Using newly developed spectroscopic models to measure the divertor concentration of Ne and Ar, it is shown that the experimental detachment threshold on ASDEX Upgrade with Ar-only and mixtures of Ar+N or Ne+N scales as expected in comparison with an analytical equation derived by Kallenbach et al (2016 Plasma Phys. Control. Fusion 58 045013). However, it is found that Ar radiates more efficiently and Ne less efficiently in the scrape-off layer than the model predicts. By separately increasing the neutral beam injection power and cutting the impurity gas flow, it is shown that the partially detached and strongly detached X-point radiator scenarios reattach in & AP;100 ms and & AP;250 ms, respectively. The former timescale is set by the core energy confinement time, whereas the latter has an additional delay caused by the time required for the ionisation front to move from the X-point to the target. A simple equation with scalable geometric terms to predict the ionisation front movement time in future machines is proposed.