The OFFSET function returns a cell or an array a set distance away from a starting point. So for instance, my OFFSET function could start with the reference A1. I want to start in cell A1. Rows and cols then ask how many rows and columns do you want to move away from that starting point? If rows was two and cols or columns was two, then you would move two rows down to row three and two columns right to column C. That would now get you to cell C3. Now, that could be your answer, so your answer would be the value within that cell reference. Alternatively, you can then add in the optional height and width values. If your height and width were each three, three height and three width, then it would return nine cells. Starting in cell C3 as your first, it would then go down another two rows and across another two columns. So you'd have nine cells in total. What's the benefit of this? Well, you could put a sum formula before the OFFSET and it would sum up those nine cells? Alternatively, you could put an average formula before the OFFSET and it would average those nine cells. The great thing about the OFFSET function is it's flexible. It grows as your model grows. If I have three scenarios and I want to go down three rows, it will work perfectly. If I add a fourth scenario, I can then say I want to go down four rows and it will work perfectly. I don't need to change the formula. The downside of the OFFSET function, it is quite difficult to audit. You get to the OFFSET, you then have to work out where it's pointing rather than it easily showing you.