Hi Graeme,
You definitely need to set up an annulus of some width to measure and subtract the background. If not, as Andy says you include light from the galaxy (and light pollution etc) with the supernova in the brightness measurement. Too narrow though and the background level measurement becomes noisy. To wide (or too far out) and it becomes unrepresentative of the background where the target is. Perhaps you could try annuluses (annuli?) of different non zero widths and see how much the result varies ?