Characterizing the Explanations for Existence that Don't Work

What would an "explanation" for existence look like?

Take "explanation" to mean you start without assuming the existence of anything, and end by concluding something exists.

Here's an attempt at an such an explanation.

Okay, suppose Sally is a purple unicorn, and consider

(1) Sally is tall and shy

(2) Sally is tall

So (1) implies (2) but not the converse. But this relationship between the assertions (1) and (2) exists whether or not Sally exists. Therefore, something exists.

This doesn't assume the 1st-order properties of Sally are instantiated. The claim is there exists a relationship between (1) and (2) even though Sally is a purple unicorn. The property (1) that Sally would have if she existed is necessarily related to the property (2) that Sally would have if she existed. So the relation exists.

This attempt at an explanation for the existence of something doesn't work (I assume), no doubt for several reasons. But what's needed is a characterization of the class(es) of explanations that don't work.