You'll need a couple known points on your print at the least. If you don't have any good existing points to measure from you'll have to get survey to give you a few stakes to work from. Work from your known existing point, be it a building, or a drain, or whatever, and measure out to your curb, etc that you're trying to locate. You can measure from two points to double check. You can also use a triangulation method when measuring off two points.
Say you're trying to locate a certain new catch basin to be constructed. You can scale a measurement off perhaps the existing building, and a measurement off a light pole or any other fixed point that exists on the print. You want the two known points and the one you're trying to find to be in a triangle, ideally with the two knowns not being very close to each other. Make an arc at the appropriate distance from each known point, and their intersection will be the point you're looking for.
Islands can often be done easily because they're generally parallel to a road, sidewalk, or each other. Sometimes pulling a rope is exactly how to locate certain things.
Regarding elevations, you're going to have your absolutes that cannot be changed. This would include the road that your entry ties into, the building height, I'm assuming the pump/slab around the pump island would have a non-negotiable given height (I've never done a gas station). You will be looking at your catch basins or runoff locations and comparing them to the area that you need to drain. Now do you have an actual grading plan from a competent engineer that shows you elevations throughout the work area, including high points? Or do work with engineers like I always end up with.. they draw a circle and say "grade area to drain." When I show up to grade something I'll look over all these things. Determine if the plan is actually helpful, look at all my low points, high points, fixed points, see if anything needs to be changed. You should go per plan if at all possible, but you also need to verify in-field if that plan actually will work. Few years back I had a plan telling me to place a catch basin 2' above existing grade in an area that was not to be raised. Confirmed with surveyor (who happened to be on site for another part of project) that my math was correct, made the decision to place the basin at an elevation that would actually work.
There are going to be very critical portions of your grading, to me these are high points, flow lines, and any areas of unavoidably slow pitch. I'm not sure how big these gas stations are you're doing, but most in my area don't contain a huge area of grading for asphalt. As such, I will keep my grade stakes to a minimum. I would stake out my island locations, using the locating stakes for elevation as well. I would probably also stake the outer perimeter of my paving, for location and elevation as well. If you have any transition points in your grade you'll want to stake them as well. This being any high points or swales/flow lines.
Some of this comes down to having an eye for grade, having experience, and having a little creativity. You have to recognize where these high points and swales need to be constructed. You'll have to generate an elevation for them. Stake those important points first.
I'm sure I'm rambling here, as it try to type this out while also entertaining my daughter lol. I hope this is somewhat readable and helpful. Please ask any questions and I can try to explain better or elaborate. Could you maybe supply a print or portion of one, and I could give you my thoughts on how I would lay out or set my grading points? I feel like this is certainly something easier shown, if we had an example I might be able to explain better.