Where did my time go?
Well, right now, it's going into writing this page. But tomorrow, all I'll remember is that I didn't get much done today!
A solution I've used in the past — and am getting very close to doing so again — is time sampling.
As an electrical and software engineer, I was very familiar with the concept. For more than you ever wanted to know, look for the works of Joseph Fourier, Harry Nyquist, and Claude Shannon, who evolved the concept over the past 100 years or so into the generalized science of information theory.
"But how will giving me a bunch of new links to waste time on help me make better use of my time?"
Oops. You caught me in a digression. Sorry!
 Time sampling
The basic theory is that if you sample something at regular intervals, you have a powerful dataset for later manipulation and analysis.
So, first get yourself a timer. A watch with a count-down function works fine, so does an old, mechanical kitchen timer from the thrift store, or even an hourglass — if you remember to watch it. But it is important that the interval be fairly regular. This doesn't work if you spend five minutes on one interval, and sixty on the next.
Next, take a sheet of paper and draw some simple columns on it. The columns need to be wide enough to put one to three or four words in them.
Now, pick an interval. Ten minutes is on the short side; and hour is probably too long. But it depends on how long your typical task is. Generally speaking, you want the interval to be no longer than half the time of the shortest task you want to measure.
Now, set your watch/timer/hourglass/whatever, and begin your day. Each time the timer goes off, write on one line in the column a few words about what you were doing, then reset and re-start the timer.
At the end of the day, you'll have an impressive amount of raw data to play with that may yield surprising results!
First, note the quantum theory principle that observing a system necessarily impacts the system. Y'know, SchrÃ¶dinger's cat and all. Things that require uninterrupted concentration will take longer when subjected to ten-minute interruptions!
It will help your analysis to enter your data into a spreadsheet or simple database. (But it may hurt your time crunch to take time to do this!)
Sort the data by numbers of occurrences -- in other words, make a histogram from it. This first-order analysis may yield surprises -- did I really spend that much time reading blogs? :-)
The more sophisticated may want to try a fourier transform (named after Joseph) of the data, which will give you a frequency plot, which is handy for understanding periodic habits.
Now, you've negatively impacted your focus-requiring activities with a stupid beeping watch, and forked over more time to drawing pretty graphs — what's next?
Once I've sorted and counted and made my population histogram, I color the bars three colors: green for things that are productive, that I should be doing, red for things that are decidedly non-productive, that I shouldn't be doing, and yellow for those things that are "gray areas", that may be aiding or subtracting from my productivity.
Now you can easily see how much of your time is helping, how much is hurting, and how much you need to simply understand better before making a judgement.
Run your sampling again, but this time, have your colored histogram at hand. Whenever you catch yourself sampling a red activity, just stop it, dammit! Or maybe give yourself a mild electric shock or something, to negatively condition yourself to that activity.
I once implemented this as a computer program, with a window that popped up with a self-expanding speed list of previously entered activities, but it got old — meaning, the computer and operating system it ran on no longer exists. If someone wants to program a modern version, I'd be happy to help out.