In the context of using magnetic nanoparticles for heat-mediated applications, the need of an accurate knowledge of the local (at the nanoparticle level) heat generation in addition to the usually studied global counterpart has been recently highlighted. Such a need requires accurate knowledge of the links among the intrinsic particle properties, system characteristics and experimental conditions. In this work we have investigated the role of the particles' anisotropy polydispersity in relation to the amplitude (Hmax) of the AC magnetic field using a Monte Carlo technique. Our results indicate that it is better to use particles with large anisotropy for enhancing global heating, whereas for achieving homogeneous local heating it is better to use lower anisotropy particles. The latter ensures that most of the system undergoes major-loop hysteresis conditions, which is the key-point. This is equivalent to say that low-anisotropy particles (i.e. with less heating capability) may be better for accurate heat-mediated applications, which goes against some research trends in the literature that seek for large anisotropy (and hence heating) values.