Numerical models of ice‐ocean interactions typically rely upon a parameterization for the transport of heat and salt to the ice face that has not been satisfactorily validated by observational or experimental data. We compare laboratory experiments of ice‐saltwater interactions to a common numerical parameterization and find a significant disagreement in the dependence of the melt rate on the fluid velocity. We suggest a resolution to this disagreement based on a theoretical analysis of the boundary layer next to a vertical heated plate, which results in a threshold fluid velocity of approximately 4 cm/s at driving temperatures between 0.5 and C, above which the form of the parameterization should be valid.