We present a new paradigm for the simulation of arrays of Imaging Atmospheric Cherenkov Telescopes (IACTs) which overcomes limitations of current approaches. Up to now, all major IACT experiments rely on the same Monte-Carlo simulation strategy, using predefined observation and instrument settings. Simulations with varying parameters are generated to provide better estimates of the Instrument Response Functions (IRFs) of different observations. However, a large fraction of the simulation configuration remains preserved, leading to complete negligence of all related influences. Additionally, the simulation scheme relies on interpolations between different array configurations, which are never fully reproducing the actual configuration for a given observation. Interpolations are usually performed on zenith angles, off-axis angles, array multiplicity, and the optical response of the instrument. With the advent of hybrid systems consisting of a large number of IACTs with different sizes, types, and camera configurations, the complexity of the interpolation and the size of the phase space becomes increasingly prohibitive. Going beyond the existing approaches, we introduce a new simulation and analysis concept which takes into account the actual observation conditions as well as individual telescope configurations of each observation run of a given data set. These run-wise simulations (RWS) thus exhibit considerably reduced systematic uncertainties compared to the existing approach, and are also more computationally efficient and simple. The RWS framework has been implemented in the H.E.S.S. software and tested, and is already being exploited in science analysis.