Soil functions, including climate regulation and the cycling of water and nutrients, are of central importance for a number of environmental issues of great societal concern. To understand and manage these functions, it is crucial to be able to quantify the structure of soils, now increasingly referred to as their “architecture,” as it constraints the physical, chemical and biological processes in soils. This quantification was traditionally approached from two different angles, one focused on aggregates of the solid phase, and the other on the pore space. The recent development of sophisticated, non-disturbing imaging techniques has led to significant progress in the description of soil architecture, in terms of both the pore space and the spatial configuration of mineral and organic materials. We now have direct access to virtually all aspects of soil architecture. In the present article, we review how this affects the perception of soil architecture specifically when trying to describe the functions of soils. A key conclusion of our analysis is that soil architecture, in that context, imperatively needs to be explored in its natural state, with as little disturbance as possible. The same requirement applies to the key processes taking place in the hierarchical soil pore network, including those contributing to the emergence of a heterogeneous organo-mineral soil matrix by various mixing processes, such as bioturbation, diffusion, microbial metabolism and organo-mineral interactions. Artificially isolated aggregates are fundamentally inappropriate for deriving conclusions about the functioning of an intact soil. To fully account for soil functions, we argue that a holistic approach that centres on the pore space is mandatory while the dismantlement of soils into chunks may still be carried out to study the binding of soil solid components. In the future, significant progress is expected along this holistic direction, as new, advanced technologies become available.