The winds of massive stars remove a significant fraction of their mass, strongly impacting their evolution. As a star evolves, the rate at which it loses mass changes. In stellar evolution codes, different mass-loss recipes are employed for different evolutionary stages. The choice of the recipes is user-dependent and the conditions for switching between them are poorly defined. Focusing on hot stars, we aim to produce a physically motivated, empirically calibrated mass-loss recipe suitable for a wide range of metallicities. We want to provide a ready-to-use universal recipe that eliminates the need for switching between recipes for hot stars during stellar evolution calculations. We compile a sample of hot stars with reliable stellar and wind parameters in the Galaxy and the Magellanic Clouds. Our sample spans effective temperatures from T~=12kK-100kK and initial masses from Mini~=15M_{sun}_ to 150M_{sun}_. The sample is used to determine the dependence of the mass-loss rate on the basic stellar parameters. We find that independent of evolutionary stage and temperature, the wind mass-loss rate is a function of the electron-scattering Eddington parameter ({GAMMA}e) and metallicity (Z), being in line with expectations of radiation-driven wind theory. Our derived scaling relation provides an adequate ({DELTA}log((dM/dt)/(M_{sun}_/yr))=0.43) and broadly applicable mass-loss recipe for hot stars. The newly derived mass-loss recipe covers nearly the entire parameter space of hot stars with UV radiation-driven winds and eliminates the need for interpolation between mass-loss formulae at different evolutionary stages when applied in stellar evolution models. Examples of stellar evolution calculations using our new recipe reveal that the predictions on the ionizing fluxes and final fates of massive stars, especially at low metallicity, differ significantly from models that use the standard mass-loss rates, impacting our understanding of stellar populations at low metallicity and in the young Universe.