A debate has arisen regarding the importance of stationary versus eruptive mass loss for massive star evolution. The reason is that stellar winds have been found to be clumped, which results in the reduction of unclumped empirical mass-loss rates. Most stellar evolution models employ theoretical mass-loss rates which are already reduced by a moderate factor of ~2-3 compared to non-corrected empirical rates. A key question is whether these reduced rates are of the correct order of magnitude, or if they should be reduced even further, which would mean that the alternative of eruptive mass loss becomes necessary. Here we introduce the transition mass-loss rate (dM/dt)_trans between O and Wolf-Rayet (WR) stars. Its novelty is that it is model independent. All that is required is postulating the spectroscopic transition point in a given data-set, and determining the stellar luminosity, which is far less model dependent than the mass-loss rate. The transition mass-loss rate is subsequently used to calibrate stellar wind strength by its application to the Of/WNh stars in the Arches cluster. Good agreement is found with two alternative modelling/theoretical results, suggesting that the rates provided by current theoretical models are of the right order of magnitude in the ~50Msun mass range. Our results do not confirm the specific need for eruptive mass loss as Luminous Blue Variables, and current stellar evolution modelling for Galactic massive stars seems sound. Mass loss through alternative mechanisms might still become necessary at lower masses, and/or metallicities, and the quantification of alternative mass loss is desirable.