Fine knowledge of the inertia matrix can be crucial to achieving attitude control performance for agile satellites. Moreover, for missions that do not plan inertia calibration manoeuvres, such as the CNES Microcarb mission, the performance of the estimation is directly linked to the inertia's observability during mission guidance manoeuvres. In this context, first two identification methods are compared: the Instrumental Variable (IV) and the Unscented Kalman Filter (UKF). Then, observability metrics are analysed to identify the best range of the dataset suitable for inertia estimation. Finally, we deduced a process based on observability analyses and identification to perform the estimation.