Several authors have estimated the achievable parameter accuracy from measuring the gravitational wave (GW) signature of an inspiraling supermassive black hole binary with the Laser Interferometer Space Antenna (LISA), with the most recent work including the final merger in the calculation for nonspinning binaries by using results from numerical relativity. Separately, much work has gone into estimating the possible timescales over which an electromagnetic (EM) signature might accompany the GW signal. A coincident GW/EM measurement could be used as a “standard siren” for constraining the dark energy equation of state, analogous to using type Ia supernovae as standard candles. These estimates have predominantly assumed an initially thin alpha-disc model that has been hollowed out by torque from the binary, and have calculated the evolution in a pseudo-Newtonian potential. However, since accretion disks supported by torque from the inspiralling binary will likely become marginally thick, the actual evolution of the inner disk will likely be dominated by the magneto-rotational instability (MRI), so it has not been clear how accurate these previous estimates would prove to be. I will present our work on constraining the source location with complete inspiral-merger-ringdown GW signals, including higher harmonic signal content. I will also discuss potential mechanisms for EM counterparts to accompany the GW observation, and will present my recent work simulating the evolution of an initially hollow, moderately thick accretion disk using axisymmetric general relativistic magneto-hydrodynamic simulations, and the substantial difference in the predicted timescale for the onset of a detectable EM counterpart signal compared to previous estimates.