Cloud cover and relative humidity in the Tropics are strongly influenced by organized atmospheric convection, which occurs across a range of spatial and temporal scales. One mode of organization that is found in idealized numerical modelling simulations is self‐aggregation, a spontaneous transition from randomly distributed convection to organized convection despite homogeneous boundary conditions. We explore the influence of domain geometry on the mechanisms, growth rates and length‐scales of self‐aggregation of tropical convection. We simulate radiative–convective equilibrium with the System for Atmospheric Modeling (SAM), in a non‐rotating, highly elongated three‐dimensional (3D) channel domain of length >104 km, with interactive radiation and surface fluxes and fixed sea‐surface temperature varying from 280–310 K. Convection self‐aggregates into multiple moist and dry bands across this full range of temperatures. As convection aggregates, we find a decrease in upper tropospheric cloud fraction but an increase in lower tropospheric cloud fraction; this sensitivity of clouds to aggregation agrees with observations in the upper troposphere but not in the lower troposphere. An advantage of the channel geometry is that a separation distance between convectively active regions can be defined; we present a theory for this distance based on boundary layer. We find that surface fluxes and radiative heating act as positive feedback mechanisms, favouring self‐aggregation, but advection of moist static energy acts as a negative feedback, opposing self‐aggregation, for nearly all temperatures and times. Early in the process of self‐aggregation, surface fluxes are a positive feedback at all temperatures, shortwave radiation is a strong positive feedback at low surface temperatures but weakens at higher temperatures and longwave radiation is a negative feedback at low temperatures but becomes a positive feedback for temperatures greater than 295–300 K. Clouds contribute strongly to the radiative feedback, especially at low temperatures.