I like those three final steps. Ideas are much easier than refinement and iteration. This is not a slam, but an observation. We learn from each creation.

Now, if I may suggest some things, some potential lessons, perhaps, to be picked from this game.

First, I think you should look not at how you consider 2048, but how those who play it, and presumably enjoy it, do. Yes, as the game designer you should enjoy your own game, but it's not very beneficial to disregard how others look at their own enjoyment and spent time.

I guess if you don't find 2048 to be fun, or a game, then it may not be a good foundation anyway.

Second, and related to the immediate point above, I'm not exactly sure that match three would make things all that different from the 2048 situation. In classic 2048 situation, you move one thing at a time even so.

Third, let's go over why do people like 2048? It helps remember that it's a mobile game, or a small distraction. It's a puzzle. And also a game, because it is treated as such. Is it a good game? A good puzzle? A good puzzle-game? Perhaps not, but I think its goal is to give you the taste of a puzzle-game, when you have but a moment to spare. As such, it's really well designed.
But it breaks when you try to put something meatier on it. It wasn't designed for that. And the cleanest most pared down designs often suffer the most when you try to add to them.

Fourth, I agree that there are way too many things to keep track of in the game you envisioned, especially considering the framework thus described. I do find it interesting to compare it to BattleCon. But it's even worse here, because the choices are false. Or all bad. But are "All bad" decisions wrong if you got there by your own doing?

Feels to me like the scaling of the opposition should be scaled, a simple fix, that along some others would help.