Getting your frequencies exactly right is not going to be possible, and I think your goal when using tools like this should be to explore the general features of certain types of situations (which types of boards you bet more or less often, when you have a large check-raising range and how it's composed, which kinds of hands tend to be better for betting or checking in various situations, etc.) rather than trying to memorize or implement the exact GTO solutions to various situations. Among other things, there's going to be a margin of error simply based on the starting range you assign to your opponent, so I think trying to balance down to an exact % isn't going to be possible.
That said, yes, some sort of random number generator is the only truly random way to do it. Using suits becomes problematic because you will end up with too many or not enough flushes in some of your ranges on certain runouts, and that is definitely the sort of thing that better players will try to exploit. Likewise, defaulting to always betting because the solution says to bet 60% and check 40% is not a good solution and does not help to balance your c-betting range. In fact, it helps to unbalance both your c-betting range and more problematically your checking range. Will Tipton's Expert Heads Up NLHE is very good on the problems caused by players who default to betting with all of the hands that are indifferent between c-betting and checking and how to exploit them.