Cluster number counts can constrain the properties of dark energy if and only if the evolution in the relationship between observable quantities and the cluster mass can be calibrated. Next generation surveys with ~10000 clusters will have sufficient statistics to enable some degree of self-calibration. The excess variance of counts due to the clustering of clusters provides such an opportunity and can be measured from the survey without additional observational cost. It can minimize the degradation in dark energy constraints due to an unknown power law evolution in the mass-observable relation improving constraints on the dark energy equation of state by a factor of 2 or more to sigma(w)=0.06 for a deep 4000 deg2 survey.