The median is pretty magical.
If we have data from a normal distribution, then we want to know the mean. But guess what, the median is expected to be the mean in this case.
If we have data from a log-normal distribution, then we want the geometric mean. But guess what, the median is expected to be the geometric mean in this case.
It takes work to figure out which sort of distribution we have: normal or log-normal, or something else.
It then takes work to calculate the mean or geometric mean.
Taking the median requires little work: sorting everything, and looking at the thing in the middle.