The clue is in the very simplistic question in the title, both!
But it's far too easy to get drawn down either path exclusively. Why we should use testing seems pretty obvious but there's real benefit in blind trials of things, one you've weighed up the downsides.
It's very rare indeed to be able to tell from testing, for certain, if something will harm or help.
There is no way to test to see if someone will respond incredibly well to B1. None. Overton has done all the various assays with many, many patients and they are just not predictive (there's a good reason for this, which is essentially that levels can vary between tissues even within the same organ). He stopped using them. I didn't start taking B1 because I thought it would help anything in particular; I just had some and figured I'd throw it in and see. It's been by far the biggest effect of any supplement I've ever taken.
My genetics suggest I'm very sensitive to methyl donors. Yet I do very, very well on really high doses of methyl B12 and methyl B9.
Ammonia testing is almost useless: Levels fluctuate across minutes. If you get a high reading, yes, you've got high ammonia. Now you know. But that same person could get a low reading half an hour later. The only way to see if reducing ammonia is needed is prospective supplementation.
Years ago, I started taking Vitamin D and within two weeks my chronically stiff calf muscles that I'd sit and massage every evening finally relaxed. I could also stand on a balance board that raised my toes without bending my knees to compensate for my tight calves and I'd never been able to do that. Could I have tested? Yes, I did eventually, but that could also have held me back from a simple and safe trial.
I had no indication that Vitamin E (Tocotrienol form, specifically) might be especially good for my son but it certainly seemed to help cognition rapidly.
Etc, etc, etc.
But notice I'm just mentioning the successes. Obviously, when things got worse, I stopped.
But maybe not so obviously, when nothing happened, I generally stopped too.
Generally, because sometimes there are things which have a long-term effect that's hard to discern, but that I'm convinced by the literature have enough value to continue with. I've never seen a discernible effect from Molybdenum, but I'm convinced that it's safe enough and is very likely to be helping his sulphur/ammonia processing issues.
Does that mean that testing is useless? Of course not! Testing is vital too! It's effectively the other half of the picture from the "try it and see" half.
But doing either one alone is making life very hard for yourself indeed.
If you just stick to what test results tell you, you can miss vital parts of the puzzle, hence the above examples.
But the same is true of testing; unless you actually check how a gut microbiome is set up, you can't know which things to feed and which things to kill. Unless you have a plasma blood test (the most accurate way to test zinc) to show zinc/copper balance, you can't know what it actually is and whether it needs high dose zinc to correct it. Etc, etc, etc!
My point here is that not having testing available, and even relying too heavily on test results (see the note on Ammonia above), is often not a reason to trial something to see what the actual effects of it are.