I suspect most hams have an inflated notion about their power output and the efficiency of their transmitter. A well constructed open wire line running without standing waves has been said to be close 98% efficient. There is always some loss in any matching network or tuner; a figure long seen tossed about in technical articles and handbooks is that the typical transmatch or tuner runs about 90% efficient (the Johnson KW Matchbox is supposed to be slightly better). Considering the similarity between the plate tank circuit in the transmitter and an antenna tuner, the tank circuit should run about the same, or maybe even less since the antenna tuner typically runs at a lower loaded Q than does the final amplifier tank circuit.
Looking at some hypothetical examples, let's say a final amplifier working through a tuner and flat (untuned) open wire line to the antenna, generates exactly 100 watts of steady carrier at the plate of the tube. The plate tank circuit losses reduce the output to 90 watts. The tuner drops that another 10%, so the input to the transmitter end of the feed line is 81 watts. Assuming the OWL is 98% efficient, about 79 watts actually make it to the antenna.
If the same final feeds the antenna through a reasonably well matched coax line without a transmatch, the figure would be close to the same, assuming the coax to be about 90% efficient. If we use a transmatch in combination with the coax, lop off another 10% of the power to the antenna.
If we also assume a nominal 75% plate efficiency at the class-C final per the tube data sheets, in order to generate 100 watts output the transmitter runs 133 watts DC input. Therefore, taking into account plate tank circuit losses, antenna tuner losses and feedline losses, the total real efficiency of that same final (comparing DC input to RF input to the antenna) is closer to 60%, and 54% if a transmatch is used with a coax feed line, and these figures are representative of the "best case" scenario of a typical ham radio configuration. I seriously question claims I have heard of class-C finals delivering rf to the antenna measured at 80% efficiency.
If the final is running in linear service, assuming a typical 60% peak efficiency at the final tubes, the calculations above give an overall peak efficiency closer to 48% (43% with transmatch working into coax). And that's just at the crest of the signal peaks, since the efficiency of a linear amplifier varies in direct proportion to the amplitude of the signal. The overall average efficiency is much lower, particularly with full carrier AM, but also with SSB whose average signal level is much lower than the maximum at voice peaks.
Still, with a good antenna, a 20% reduction in power due to network and feedline losses will have a negligible effect on the signal at the other end, so making a major issue about whether one actually has 600 or 750 watts of signal arriving at the antenna is much ado about nothing.