It's not some kind of magical unmeasurable quality; it's just physics.
What dictates the max speed of limb tips? Materials (stiffness of a composite beam), profile (thickness, width, taper), curvature, overall mass and mass distribution (also governed by materials and dimensions). The ideal measure eliminating most possible variables is the limb tip acceleration and velocity curve in thin air when released from a constant deflection, (no string or anything) but that's damaging to the limbs.
So to remove as many other variables as possible: use a shooting machine to remove the squidgy bit (the archer), use a reference riser, string and arrow. Pick (a riser with) typical/standard limb angles, brace height, draw length, arrow mass that are all "middle of the road" of the recommended ranges. Spine? Well my gut feel is that it isn't that relevant for limb comparison purposes. To get the best possible arrow speed (highest efficiency) it should be as stiff as possible for the arrow dimensions chosen, to minimise energy lost to flex.
This gives you a comparative velocity, not a velocity you should necessarily expect in the real world. With the arrow mass and velocity you've got output (kinetic) energy. Measure draw force curve and calculate input energy (area under the curve). You can then get a figure for input energy and hence efficiency. The lost energy is in the kinetic energy imparted to the string and the limbs, any vibration in the limbs, arrow flex, sound energy of the shot, air resistance, and in any plasticity/hysteresis of the limbs or string (permanent deformation, loss of springiness, stretch of the string etc, lost in the form of heat).
In addition to the middle of the road values for arrow mass I'd also pick a high and low mass arrow shaft trying to keep everything else the same and calculate the efficiency for each. In most cases this will be the highest mass as that is able to pick up the most energy from the limbs. So to give the best representation of the potential for those limbs perhaps an average of the three would best.
If you were looking for a topic for a doctoral thesis in mechanics you could then look at deriving formulae to predict this from measurable attributes on the limbs themselves. Plug your data into simulations etc. There's probably a dozen or more of these been done by various people.
I suspect manufacturers know all this but really don't want comparison tables "out there".