Explicit time integration schemes coupled with Galerkin discretizations of time dependent PDEs in structural dynamics require solving a linear system with the mass matrix at each time step. The repeated solution of those linear systems has long been acknowledged as one of the most expensive steps in the solution process but is further exacerbated in isogeometric analysis [Collier et al. 2012]. Moreover, the stringent constraint on the step size stemming from the outlier eigenvalues leads to an increasingly large number of linear systems. Instead of solving those linear systems “exactly”, practitioners resort to ad hoc approximations, with mass lumping being one of its best known examples. While some of these techniques have a sound theory for classical finite element methods, their extension to isogeometric analysis is an active research topic. Some appealing mathematical properties of the row-sum technique were recently unraveled in [Voet et al. 2023] and generalized to banded matrices and Kronecker products. In this talk, we further extend the framework to nontrivial settings, including general single-patch, multi-patch and trimmed geometries. By exploiting the hierarchical block structure of the isogeometric mass matrix, we first define a block analogue of the row-sum technique and define a sequence of block-banded matrices generalizing the results in [Voet et al. 2023]. We further define hierarchical lumped mass matrices as a recursive application of block mass lumping down the hierarchical structure of the mass matrix. Both methods are straightforwardly extended to multi-patch geometries and are easily adjusted to trimmed geometries. These new mass lumping techniques may significantly improve the accuracy of a dynamical simulation while retaining the simplicity of the row-sum technique.