As far as I'm concerned, your intuitions through play are correct. Redesigning your Alhambra is an action of last resort, rather than an action a player is likely to take multiple times per game. Quite often I don't redesign my Alhambra at all during the course of a game. If you can construct a reasonably expansive Alhambra where it is easy to add new tiles no matter what their configuration of walls, you may well not need to resort to a redesign at any point in the game.
On the other hand, just because a strategy is not that universally great, doesn't mean you shouldn't use it if circumstances dictate that it is the right one. I find that, in the early game, going all-out for walls tends to be beneficial, and if you do this you may end up in a situation where there are tiles you want to buy but can't currently place. In such cases, it makes sense to grab those tiles for your reserve and use a redesign action at some point in the second or third round to make your Alhambra easier to expand. Likewise, it would often seem foolish to pass up on a tile that you can buy for exact money, just because you can't place it. You can offset the action you lose by placing the tile in your reserve with the action you've gained by buying for exact.
Generally taking a good money card, or combination of money cards, will be the correct play, before or after buying a tile for exact price. On the other hand, I usually find that at least once per game you will be confronted with a mediocre selection of money (let's say 5 5 4 4, in unappealing colours). Such occasions give you a prime opportunity for a redesign of your Alhambra, and dump the inferior money picks on your opponents instead of yourself. Would you really rather pick up a poor money card, and open up much better picks for your neighbour? Sometimes picking up the money that is available is an actively unappealing option, and this gives you a prime chance to perform a useful Alhambra redesign action instead.
In short, most of the time an Alhambra redesign is not the optimum play, but there will usually be at least one occasion during a game where there isn't much else you need to do with your turn, making it not a bad play at all. The player who remains flexible and gives himself the opportunity to change their strategy via a well-timed Alhambra redesign will win more games, on average, than the player who fixates on a one-track strategy that can be stymied by a bad run of luck. (Case in point: I played against someone who ended the game with about 9 tiles in their reserve the other week. If they'd been imaginative enough to do an Alhambra redesign to open up their board at some point, instead of waiting for the tiles to arrive that would fit into their wall arrangement... they'd definitely have done a lot better in that game.)