One missing element in discussions involving measurements is how the hearer themself would measure -- their hearing, their sensitivities, their perceptual inclinations. I’m not talking about subjective (interpretative) taste. I’m talking about how they are set up as biological and psychological beings.
In other words, even IF we were to allow that objective measurements are the only thing that matter, we still don’t have the most primitive measurement data about the hearer/experiencer themself. We have (some) measurements for only one side of the gear-listener circuit, i.e., the gear.
So, lacking the listener half of the measurement picture, we then go and see how the gear-half compares with what people subjectively interpret as what sounds good. What a forlorn enterprise.
With these caveats in mind, my approach is as follows -- use those measurements which can describe whether some bit of gear will work with another -- "Is X amp powerful enough for Y speaker?" for example. And then allow that to delimit a massive number of possibilities to a more manageable number. After that, one has to listen.