Measuring your own productivity for your own purposes: Not really much point; people tend to match their level of performance with their levels of desire and skill. Dollars earned is a lagging metric but it's probably the best one.

Measuring the productivity of others (say, your employees): Goals met versus goals set (by you), assuming you're realistic in measuring the median rate of goal achievement in a pool of available talent. That's pretty hard to get right, particularly when delays aren't always employee-related (servers go down, your supplier's software had a bug in it nobody noticed until your employee had to track it down).

Nobody much here is in the business of managing throngs of interchangeable laborers; you're usually lucky to have the talent you retain and are constantly trying to fill positions created by company growth and by turnover. Unless a coder really disappoints by any metric, you're not likely to fire a less-than-median performer because replacing them is so difficult, so a measurement of productivity tends not to be something you can do anything with.

A much more interesting question is how you can maximize productivity for both cases.

I would recommend the ideas and techniques from this book -- Codermetrics.

It's software development centric but the idea is very interesting for measuring personal performance and productivity.

Todo list primarily. Striking things off a piece of paper is oddly satisfying.

I basically break down my tasks into stuff that can be done in a few hours.