Information Content (IC) analysis can be used before an instrument is built to estimate its retrieval uncertainties and analyse their sensitivity to several factors. It is a very useful method to define/optimise satellite instruments. IC has shown its potential to compare instrument concepts in the infrared or the microwaves. IC is based on some hypotheses such as the the gaussian character of the Radiative Transfer (RT) and instrument errors, the first guess errors (Gaussian character, std and correlation structure), or the linearisation of the RT around a first guess. These hypotheses are easier to define for simple atmospheric situations. However, even in the clear‐sky case, their complexity has never ceased to increase towards more realism, to optimise the assimilation of satellite measurements in the Numerical Weather Prediction (NWP) systems. In the cloudy/precipitating case, these hypotheses are even more difficult to define in a realistic way as many factors are still very difficult to quantify. In this study, several tools are introduced to specify more realistic IC hypotheses than the current practice. We focus on the microwave observations as this is more pertinent for clouds and precipitation. Although not perfect, the proposed solutions are a new step towards more realistic IC assumptions of cloudy/precipitating scenes. A state‐dependence of the RT errors is introduced, the first guess errors have a more complex vertical structure, the IC is performed simultaneously on all the hydrometeors to take into account the contamination effect of the RT input uncertainties, and the IC is performed on a diversified set of cloudy/precipitating scenes with well‐defined hydrometeor assumptions. The method presented in this study is illustrated using the HYperspectral Microwave Sensor (HYMS) instrument concept with channels between 6.9 and 874 GHz (millimeter and sub‐millimeter regions). HYMS is considered as a potential next generation microwave sounder.