OK, let me give you guys some insights. I just released the game, so some of my comments apply to the predecessor (Tropical Stormfront) I released about half a year ago.
What Went Right1) On the technical side the game is implemented using libgdx & OpenGL ES 2.0. Using libgdx development was a breeze! I can highly recommend it.
2) I use getdown for deployment on Desktop; I had some technical problems initially but managed to eventually fix them all. I can definitely recommend it over webstart!
3) Development overall went well. Also, I had a great artist I worked with (Nick Lee) who did an excellent job; also provided some feedback regarding game-design (couldn't have done it without him). Unfortunatley, he is planning to get a full-time job soon so... :/
4) After I implemented the AI I actually enjoyed the game myself (I am aware I am biased). I believe I played the Hawaii invasion about 20x until I finally beat it - not because I had to but just because it was a challenge. Anyhow, I thought the AI turned out pretty well (biased again).
5) The game got a great rating on Android (and generelly very mixed ratings for Desktop: either good or very bad).
What Went Wrong1) I guess the biggest mistake I made was to create an RTS game. It's a lot of work, especially the real-time AI, balancing the units, animated graphics, real-time collision detection and movement, network-play (cross-platform), making it run at decent speeds on Android etc. I have been asked several times so far to make it a turn-based game. Also, I have been told several times how it wouldn't compare to C&C (mostly from Desktop players). Should have made a turn-based game/platformer instead.
2) Desktop sales (especially for Tropical Stormfront the predecessor) have been abmissal. I don't except it much different for Desert Stormfront. Android sales haven't been great either, but a lot better than desktop. At least on Android the players seem to generally enjoy the game.
3) The name of the game. Also the scenery. Not sure if the Middle East was a wise choice.
4) I should have played more RTS games beforehand. Things have change a lot since I played Red Alert.
5) I should have done a lot more testing with especially Desktop players. I think I'll try to find some testers for my next game a lot earlier on. Not just a few weeks before I am done with the game to get some insights into game controls, functionality etc. On the other hand it takes a lot of time to find willing individals for testing let alone getting useful feedback.
@appel re-gfx: I worked with "Nick Lee" (from London) - great artist. Also very responsive. I believe he used 3D Studio for the units/structures and rendered them out to 2D isometric. The tiles have been done by hand. I used Photoshop to grid them up etc.