In my perfect world, GW would blow up their whole system -which I think it actually far closer to the spreadsheet formula system where x=y then people want to believe- and take a top down approach. Something like:
Consider most decent size competitive AoS events are five rounds and 2k points, that's your baseline. Create five different scenarios that are relatively balanced but slightly reward a specific type of strategy. For example, maybe a Siege type scenario or zone control will favor a tankier army a bit more. A Supply Raid scenario favors a fast elite army, so and so forth...
Now take your factions and have some number of "shells" or archetypes for each of them. You could probably break this up into Tanky, Balanced, and Glass Cannon or something along those lines. How this manifests in the game is where they get to be creative. Stormcasts tanky shell is from good armor saves and defensive buffs, whereas Death keeps their horde alive by summoning.
Now you say this shell is going to good at objectives 1 and 2, it'll be okay at objective 3, and weak at objective 4 and 5. Now Shell B can be good at objectives 2 and 3, okay at 4, and weak at 5 and 1... This is where the vast majority of balancing should take place
Then you take your pile of models and say, "How can I get the most variety of cool stuff on the table" and you start shoving it in different shells through different battleline or Hero requirements or restrictions, incentives and synergy. Whatever point cost a unit or model arrives at to fit in an archetype at this point is mostly arbitrary on a macro level, but still allows for some micro adjustments to be made, but your actual game balances comes from an adherence to defined objectives and scenarios and rock/paper/scissors/lizard/spock type approach of macro balance.