Hello there, I might be this this forum's next one-post wonder, but I've got a background in Physics and energy-efficiency so I think I might be able to offer some useful advice on this thread...
Anyway, it looks like I can't post links until I've posted to 15 or more threads, so please message me if the links I was going to post sound like they might be helpful.
I think the most important thing to realize is that degree-day-based calculations are generally very approximate, as there are far too many factors that it's very difficult to account for, and that can affect results considerably. I wrote an article about the problems with degree days that I think you'd find useful. (message me for the link)
Anyway, the standard thing to do would be to correlate kWh energy consumption against degree days to determine a "baseline" consumption against which you'd compare future values. Once you have a past-performance-based formula that describes how kWh depend on degree days, you can plug in new figures to see if they show improvement compared to the "baseline". The article I referenced above describes this in quite a bit more detail.
It's all rather complicated by the fact that you have both heating and cooling going on. Again, the article above has more on this issue, but essentially you should do your best to separate out the heating and cooling consumption (ideally you'd have them both metered separately), and perhaps just ignore any periods when both are operating together. Degree-day-based calculations are inaccurate at the best of times, but they become much more so when outside temperatures are near to the base temperature of the building, and when you have heating and cooling operating on the same day.
If your heating and cooling is set up well you'll need heating and cooling degree days with different base temperatures. That's because ideally the temperature below which the heating comes on would be a few degrees lower than the temperature above which the cooling comes on (so that the two don't come on together, battle it out, and consume huge amounts of energy unnecessarily).
I'd suggest you estimate the base temperature of your building (or ideally the separately-metered areas of your building) by starting with the temperature that your building is heated to, and then knocking off a few degrees for "internal heat gains" - the free heating that you get from people and equipment. Once you've estimated the base temperature (sometimes called the "balance point"), try correlating kWh with degree days for a few base temperatures around that point. It's impossible to estimate the best base temperature exactly, but you can find it experimentally by correlating with different base temperatures until you find the best fit.
My company recently developed a free website that you can use to calculate degree days to any base temperature, and for loads of locations worldwide (particularly loads in the US). Again, please message me and I'll send you the link. This probably sounds like a plug... and it is... but to be fair I'm not aware of any other source of degree days that offers all the options you get at our site - options that are rather important for energy-efficiency-related calculations.
If you've got weekly or monthly meter readings, you can use our site to generate weekly or monthly degree day values. Or, if you have less regular meter readings, you can get daily degree days from our site, and then sum them together into blocks that correspond with the periods over which you have measured kWh values.
