In this talk I will elaborate on a certain class of Newton-Okounkov bodies that one can associate to "nice" compactifications of cluster varieties. In particular, I will explain how this approach recovers Rietsch--Williams' construction of Newton--Okounkov bodies for Grassmannians. In order to make the precise connection it will be necessary to explain how the Marsh--Rietsch potential and the Gross--Hacking--Keel--Kontsevich potential for Grassmannians are related. Finally, I will draw some consequences from this relation such as an isomorphism of the toric degenerations obtained by Rietsch-Willimas and the toric degenerations obtained by the celebrated "principal coefficient" construction. Time permitting, I will briefly elaborate on the interpretation of these results from the viewpoint of the representation theory of the associated dimer algebra.