Luminescence studies are used to investigate the local properties of various light-emitting materials. A critical issue of these studies is presented that the signals often lack all advantages of luminescence-studies of high locality, and may originate from an extended spatial region of even a few millimeters in size or the whole sample, i.e., places other than intended for investigation. This is a key problem for research and development in photonics. Due to this nonlocality, information indicating defects, irregularities, nonuniformities and inhomogeneities is lost. The issue refers to typical structures with a strong built-in electric field. Such fields exist intentionally in most photonic structures and occur unintentionally in many other materials investigated by applied physics. We reveal [using test samples prepared with focused ion beam (FIB) on an AlGaAs/GaAs laser heterostructure with an InGaAs quantum well (QW)] that nonlocality increases at low temperatures. This is contrary to the widely expected outcome, as low-temperature luminescence measurements are usually assumed to be free from disturbances. We explain many effects observed due to nonlocality in luminescence studies and prove that separation of the investigated area by focused ion beam milling is a practical solution enabling truly local luminescence measurements. All conclusions drawn using the example of cathodoluminescence are useful for other luminescence techniques.