There's a perception amongst FFXIV players that action/talent tree design doesn't ever work because there is always a meta pick. That's actually false.

There's a clear cut meta pick only if you can directly compare actions to each other. For example, if you were choosing between Shadow Wall and Vengeance as defensive options on an action tree, you can say that Vengeance is a better action because it has an offensive buff that the other doesn't. But what if you were comparing Shukuchi, En Avant, and the new Hell's Egress/Ingress (RPR) as movement abilities? That's a bit more difficult of a comparison to make, and player preference may start to enter the picture. In some cases, you may have choices between a marginally mathematically better option when executed perfectly and a simpler option that is much more forgiving. You still see a lot of variation in player choice when you lay out these types of options.

You only run into problems when your choices are partially homogenized, because you can do a direct comparison and a clear winner emerges. If choices are completely homogenized, then you have balance but it's boring. If you can't form a proper comparison between the choices at all, then you also have balance, but you also have variety. People are always afraid to try new things, but I really think it would be worth a trial of a simple action/talent tree to see if it works in this game.

Also, as an aside, your priorities during progression are completely different from farm. When nobody knows a fight or how mechanics works, players play a lot more defensively, because your goal is to see more of the fight to gather more data about it. So I wouldn't generalize across the two.