Ideally we look at both calculated and measured output to judge a lamp. Kind of like the grams/W thing, we really need g/W based on dissipation W and g/W based on system W to get the whole picture.
The PPFD calculation is inaccurate when we are not sure exactly what current the LEDs are running at, what bin we are dealing with, where that led falls within the bin range, uncertainty of the junction temp/temp droop and how much loss relating to optics and walls. That said the calculation is very helpful top get a general idea where we stand in terms of intensity and VERY useful to compare the potential of 2 different designs.
PAR meter PPFD measurement is inaccurate due to variations in the sensor sensitivity, shape and angle, low and varying sensitivity in the deep red range, variations in ambient temp/circulation/ventilation, variations in reflectivity of location/angle of walls, was the lamp fully warmed up after temp droop has stabilized. Apogee provides an approximate correction factor for the deep red problem, but this is really just another calculation not a measurement and it introduces several more sources or error. There would have to be a specific correction factor for a lamp like the SK which has a mix of reds deep reds and warm whites. So unless we have a calibrated spectroradiometer, PAR measurement has significant margin or error as well. It is very useful for checking uniformity, temp droop and for verifying the output is near what is expected from the calculated PPFD.
So there is a margin of error in both methods and on top of that we still do not know how much importance to assign to the 700nm+ range, which is higher for warm whites and HIDs but low for deep reds. There is disagreement regarding the PPFD requirements of 2D canopy vs 3D canopy. Uniformity and CRi/SPD can shift the efficacy of PPFD. All that said, an honest, reproducible side by side measuring grams/dissipation W can add another very useful way to compare 2 lights and that can be easily measured with very good precision and accuracy.