This is a talk about waveform modelling for gravitational-wave astronomy. Abstracts for such talks invariably begin with sentiments like, "Accurate models are essential to measure the properties of the black holes and neutron stars that produce our gravitational-wave observations," but with the unspoken understanding that of course such models will be available. And indeed, models successfully stayed ahead of improving detector sensitivities for most of the ~80 detections in the first three LIGO-Virgo-KAGRA observing runs. But they failed spectacularly for one of the most interesting signals in O3, and we can expect more of the same in future observing runs. I will summarise the current situation and the importance of incorporating systematics errors into future measurements.