You’re more expert than I. Certainly skilled play consists of reducing the search space as much as possible in as few tries as possible. And at the opposite extreme, there’s willy nilly guessing uninformed by the try-by-try feedback which amounts to nothing but luck.
But on the occasion you don’t nail that skilled play part, or the nature of the word makes that impossible, you’d be left with X number of plausible last letters, and less than X number of remaining tries. At that point it amounts to luck. Or said another way, it amounts to you rank-ordering your list of potential guesses such that the right answer occurs early enough in your list versus your remaining tries. You can call that skill in mind-reading the puzzle maker’s mind, or luck, or chance, or whatever other term you want. But it definitely is not a deterministic process at that point.
As an example with PRATE as the goal …
Someone picks CRATE as a first try. They’ve nailed 4 letters in place on their first try and now have 21 potential choices for the first position. Bravo! Well done. But what now?
All 21 choices do not form English words. But more than 5 do. Even assuming this person knows every single English word of the form _RATE with no gaps or mistakes, they’re left with having to pick 5 from a field of 21. There’s not even any a priori assurance that the likelihood of the right choice is among the 5/21 ~= 25% of the most common words. And as pointed out above by others, there are many different ideas of “most common” depending on which corpus is being counted.
So the strategy, then, is to guess words which will eliminate as many of the starting letters as possible. Just guessing again and again out of too large a field will be a game of luck. But that doesn’t mean you can’t play a skill game.
A recent one that nearly got me was “nutty.” I had all the letters by the third guess, but I wasn’t to know that. And none of them were in the correct spot.
So I guessed down to my emergency fifth word, which left me with Q, V,X, Z, and F unaccounted for. And then I stared at the game until I had run through every possible permutation of those unused letters before I smacked myself in the forehead and used two "t"s to win the puzzle. Double letters and QU combos are the ones that I occasionally forget to look for until after I have become quite frustrated.
I had gotten really lucky and got PRATE in two. My first guess was PLATE and there was literally nothing else that could fit. (Btw I would never use any Wordle guessing bot, that’s cheating.)
I suspect most folks posting here have larger than average vocabularies. That said, I have the feeling there’s a subset of brain space allocated specifically for words that are useful pretty much only for games like Wordle or Scrabble and the like. How many of your neurons are devoted solely to improving triple letter scores for use in words that have Q, J or X in them? Has the word qis popped up in your conversation recently? What’s stored in the pyx on your shelves?
That depends on whether you play on "hard mode’ or not.
With hard mode on, you have to use the letters you’ve already found, and you have to use them in places where you’ve already found them. So given your example, or the classic -OUND trap (8 valid possibles, btw) you’re stuck, and you’d better grab that rabbit’s foot keychain or four-leaf clover.
But folks who don’t play on hard mode have another option both devious and skillful. They can abandon the direct quest for the solution and instead play a “sleuth word” as we have come to call them in the main thread. A sleuth word uses a bunch of the letters that might fit in the trap, so you can reduce your X below (or to) the number of remaining tries by burning several letters with a single guess. So given your example, a sleuth word might be POINT after CRATE to check PRATE, ORATE, and IRATE with a single guess. I have a lot of admiration for sleuth word players (I stay on hard mode) because they sometimes come up with very creative sleuths to get a win and carry on a streak.