A new paper, written by S. A. Mewes as first author, appeared recently in JCTC: Benchmarking Excited-State Calculations Using Exciton Properties. In this work, we evaluated the possibility of benchmarking excited-state methods using a set of exciton analysis methods developed previously.
The main result is that once you look at things in detail, things can become much more complicated than you would have presumed otherwise. Just because a method provides good energies, it does not necessarily give good wave functions. In particular we evaluate the effects of electron-hole correlation, a dynamic charge transfer effect that is particularly difficult to capture with TDDFT.