Yeh, I don't know how big the difference in RDPS would be between an AST that got Ewer, Balance, Ewer, Balance, over and over for the whole raid (maximum raidbuff windows), vs one that got bad draws like, idk, 3/8 possible raidbuff windows because they couldn't find the Ewer/Spires they needed. But if we kept the cards as Single Target only, it removes that second layer of RNG, of 'OK I got the good card Spread, now I need this second card to burn to power my first card', which should make it easier to balance. This would hopefully strike a balance (heh) between people who want the cards to feel like they actually do different things, and the people who want the cards to be balanced in output so they can't get screwed by bad RNG for logs. Personally, I'd be willing to sacrifice Royal Road interactions if it meant the base cards had different effects again, I think.
If we went with eg Balance is 10% all damage, Ewer is 10% Magic damage, and Spire is 10% Physical damage, well those are all theoretically identical effects when played on the 'correct role'. Balance is a wildcard and works on anyone, Spire to physicals, Ewer to magicals. But there'd be some slight differences to consider even with this. For example, if you were able to see the future and know your 2min window would be Spire, Ewer, Balance, in that order, and your comp is MNK NIN DNC BLM, who do you use each card on? Some people might say Ewer BLM, Spire NIN, Balance MNK. But if you Balance the NIN instead, part of their 2min window is the ninjutsu, which is magic damage. So they benefit more from the all-round effect of Balance compared to the 100% physical MNK. With the idea of Arrow giving autoattack speed, it could potentially be better to give to the PLD during prog specifically, as while it'd be less damage output, it'd also mean they generate much more gauge to use on Sheltron. Shame SE removed the 'autoattack gives gauge' from other classes like NIN, it could have been interesting to have specific jobs be Arrow Priority targets in that regard.