That sounds more like phenetics, which is a classification of organisms based on observable similarities. Which is also pretty much what Socionics is.
A phylogenetic tree is not necessarily based on observations, since it requires an interpretation of what the DNA means. Which is a theory, not an observation.
I think you have the cart before the horse. We make categories because we have explanations of some kind (which may only exist in our heads) in order to categorize them in a certain way. We may rely on observable similarities, or we may rely on non-observables, such as interpretations of what the DNA means. And those require theories to do so.
The point is that you couldn't possibly have come up with Darwin's theory of evolution by just categorizing similarities. No amount of categorizing would have come up with that. It required a completely new way of thinking that things gradually evolved into something different, as in, there are similarities, even though there are no immediate similarities between them. You'd have to take a guess and imagine how things were in the way back before time. This required a bold guess to make that giant leap.
It couldn't have been done with an observation, but it could be done with a theory. This is why Darwin made a guess, and not just recognize some patterns or categorize a bunch of organisms.