I have a system that toggles between a google spreadsheet for my budgeting and projections, then mint to track my spending (which then informs my spreadsheet budgeting and makes it more accurate over time).
Mint is first, it pulls in all of my accounts and transactions from my past year's spending, I can see at an annualized level what my general spending patterns for all the categories looks like. With these numbers, I then input into my historical spending tab in my spreadsheet, do some math to inflation-adjust the numbers into today's dollars, and then average the category spending across time (to account for annualized spending of large purchases as well).
This averaged spending goes into my budget, or rather what I refer to as my spending plan. Over time, this budget is incredibly accurate, but it isn't accurate month-to-month (ie we may buy bulk groceries that month and put the grocery category temporarily "over-budget", or a quarterly bill will be paid). However, to account for this natural ebb and flow, we keep a month buffer of cash in our checking account.
Anyway, the result is, over the long run, on average we spend $X per month. On some months it will be +/- $X, but we don't get worked up over that much, entirely because over the last few years our annualized spending has been so incredibly consistent, and our "budget" includes those weird one-off purchases that we have made in the past.
Then it is just a matter of filling in the other parts of my spreadsheet to get what I am ultimately most interested in: what is my surplus to save and invest (on average) each month?
So I have my income, I also take out this year's taxes from MDM's case-study spreadsheet, and then with whatever I have left over, I begin allocating the surplus based on the Investment Order that floats around this forum. There's some fun pre-tax vs post-tax things in there that will then again change the numbers a bit so I'm always tweaking my spreadsheets to be more accurate/refined.
Anyway, initially setting up this system was pretty time intensive and it is still occasionally a work in progress. But now that it is mostly running smoothly, it requires minor checks in mint each week to update any wonky autocategorizations, minor updates if things change on the income side, and then each year about two hours of work to pull in that year's spending and recalculate our data-driven spending plan for the upcoming year and refresh any auto-savings if there are changes there.