We investigate the consequences of the recent proposal that the cosmological vacuum energy of a quantum field should be computed as the difference between the naive result in FLRW and the result in a reference geometry, Minkowski space-time. Firstly, this subtraction scheme eliminates the old cosmological constant problem by definition.

Secondly, we find that the remaining vacuum energy followes the evolution of the total energy density of the Universe if and only if the vacuum energy is separately conserved. This means that if the vacuum doesn't interact with any other component, its cosmological effect can be reabsorbed into the definition of Newton's constant and, as a consequence, the vacuum energy has no cosmological effect.

On the other hand, if there is an (effective) interaction between the zero-point fluctuations and some other field, they cannot be absorbed into a renormalisation of Newton's constant and have a cosmological effect. If an ultralight scalar field with a mass smaller than the Hubble rate is present, such an effective interaction is a natural consequence. We construct a simple implementation of these ideas where an additional dark energy component interacts with the zero-point fluctuations. We see that this naturally leads to an "early dark energy" scenario where the total dark energy cannot be neglected in the early Universe and during matter domination. We test this model against CMB, SNe and BAO data: the results are consistend with the cosmological constant model, however, our toy model is not ruled out yet.