The efficiency plot is only one piece, like so many other specs. The efficiency of any amp goes to 0 (or below) at low to no power. It would be nice to compare the actual power consumption of comparable amplifiers, e.g. a 100 W class AB+G/H to a 100 W class D amplifier, from idle to full power. Preferably with a log x axis to highlight say the 10 mW to 10 W region where most amps probably operate most of the time. IIRC my Kill-a-Watt says my four-channel Buckeye NC252MP burns a few watts at idle where as my two-channel Emotiva class AB burns ~30 W or more, but I do not have a class G or H amp to compare (somebody want to loan me their AHB2?

)
I am not sure why the rash of "class D killer" threads. My microsecond look says the "killers" are better at HF distortion in some cases, worse at lower frequencies, and are generally less efficient but I have not seen an actual power in vs. power out plot. I am with
@pma,
@restorer-john,
et. al. that I would like more data, including actual noise levels in uV or whatever.
Other than anecdotal "I hate class D, because it's bad" posts I am not sure I have read an actual blind listening test showing comparable amps with inaudible'ish distortion are audibly different. My own tests were far too limited to be of value and were not blind. I was not expecting a difference and got what I was expecting, surprise, but expectation bias works both ways. I have no desire to "kill" any class; they all have their pros and cons, no need to kill one to highlight another ("You don't need to put out another's candle to make your own shine brighter" -- William Adam).