Chemotaxis is the response of a particle to a gradient in the chemical composition of the environment. While it was originally observed for biological organisms, it is of great interest in the context of synthetic active particles such as nanomotors. Experimental demonstration of chemotaxis for chemically-powered colloidal nanomotor was reported in the literature in the context of chemo-attraction in a still fluid or in a microfluidic channel where the gradient is sustained by a specific inlet geometry. In this work, we use mesoscopic particle-based simulations of the colloid and solvent to demonstrate chemotaxis in a microfluidic channel. On the basis of this particle-based model, we evaluate the chemical concentration profiles in the presence of passive or chemically active colloids, compute the chemotactic force acting upon them and propose a stochastic model that rationalises our findings on colloidal chemotaxis. Our model is also able to explain the results of an earlier simulation work that uses a simpler geometry and to extend its interpretation.