Each trial has two measurements. Keep them on the same row before sorting or removing incomplete records so the formula compares the intended pairs.
Start with four pairs whose result you can predict
Enter 1, 2, 3, 4 in B2:B5 and 2, 4, 6, 8 in C2:C5. Use =CORREL(B2:B5,C2:C5). Every Y is double its X, so the expected correlation is 1.

Illustrated correlation check. Keep each X next to its own Y.
Test a constant column in a copy
Set all four Y values to 4. The calculation is undefined because Y has no variation. If a real dataset errors, check both ranges for spread, length, and missing observations. Use a scatter plot to spot curves and unusual points, and include the number of pairs when reporting correlation.

Illustrated constant-column check. Both variables need spread.



