Think of a surge protector as a variable resistor that is sensitive to voltage above a certain point. When the SPD sees voltage above the point it is sensitive, its resistance changes from what is close to being an open circuit to what is close to being a short circuit.
For the very short period of time the SPD is a very low resistance it conducts. Ohm's law says that while it is in a low resistance state, the voltage drop across it won't be real high despite a fair amount of current flowing through it for a short time (microseconds typically).
I think it makes very little difference if the SPD is placed right at the main or on a branch CB. It does appear that keeping the leads as short as possible and keeping them straight is important. The transients it protects from have high frequency components and even a small amount of inductance can add impedance to the wiring, which would result in the SPD still working but you could have voltage drop across the wiring due to the higher impedance caused by coiling up lead wires.
There are some interesting ramifications to how a SPD works. A lot of the energy of a surge would appear to actually be dissipated as heat in the wiring rather than being absorbed into the SPD as most of us tend to think happens.
Say you have a 1000V surge comes down from the POCO and hits the panelboard from L-L. Instead of 240V your have 1000V for a few microseconds. The SPD turns itself into a low resistance path. However, your current path is still your service wires and then thru your SPD to complete the circuit. It is likely the resistance in the service wiring far exceeds that within the SPD, so I^2R says most of the energy will be dissipated in the service wires.