Proteins and protein assemblies often tether interaction partners to strengthen interactions, to regulate activity through auto-inhibition or -activation, or to boost enzyme catalysis. Tethered reactions are regulated by the architecture of tether, which define an effective concentration of the interactors. Effective concentrations can be estimated theoretically for simple linkers via polymer models, but there is currently no general method for estimating effective concentrations for complex linker architectures consisting of both flexible and folded domains. We describe how effective concentrations can be estimated computationally for any protein linker architecture by defining a realistic conformational ensemble. We benchmark against prediction from a worm-like chain and values measured by competition experiments, and find minor differences likely due to excluded volume effects. Systematic variation of the properties of flexible and folded segments show that the effective concentration is mainly determined by the combination of the total length of flexible segments and the distance between termini of the folded domains. We show that a folded domain in a disordered linker can increase the effective concentration beyond what can be achieved by a fully disordered linker by focusing the end-to-end distance at the appropriate spacing. This suggest that complex linker architecture may have advantages over simple flexible linker, and emphasize that annotation as a linker should depend on the molecular context.