Indeed. There’s a few things I’m noticing here upfront per say. It’s all a matter of design intent – the proverbial words I’m probably going to say every day because it’s just the way things are (or rather are they), pun intended.
I’ve circled a few details here where geometry is being propagated throughout the said surface, that manifest from the adjacent geometry:
So a designer must ask themselves, is this how I want the geometry to be propagated? Or is it just a deterministic sequence in an otherwise ‘unrefined’ workflow?
The sharp vertices can be a problem if one is desiring to make a ‘single’ surface, especially when there’s more than ‘one’ sharp vertice within said to be ‘single’ surface.
The adjacent ‘fillets’ aren’t as bad, relatively speaking.
Next I would look at the composition of said curves that comprise said surface.
I would consider their degrees, poly’s, kinks, paired cross sections etc.
Depending on the situation and workflow, a degree 3 surface might be inevitable in certain cases, but I know lot’s of members here really like degree 5 – so all depends on workflow, design intent, etc.
I like doing network surface most of the time. It’s too bad Rhino doesn’t network degree 5 off the bat. Some reason they’re deg-3 
to my knowledge anyway. So rebuilding them later if ‘netwrksrf’ is used, may be inevitable to alter degrees. Now I’m questioning myself, maybe I’ll have to test this theory… Always, seems like I’m everlastingly scratching the surface of Rhino’s capabilities. I should have converted to Rhino7 much sooner. Maybe I’m stuck in Rhino5 mentality.
I wonder if R8 will be better at netwrksrf’ing…
At any rate, these are the geometric constituent entity locations that catch my eye in the situation I’m perceiving:
These are the locations that catch my attention when considering revising the surface(s) in question.
There’s a couple spots I missed at first, but you get the idea.
Any time there’s a ‘kink’, ‘change in curvature’, or a ‘span’ that’s very ‘acute’ or ‘obtuse’, then possibly there can end up being some ‘trouble’.
It’s possible however, that the design intends things to be this way, and maybe all it needs is some very slight adjustment rather than complete revamp and redesign.
Some members, also really like trimming surfaces all over the place, so I guess that’s something to consider too. 
My origins with Rhino were solely based on reverse engineering the most highly complex-organic-geometries you can imagine, and I spent most of my time compiling data that had the least number of trimmed surfaces as possible, so that’s why I kinda lean in that direction so much.
Interestingly though, I don’t think ppl realize you can do both if you want, and pick one, or go back n’ forth if you have to. I might try to demonstrate this more and more over time.
@Ryan14 This is interesting. I like the crosswise span alignment. I’m sure there’s room for certain design permutations that allow for the ‘blend’ type versions. Those versions might take more time to ‘close’ up due to possible requisite to revamp some of the current geometries adjacent to the blend.
I’ll try a quick shorter version to start, and then maybe other versions later…
Here’s a first attempt. Without knowing more about the design intent, it’s the quickest thing I’d come up with. Obviously there’s many ‘naked’ edges that should be fixed before moving forward with anything else. And the design intent should probably be revised in doing so.
Mouse_emod.3dm (14.3 MB)