We consider a free boundary problem for the p-Laplace operator which is related to the so-called Bernoulli free boundary problem. In this formulation, the classical boundary gradient condition is replaced by a condition on the distance between two different level surfaces of the solution. For suitable scalings our model converges to the classical Bernoulli problem; one of the advantages in this new formulation is that one does not need to consider the boundary gradient. We shall study this problem in convex and other regimes, and establish existence and qualitative theory. This is joint work with M. Gualdani and H. Shahgholian.