In my first year on the Dexcom, I wore about fifteen sensors on my arms, exactly four on my legs, about ten on my stomach, and only two on my butt. The two I'd worn on my butt weren't stellar in terms of performance or longevity, but they did both last a week, and I figured I'd give the location another two tries.
Those two tries were my most recent two sensors. The first one was, from the get go, not so great; it kept being off and having ???s. It also started hurting after the third day. But it got to the end of the week still working, not being quite off enough to cause errors, and since I was out and didn't want to change sensors, I restarted it. It went to ??? within a few hours and I pulled it after 8 days of wear. Oddly enough, I had a brownish-purple bruise under the plastic part of the sensor.
The most recent one worked well from the beginning. It sometimes overstated or understated a low or high but wasn't wrong about whether I was high or low, had no ??? episodes, and didn't hurt, except when I caught it on door knobs (my butt is at door knob height, apparently). It was a no-brainer to try to restart it today when it hit the one week mark, but it failed to restart; the initial calibration caused a SENSOR ERROR 1 and it didn't change its mind about that even after six hours, so I pulled it. It's not the first sensor that seemed to have been working well that wouldn't restart.
My general feeling about sensors on my butt is that they are less likely to work well than sensors on my arm or abdomen but are probably worthwhile just for the sake of site rotation. Certainly, I don't mind moving the Lantus shots to my abdomen while wearing the sensors on my butt.

But thinking about the last few sensors has made me wonder about how to rate sensors on quality instead of longevity. Minimed's software uploads allowed a really easy comparison so you could see how the sensors were doing in terms of accuracy. The closest Dexcom comes to offering a quality evaluation of itself is to show how many data points I got on a given day and while that's certainly interesting, it's not enough to show quality, since a data point could be missing for many reasons other than ??? or sensor error, and a data point that's present doesn't prove it was any good.

This may also be an opportune time to point out, for those of you reading this through a feed, that I have some about me pages at the top of this blog. I had three and have just added a fourth, about the Dexcom sensors.