If you measure the TK70 at 3 meters, its too close for beam convergence, and you'll get a low cd number.
Also, remember that the ANSI ratings are based on only THREE total samples, and that LEDs range in performance as much as +/- 15%.
So, if one of us takes ONE measurement, of one light, statistically, its less likely to represent the entire model line except by chance.
A +/-15% range is therefore a way to ball park the performance, so that if a light is supposed to be 8,000 lumens, and we get 4,000 L, we can be pretty sure its not due to random batch variations...but if we get 6,800 L, or 9,200 L, we might figue that the claimed 8,000 L could be reasonable.
That means your TK70 might be in spec, or low, or high, and, so could your RC40, etc. It doesn't mean some other guy's version of your light puts out the same as yours does. Sometimes you get the runt of the litter, and some times you get the pick of the litter, etc.
At Photon Fest 18 for example, my TK70 measured ~ 2,800 L, over spec, but, some others measured were on the lower side, etc. You just need to see it as YOUR light's specs vs the MODEL's specs.