On the homogenization of means