Rashba spin-orbit coupling enables electric control of spin states, promising enormous advances from conventional charge-based computing. Until now, a general scheme or a descriptor to find an optimal system with isolated spin states with large tunable splitting is still lacking. Here, based on first-principles calculations, we explore the microscopic physicochemical mechanism responsible for the Rashba effect in 2D van der Waals heterobilayers. We find that the difference in the Born effective charge of atoms at the interface can be used as a single-layer descriptor to predict heteropairs with large Rashba splitting, thus reducing the scaling factor in materials search. Moreover, we discover that for most 2D materials, the routinely used Rashba parameter alpha(R) is not a good gauge of the effect's strength. From our general scheme, MoTe2 vertical bar Tl2O and MoTe2 vertical bar PtS2, with spin splitting above 120 meV, Rashba energy E-R = 94 meV, and wavenumber difference 2k(0) = 0.36 A(-1) ("effective" alpha(R) > 1 eVA), emerge as the best candidates for spin transistors at room temperature.