Let's say we're conducting an interferometry experiment. The experiment is such that we're reflecting collimated laser light off some object and then using the information contained in the reflected light from the object to perform some measurement. Does being closer to the object (say, 1 metre instead of 10 metres) produce a better signal/measurement? The answer seems obvious, but I'd like to double-check and perhaps get some details about the physics. My thoughts here are that, unless the object has some kind of perfect reflectivity and the path is in a vacuum, not all of the reflected rays (even if the beam was collimated) will be reflected at a parallel angle, and, so, there will be some divergence that will cause the returning rays to miss the detector, resulting in lower signal / inferior measurement. Furthermore, it seems to me that, the rougher the surface of the object is, the better signal/measurement we'll get the closer we are, compared to, say, a mirror, right? (Although, I suspect this last point is confusing scattering with reflection, right?)
It really depends on what you want to do. Generally speaking, the farther you place the detector, the lower the amplitude you will measure of the scattered signal, and therefore you would want to detect the signal closer to the scatterer to get a better signal to noise ratio. On the other hand, perhaps you are going to analyse your measurement with some far-field theoretical solutions (approximations of the scattered signal far from the scatterer) and in this case you don't want to detect too close because your model no longer applies.
As for the angles, the detected signal changes depending on the angle in which you place your detector (the angle with respect to the incident signal and the scatterer). I would guess that the back-scattered signal is usually the one that contains for information about the scatterer, but this is only a guess and this probably depends on the characteristics of the region of the surface where the incident signal meets the scatterer.