The Unity devs are actually pretty good in general, it's just that whatever 4-digit dollar figure they're getting from individual companies with 2 seats (like us) just doesn't warrant them paying attention to edge cases that only we hit (we're doing a lot of our own thing; we weren't looking for an engine to mold our style, we were looking for something that was deployable and would let us do what we were already doing). And if it's a bigger issue that a bunch of people are hitting then they do try to fix it. Certainly the current version of Unity is quite an improvement over what they had when we started on it. But most of the improvements are for their main audience, not a couple of dudes making strictly-2D games.
The main problem with the old engine (which I miss dearly) is that it required .NET. Therefore it could take 1-2 hours to install the game on some systems that didn't have .NET (SlimDX too, but it was less of an issue). Further, it's not all that uncommon for folks to have messed up .NET installations, and since .NET is so deeply embedded into the OS there were more than a few customers who could not get AIW to run properly (or at all) until they actually reinstalled their OS.
Unity games, on the other hand, can be run off a jumpdrive and zero installation (we do actually have an installer, but all it really does is copy files).
Various graphical tricks that we use
all the time now (like UV animation, etc) are also really easy in Unity, but were really hard/impossible in our old engine.
But in general, if Microsoft really cared about making .NET apps deployable in a reasonable fashion, we might still be using it