I originally chose the Motorola Razr i (no longer sold, sadly) specifically because of its great battery life. When I first got it, I was getting nearly seven days between charging, on 2+3g with no data. Even after eight months of use, that was still at 5–6 days. Suddenly, though, the battery life started to drop dramatically: by the time of Drupalcon Amsterdam, I was only getting 1.5–2 days between charging.
The phone was only about a year old, but even at the best of times warranties are a pain to deal with: and as I say, this phone is no longer manufactured. So, inspired by the Restart Project, and helped by this video, I decided to try my hand at replacing it. To my astonishment (especially given a third-party battery) it worked: battery life was back up to 5–6 days.
Sadly, this didn't last, and after the first couple of dischargings, I was back to where I was, scurrying around every two days for a charger. What was going on? My first thought was some kind of power regulator had gone in the hardware: but then why was the replacement battery OK for a charge or two, before the same discharge profile kicked in again?
Power reporting in the phone's settings interface showed nothing particularly hogging the battery. But, as a last-ditch attempt, I did what surely could not have helped: I deleted some apps that I hadn't even been using.
My phone is now back to (at least) 4–5 days between charging, and shows no sign of the battery problems reappearing. I've obviously used the apps I've kept—Twitter, Chrome—but haven't reinstalled anything that might have caused the problem. In all honesty, I can barely remember what apps I removed: this was not a scientific test, and I therefore don't want to point the finger at any app in particular. I'll only mention that many of the apps had a geolocation component (although GPS was almost always turned off on the phone, so there you go.)
I remember a while ago seeing someone tweet about how presumptuous it was that people quit applications on their mobile phone to conserve power. Their point was: why are you trusting your own method of power management any more than the mobile OS's own? Well, now I know why: if anything, within an OS where you have such limited reporting at your immediate disposal, merely quitting some apps doesn't go far enough.