Now, if you’re correcting white balance, what you’re doing is picking an area you want to be white (or neutral gray, also works as well), and telling the program to make it R=B=G, and the modify there rest of the image by the same operation. It makes sense that the starting values on the surface you pick aren’t equal, because your intent is to correct them.
The program will take that patch and find multipliers that will make all three channels equal, usually using G as the anchor. And then, it will walk the entire image and multiply each R, G, and B by the corresponding multiplier. And then, R, G, and B in your selected patch will be equal…