Offset and gain work together. Consider the equation for a straight line: y = mx +b
This analogy is not perfect, but look at "gain" as the slope, m, and offset is b.
Starting at room temperature, I set offset to make the two readings (reference and, say, oven right) about equal. Then ran a constant temp profile to about 150 to 200°. At that point, set gain so they again match. Had to repeat once once or twice making adjustments until they matched fairly closely. Then repeat for the other oven TC.
Finally, I ran the leaded solder profile and compared the reported temp of the oven, which is an average of the two TC's with my reference put in the middle. That is where I found it helpful to increase the minimal fan speed to get better mixing.
The final profile showed pretty good match between reported temp on the screen and the desired profile on the screen up to the final rapid heat phase. At that point, reported temp lagged a little. At its peak, reported temp was about 4 or 5° low (I don't recall exactly), which was just fine for my Kester leaded solder paste.