CORREL in Zoho Sheet: Pearson Correlation with Worked Examples

Use CORREL in Zoho Sheet when you want one number describing the direction and strength of a linear association between two sets of paired observations. If your numeric activity measures are in A2:A7 and qualified leads are in B2:B7, enter:

=CORREL(A2:A7;B2:B7)

For the six-row practice dataset below, the independently calculated expected value is approximately 0.965648520. A coefficient near +1 indicates a strong positive linear association in these observations; near -1 indicates a strong negative association; near zero means little linear association, not necessarily no relationship at all. Zoho describes the function as returning the Pearson correlation coefficient: official CORREL reference (checked 2026-10-12).

Validation disclosure: The formulas and numbers on this page have been checked against documentation and independently calculated using a mathematical model. They have not been entered or executed in native Zoho Sheet. Treat the numeric outputs as reproducible expectations pending native testing.

Zoho’s documented syntax is:

CORREL(x; y)

There are two required arguments and no documented optional parameters or defaults in the Zoho CORREL guide. Zoho’s English help examples use a semicolon between arguments. When working in another interface locale, check the formula editor rather than assuming another separator works.

The two ranges represent corresponding observations: row 2 in column A is paired with row 2 in column B, and so on. A correlation value is not a slope, a percentage change, a probability that one variable causes another, or a guarantee about future observations. Do not mistake two upward-moving time series for evidence of causation.

Availability and version scope

The function is listed in Zoho Sheet’s official statistical-function help as of October 12, 2026. Neither its individual function page nor the Zoho Sheet product change log establishes an original release date, a minimum account plan, or identical behavior across browser, mobile, offline and desktop editions. Those details are not independently verified here. The existence of an official cloud-help page is not evidence of an execution test on every platform.

Create a blank Zoho Sheet, enter the headings in row 1, then type the six numeric rows exactly as shown.

1. Select a blank cell such as F2 and enter =CORREL(A2:A7;B2:B7).
2. The model-predicted result is 0.965648520 (rounded to nine decimal places). This means the two variables in these six rows are strongly and positively associated linearly.
3. In F3, enter =CORREL(A2:A7;C2:C7). Its model-predicted result is -0.965648520; column C deliberately reverses the trend because every value is 35 minus its corresponding column-B value.
4. In F4, enter =CORREL(A2:A4;B2:B4). The model-predicted result is 0.654653671, showing that a three-row subset can tell a different story.
5. In F5, enter =CORREL(A2:A7;A2:A7). With varying valid numbers, correlation of a variable with itself is mathematically 1.

Why these predictions are reproducible: For the six full pairs, the means are 7 (A) and 17.166666667 (B). The sum of the paired deviation products is 77, and the sums of squared deviations are 70 and 90.833333333 respectively. Pearson’s r = 77 / SQRT(70 × 90.833333333) ≈ 0.965648520. These numbers were checked independently, not observed in the Zoho application.

For further context, Zoho’s PEARSON help defines the same type of coefficient. It does not settle every edge-case equivalence between the two implementations, so prefer one documented function consistently and separately check unusual inputs.

A single high correlation can hide changes over time. Instead of computing one number for the whole dataset, calculate short-window correlations:

=CORREL(A2:A5;B2:B5)
=CORREL(A4:A7;B4:B7)

With the sample dataset, the first four rows give approximately 0.877876225, while the last four rows give approximately 0.989070710 by independent calculation. The overlapping windows are for illustration only; with four points, each estimate is sensitive to one observation. For real work, choose windows before looking at the results, use more observations, and inspect a scatterplot for curved relationships and outliers.

A second diagnostic is the squared correlation. Zoho separately documents RSQ as the square of the Pearson coefficient:

=RSQ(A2:A7;B2:B7)

The model-predicted result for the dataset is 0.932477064. Squaring loses the sign: an equally strong negative correlation also has a positive squared value. In a simple ordinary least-squares linear regression with an intercept, this is related to explained variation; it is not proof of prediction accuracy on future data or of causality. The Zoho RSQ documentation was also checked on 2026-10-12.

Correlation versus covariance and regression slope

The Pearson coefficient is unitless and constrained mathematically between -1 and +1. Covariance retains the units of both variables multiplied together, and a regression slope states a change in predicted Y per unit of X. These answer different questions. Compare the COVAR guide with Zoho’s COVAR reference, and see the verified SLOPE reference if you need a regression coefficient rather than an association index.

Zoho’s own CORREL reference includes generic #NAME!, #VALUE! and #REF! explanations plus the CORREL-specific empty/length conditions; consult the source error table. It is deliberately not claimed here that every generic example was reproduced with CORREL.

Good hygiene: Pair observations in the same rows, remove accidental mixed types from both columns and rerun with a small clean control dataset. If a time series contains a trend, explore detrending and plotting the pairs; high Pearson correlation by itself is not a forecast. A small sample or one extreme point can dominate the coefficient.

The mathematical coefficient is standardized, but the cited documents do not demonstrate that all three engines pair incomplete rows, coerce data types, propagate errors or handle every locale identically. Spreadsheet locale settings can affect separators; the table shows the syntax printed in the linked English references, not a universal rule.

For variability, see STDEVP and VARP. Additional official background: Zoho PEARSON, Zoho RSQ, Zoho SLOPE and Zoho statistical function catalogue.

Evidence last checked: 2026-10-12. Official guide status: documented; independent numerical checks: completed; native Zoho Sheet tests, mobile/desktop parity checks, screenshots: not completed.