borleech I think you misunderstand the purpose of offset in the first place. If you set the mix value of each transform to 100 with all offset set to 0, the transform of the constrained bone will match the transform of the target bone. If you don't want them to match, and you want the constrained bones to have their own rotation, position, or scale, you can adjust the offset. If the offset value itself can be animated, the result will be different if the mix value of each transform is 100, but since the mix itself is animatable, this means you can adjust both the base value of the mix and how much of the mix is applied to it. In addition, if the mix is less than 100, the transform of the constrained bones themselves is also taken into account, so you can see how complicated the formula gets.
To get what you want, you should have multiple transform constraints. Bones can be constrained by multiple constraints, and a bone that is already constrained by one transform constraint can be constrained by another transform constraint. This means that you can create another transform constraint with the exact same constrained bones and the target bone, but with different offset values, and then change which one you want to mix to during the animation.