How do you decide when a groundwater model is calibrated enough?

I keep running into the same question in my own modelling work and with students: at what point do we stop tuning and call a model calibrated?

Everyone seems to have a different threshold, and reviewers often push for more.

I'd like to hear how others balance fit statistics, independent validation data, and the purpose of the model.

What's your practical rule for deciding a model is good enough to publish or use for decisions?