Most guesses are wrong. But you don't really have a choice: if you don't know how to do the problem, then what else can you do but guess and move on? :)
In general, you have two limited resources that you are managing throughout the test: time and mental energy. You will not have enough time or mental energy to do every last problem to the fullest extent, so you have to choose where to spend that time and energy.
You don't want to rush on things that you know how to do / know how to get right. You also don't want to spend 4 minutes getting a single question right, because you're just guaranteeing that you're going to run out of time or have to rush elsewhere (which usually results in careless mistakes).
So for roughly the first minute on any question, you are trying to do it, but if you're around the 1min mark and you don't know what you're doing, then spending even more time and mental effort on this problem is probably a bad idea. At most, you're supposed to spend about 2min (on quant and CR), so you've already used at least half of your time. If you're lost now, you're probably not going to figure it out and have time to do all the work with half the time (or less) left.
As a general rule, it's okay to go up to about 30 to 45 seconds over the average time for that type of question (because you will have some other questions that you answer more quickly). Beyond that, you should cut yourself off, though. If you need that much time, then something's already wrong (because there is a faster solution, but you don't know it), plus you're using up time and mental energy that you're going to need later in this section and test.
There are certain types on which I nearly always guess because I'm bad at them (like 3D geometry or really hard combinatorics). It takes too much brain energy and I'm too likely to make mistakes, so it's not worth it.
There are also certain types on which I'm immediately skeptical, because they are 'time wasters.' Roman numeral questions are a good example: they're asking me to solve 3 problems to solve one question. That sounds like a bad bargain to me. :) Unless the problem's pretty straightforward, I'm probably not going to bother.
Read this:
https://www.manhattangmat.com/blog/inde ... lly-tests/That mindset will help you make choices as you work through the test.