When I first started, I kept lots of statistics on the problem types that I got wrong; this is what many people have recommended on this site. That just left me with a lot of stats, but I wanted something that would clearly articulate my errors and what I could do to address them.
My original idea was just to use the stats and just study a particular topic to address my weakest links. But my issue with this is that I have a lot of material to cover and sticking with one particular topic for an extended period time meant I had less time to spend on other equally important areas. For example, one weakness of mine is geometry. So I could study geometry for several days/weeks or move on to studying SC or CR, which I have just recently begun studying in-depth. My argument is that geometry is one small portion of the test while SC and CR are much larger topics.
My second idea was to keep an error log that copies out every problem I've gotten wrong, what the correct answer was, and how to answer the problem correctly. This became unwieldy for certain types of problems like reading comprehension.
My last and my current method is to keep 5 different lessons learned logs - one for each topic. For every problem I get wrong, I write down the underlying lesson for that problem. In particular, I write down what to look out for, what NOT to do, and/or what to do next time. I have found this to be more helpful because instead of spending time copying down questions/answers or studying general topics, I can focus on that exact lesson.
My current study plan covers one topic each day of the week (M-F) so every night, I bring out the lessons learned log for that topic and take notes while I'm working the problems and also when I read the solutions.