Microbially induced calcium carbonate (CaCO3) precipitation (MICP) is one of the major sustainable alternatives to the artificial cementation of granular media. MICP consists of injecting the soil with bacterial- and calcium-rich solutions sequentially to form calcite bonds among the soil particles that improve the strength and stiffness of soils. The performance of MICP is governed by the underlying microscale processes of bacterial growth, reactive transport of solutes, reaction rates, crystal nucleation and growth. However, the impact of pore-scale heterogeneity on these processes during MICP is not well understood. This paper sheds light on the effect of pore-scale heterogeneity on the spatiotemporal evolution of MICP, overall chemical reaction efficiency and permeability evolution by combining two meter-long microfluidic devices of identical dimensions and porosity with homogeneous and heterogeneous porous networks and real-time monitoring. The two chips received, in triplicate, MICP treatment with an imposed flow and the same initial conditions, while the inlet and outlet pressures were periodically monitored. This paper proposes a comprehensive workflow destined to detect bacteria and crystals from time-lapse microscopy data at multiple positions along a microfluidic replica of porous media treated with MICP. CaCO3 crystals were formed 1 h after the introduction of the cementation solution (CS), and crystal growth was completed 12 h later. The average crystal growth rate was overall higher in the heterogeneous porous medium, while it became slower after the first 3 h of cementation injection. It was found that the average chemical reaction efficiency presented a peak of 34% at the middle of the chip and remained above 20% before the last 90 mm of the reactive path for the heterogeneous porous network. The homogeneous porous medium presented an overall lower average reaction efficiency, which peaked at 27% 420 mm downstream of the inlet and remained lower than 12% for the rest of the microfluidic channel. These different trends of chemical efficiency in the two networks are due to a higher number of crystals of higher average diameter in the heterogeneous medium than in the homogeneous porous medium. In the interval between 480 and 900 mm, the number of crystals in the heterogeneous porous medium is more than double the number of crystals in the homogeneous porous medium. The average diameters of the crystals were 23–46 μm in the heterogeneous porous medium, compared to 17–40 μm in the homogeneous porous medium across the whole chip. The permeability of the heterogeneous porous medium was more affected than that of the homogeneous system, while the pressure sensors effectively captured a higher decrease in the permeability during the first two hours when crystals were formed and a less prominent decrease during the subsequent seeded growth of the existing crystals, as well as the nucleation and growth of new crystals.