Each new target carries with it some overhead: pointing to the target, centering it, etc. For each exposure there is also the overhead of readout and storing the data file. So it is expected that efficiency should drop as the number of targets is increased.
In order to study this effect, we looked at HIRES data. HIRES is known to have observing projects that look at high numbers of targets per night (e.g. the planet hunters), as well as programs that look at only a small number of targets. The data were screened for a total time of at least four hours (excluding weather and engineering time), since poor weather can seriously impact duty-cycle efficiency. (Weather can either decrease or increase the efficiency, by either making it more difficult to identify and center targets, or by cutting down the number of targets per night.) The trend of dropping efficiency with increasing number of targets is shown in the figure below.
A trend line was fit to the data. The y-intercept in some sense indicates the maximum efficiency that would typically be obtained by an average observer on an average night (e.g. one without excessive telescope or instrument problems). The slope represents the extra cost of per target overhead; the observer loses a little more than a quarter percent for each target observed.