Numerical relativity waveforms serve as ground truth for detection and parameter estimation of binary black hole mergers. Most NR waveforms to date were extracted with an approximation that misses memory effects. I will report on the SXS collaboration's capacity to resolve memory effects in production NR simulations using Cauchy-characteristic evolution (CCE). I will further report on how BH perturbation and post-Newtonian theory furnish natural "BMS frames". With these BMS frames, we can perform precision ringdown modeling, extract well-defined remnant quantities, and build complete surrogate waveform models that capture memory effects.