To study the continuum limit of a microscopic model of gravity we need microscopic observables that have a clear interpretation in terms of continuum geometry. In general the construction of such observables is notoriously difficult. In the model of causal dynamical triangulations (CDT) it is clear what the microscopic observables are, but at present the only known well-behaved observables with a continuum interpretation are spatial volumes. In this talk I will demonstrate what it takes to go beyond these by introducing the moduli as observables for CDT in 2+1 dimensions with spatial topology of the torus. Measurements of these observables using computer simulations provide valuable clues concerning the effective action describing CDT in the continuum. In particular I will present numerical evidence indicating that the effective kinetic term is well described by a modified Wheeler-De Witt metric like the one appearing in Horava-Lifshitz gravity.