I'm trying to assess a particular UI element and wanted to see if there's a particular barometer for how good a UI element is (without really having the basis of a good comparison, like an A/B test). Is it sufficient to say a UI element is significantly important if 50% of users who visit the page in a single day click on it? (It's just a button). Is 50% too much?
Yes and no at the same time.
The standard is better or worse than a comparison of usability against previous iterations or parallel designs. Example: "Does this design work better than that design? Why or why not?"
Thus we see the fuzzy logic of qualitative testing.
The first step is to break down a design as to what problem it solves, and establish the variables upon which you'll be comparing designs. "Elegance" could be a variable, as could "ease of use". A/B testing could help determine which is "better" or "worse", as could basic qualitative feedback from user testing.
However, since you said without A/B testing, you can consult general design theory when looking at your design. Doing a one-off analysis of your design could help. However, I implore you to take the A/B Testing route, or any kind of comparative-analysis.
"How Good" is a matter of opinion. Your 50% mark is arbitrary to a large degree. There are buttons all over the UX-SE site that I don't click, which isn't a good mark of how usable they are; but I know that if I wanted to Share a post, I'd know that I could (although I haven't yet done so).
Long story short, do A/B testing or any kind of qualitative testing and analysis.
First, "goodness" of a UI element sounds a little fuzzy. You need some verifiable criteria to test the "goodness".
Then you apply term "significantly important" to UI element. Is it the same as "goodness" for you? I think this is you who define "importantness" of the control for your business, not a user.
Finally, you are going to apply click-through rate to a UI element, which is used for "effectiveness" measure. Pay attention, for banner ads:
Having more strictly defined "goodness", you have a lot of sources to measure (quantitative) and assess (qualitative) it.
How good is a can of Pepsi?
How good is impossible to answer
From a researcher point of view "how good?" is an impossible question to answer, at least while there's no specific criteria to look into and some comparisons to be made.
Goodness is not quantifiable on its own (unlike size, for instance). It is relative, personal and contextual. What's good in one context may not be in another (a glass of coffee will be really good for some in the morning, unless pregnant and trying to sleep after a sleepless night).
You can conduct some qualitative research into this (ask users), but in a quantitative sense, picking a number such as 50% is completely arbitrary.
The Multivariate Approach
What you may do is a multivariate testing (with and without the element) to see how much more (or less) business goals are met. But that is obviously impossible if the button is "Order Now".
There is no such thing as a good standard UI element.
Why do I say that? Because each interface varies from another because of the application type, what you're trying to achieve and what the user goals are.
How do you figure this out?
Here are things I would highly recommend:
User testing is your friend. But, user testing is also a hefty process in itself. Here is a great resource for that: http://www.nngroup.com/topic/user-testing/
Don Norman also speak about elegant design as a whole with his TED talk which I highly recommend you watch. It's phenomenal and I love this guy (he's very easy to listen to and explains everything in a simple and friendly way): http://www.ted.com/talks/don_norman_on_design_and_emotion.html
Best of luck