During the filtration of colloidal dispersions by a membrane, pores often get clogged by the suspended particles. Knowing the shape and size of the particles that cause this clog would be a great help to membrane users since they could then choose the ideal filtering device. Microfluidic technology enables the fabrication of model membranes or filters that are transparent, which allows for measuring the particle geometrical features that deposit either at the surface of the pores or on top of the fouling layer that has already formed. However, the use of microfluidic filters have been confined to the study of clog formation at the pore scale, overlooking the influence of the dynamics of the particle deposition on the clogging process. We have recently shown that looking precisely at what is deposited and how this is captured inside the pore provides new insight into the clogging process. In particular, we have found that a minute concentration of aggregates in a supposedly monodisperse dispersions are mainly responsible for pore fouling. In this paper, we use the same imaging technique to determine the entire clogging process for different types of monodisperses dispersions under various flow conditions, DLVO interactions with the pore walls, and confinements. We show that the way clogs form is appear complex but is also quite systematic in the fact that aggregates are the building blocks of the clog. Pores are clogged by progressive accumulation of aggregates with the average size of the aggregate required to cause the blockage increasing with increasing flow velocity. This work demonstrates that particle size and shape distributions of the feeding dispersion must be determined to understand which physical mechanisms are at play during the clogging process.