Now I'm doing another study (I love doing studies) that is completely unrelated, but requires me to wear a Dexcom for a couple of weeks out of the month. Because they don't want me changing anything, they have blinded the CGM so that I can't see the readings. They will give me the printouts at the end of the study, though, so that I can use them to tweak my insulin doses if needed.
My impression a few years ago, when the technology was still fairly new, was that the CGM was rarely 100 percent accurate -- but then, as any diabetic knows who has tested their blood sugar two or three times in as many minutes, regular meters are rarely 100 percent accurate either. What I found the CGM most useful for was recognizing trends and catching lows (or highs, for that matter) before they happened. Whether it was spot-on or not, the CGM showed you if your blood sugar was rising or falling, so that you could use your regular meter to verify the trend, and fix it before it became a problem.
The biggest thing I'll be looking for when I get to see the downloads from the Dexcom for this study is my overnight trends. I don't really have any other way of knowing whether I am crashing overnight and just not waking up for it, so I will be glad to see what the CGM has to say about that.
For diabetics who have never worn a CGM, I highly recommend one if you can get your insurance company to pay for it! They are incredibly helpful for determining whether your dose is correct, and for protecting against crashing overnight or when you are exercising!