The first two LIGO and Virgo observation runs have been important milestones in the gravitational wave (GW) field, thanks to the detection of GW signals from ten binary black hole systems and a binary neutron star system. In order to fully characterize the emitting source, the remnant object and its environment, electromagnetic follow-up observations at different wavelengths are essential, as learned from the GW170817/GRB170817A case. Given the quite large localization uncertainties provided by interferometers, the main challenge faced by facilities with a narrow field of view (e.g. Imaging Atmospheric Cherenkov Telescopes, IACTs) is to setup a suitable follow-up strategy in order to observe sky regions with the highest probability to host the electromagnetic (EM) counterpart of the GW signal. As member of the EM follow-up community, the MAGIC collaboration joined this effort in 2014. As the third observation run (O3) is currently ongoing, where both LIGO and Virgo are expected to have much better sensitivities, MAGIC is refining its follow-up strategy to maximize the chances of observing the EM counterparts as soon as possible. In this contribution we will describe this strategy, focusing on the different observation cases, which mainly depends on the information available from both GW and EM partner facilities.