Hi Sam -
When someone asks me a similar question, I suggest that they spend an hour or two going through their old, un-culled photos while asking themselves the question, "Which of the poor shots could I have nailed if I had a brand-new whiz-bang camera?".
There is an unexpected bit of subtlity to this approach. For one thing, it is not thinking in the abstract about how wonderful life would be to have this or that in your camera. It is personalized to precisely your shooting habits. If you don't like shooting bugs and flowers, they won't show up in this search. You wouldn't have pressed the button if you didn't want to take a picture of *that* subject in *that* situation.
When I apply this evaluation technique to my own photography, I find that the areas of improvement are rarely in the category of simple landscapes, macros, or anything with a stationary or nearly stationary subject. If I needed to, I could probably still do a reasonable job on any of these situations with a tripod, a WW2 Zeiss roll film folder camera with a Tessar lens, and then scan the results.
Nor are they in the category of typical outdoor daytime touristy pix (ie, "here we are in front of x, now we're standing in front of y", etc.). I can usually nail these with a $100 point and shoot or my iPhone.
I find that what drives me to upgrade my camera are situations that absolutely require me to get a good shot on one try. For example, I shoot quite a few events for my employer. If I don't nail the photo of the president as he is lifting the spade of dirt to kick off groundbreaking for the new building, I look like an idiot. Although I don't shoot weddings as frequently as I used to, I still do the occasional one, and again, if I don't nail the key moments or screw up some of the table shots, again, I look like a complete idiot.
So, what improvements would improve my success rate for such shots? It's always the basics. In no particular order:
1) Accuracy and speed in auto focus,
2) Accuracy in metering,
3) Improved high-ISO / low light performance,
4) Reduction in shutter lag,
5) High bit depth RAW output (so I can pull detail out of the shadows with minimal noise / posterization / banding, and with better color correction in uncontrolled lighting situations).
6) A full frame sensor is fairly important to me because it makes obtaining OOF backgrounds easier, and my lenses match up better than with a crop-frame sensor.
7) Having an accurate, on-the-fly, extendible flash system is extremely important to me. I regularly shoot with one on-camera hot shoe flash and two off-camera flashes synched to, and controlled by the main flash. Personally, I have found that the Nikon CLS has been a God-send for me.
8) Finally, I would add that the user interface is supremely important to me. I want lots of individual, dedicated knobs, wheels and buttons. I simply will not accept any camera where you have to drill down through a menu tree to access any of the common shooting parameters.
A) It is *NEVER* "more features" (eg, more preset "scenes").
B) Once sensor sizes went above around 10 Mpixels, simply having more pixels isn't a big deal to me anymore.
C) It is never adding video to a DSLR. If I want serious video, I use a video cam. If I will be happy with just some sort of record of the situation, I just use my iPhone.
C) Personally, I probably wouldn't use Internet connectivity on my DSLR very often, but I can also see where this could come in handy.
D) It is currently not the "frames per second" rating since I rarely shoot sports, and even when I did, I found shutter response on a single press much more important to nailing the critical moment.
Anyway, I know that I didn't answer your exact question, but I hope my ramblings give you an idea of how I would make such a decision.
Cheers,
Tom M