For the reasons you mentioned I prefer to sweep with a surgical cut instead of a boost.
For bothersome high frequencies, I stop the track for a couple of seconds, to reset my ears a bit, and when I start it, I usually notice the offending frequency right away, because it really bothers me. I then drag a semi-tight notchfilter over the area where I think the frequency is, and when I hit it, it gives me an instant soothing feeling, as if I just pulled a needle out of my skin. After that I try if I can make the surgical cut a bit tighter, so I can single out the offending frequency a bit more, without losing the good stuff. Sometimes a simple bell-cut works better than a full notch too.
For low/mid frequencies Im usually a bit less drastic. What really helps me here is to imagine the mix in front of me like a painting between my speakers. I try to see where every part of the mix resides on that painting and how big it is. If I have trouble "seeing" a part of the mix, I try to figure out why that is. Usually I notice at this point that a part of another instrument is overlapping it too much. I then use a not-too-wide cut on that overlapping instrument in the area where i think it's poking over the masked instrument too much, and drag it around until the other instrument becomes clearly visible on my "painting". After that, I adjust the Q and amount of the cut until both instruments are like I want them to be. It's sometimes an almost visual process to me this way. I realise it sounds a bit weird, but try it. It might help you
Frequency training really helps for all of this by the way. For example, EarPlugins Tune is a nice and free plugin that can help you become much better at recognising frequency spikes and holes.