One of the biggest points of balance is that the game works very hard to ensure that not only can its various classes contribute about equally, it goes out of its way to make sure that you can achieve a baseline level of power, and that optimization functions to give you additional power, but reigns in how much additional power you can get from optimization. So to begin with, in Pathfinder 2e, playing a Wizard or a Fighter isn't a decision about how powerful you are (contrasted with 5e, PF1e and so forth) although they may contribute to the party in somewhat different ways (martials are generally better at single target damage and big single hits and tankier, while casters are good at AOE and getting more smaller hits in due to failure effects or even just buffing, debuffing, and healing.) Then, even the classes that are talked about as maybe not measuring up (the Alchemist being the hardest to build and play well, and the only real example) are much closer than they would be in those other games. If you look at charts, the real differences in damage and other forms of effectiveness are not very large (though they may look it because having a small range makes the differences look bigger).
A big part of how the game accomplishes this, is that it has strict guidelines on what numerical bonuses you can get and from where, and on what does and doesn't stack. It also loads most of your power into things that you always get, like class features, and its multiclassing system doesn't compromise that core-- everyone gets the majority of their numbers this way. Then, you build a stable of feats on top of that, that provide relatively modest bonuses with a deficit of real feat taxes (in other words, things that are necessary to take to be viable-- the fact that the alchemist is the exception is part of what makes it problematic, the exception that proves the rule so to speak), when I say "real" to be clear, I mean to make the class function and contribute sufficiently to encounters, you can absolutely find yourself spending the majority of your feats to build into a certain concept (Druids are perfectly capable of being viable regardless of their feats with their full casting, if you want to keep up with Wildshaping it eats a LOT of your feats, because its not part of that core... although you could always use your spell slots to shift instead, interestingly and still force it to work.)
So as a result, if you follow the encounter guidelines, and your players broadly boost their key attribute (the game makes this very easy to do and still have a well rounded character), they will mostly all feel like they're contributing to encounters, and encounters will hinge more on their decisions and dice rolls than their ability to break the game by staring at optimization guides (while still rewarding them for their system mastery, my players optimize heavily when they organize a regular group and handle severe and extreme encounters much better than I see other groups discussing online.)
Finally, because so much of your power comes from the 'baseline' and because stacking bonuses up tall is relatively restricted (you won't be able to endlessly benefit your main thing past an investment of a few feats, if that) the game naturally encourages you to build wide, and sacking a feat on something that won't always be useful won't be as tough a decision. Similarly, noncombat feats are seperated from your class feat bucket so that you're never screwing yourself by taking non-combat capability, generally. Taking options is then both more intuitive ("oh this gives me a means of attacking with my reaction" or "Oh this makes my second attack likelier to hit" or "This gives me a use for my third action." instead of stacking up a massive combo of bonuses that all make your attack hit arbitrarily harder) and more self contained.