Should observers switch sides when conducting aural surveys at frog ponds?

I have been doing some research where two oberservers will stand in the middle of a frog pond facing opposite directions listening for frog calls (from 3 different species) within their half of the area. They record the number of individuals they hear calling of each species, and also their approximate location. We aim to compare this information to data collected from audio recorders paced around the pond in an array. We have done 7 surveys using this method where the observers have always stood back-to-back in the same position in the pond, facing the same direction every time. However, we are now replacing one of the oberservers with a different person. Would it be statistically better to continue the previous method of the two observers always facing the same way in the pond, or should we start randomising the direction faced/half of the pond being observed by each person? If we randomise the direction faced, would we still be able to compare this data to the previous survey data we have collected? We are probably only able to complete 4 or so more surveys (maybe 6, but this is unlikely).