reproducible exactly and which are expected only to satisfy bounds.

The entry’s shared error table includes name, type, reference and generic lookup errors. RAND has no data-range argument, so a lookup description is not a function-specific instruction. Exact behavior for invalid supplied arguments was not native-tested.

If a displayed value appears to be one, inspect additional decimal places: formatting can round a value below one to 1. A displayed value alone does not prove the domain was violated. If repeated evaluations differ, check recalculation before assuming the function is unstable or the worksheet is corrupted.