Budgets and Estimates
What about my Budget?
This is sure to get people mad: Budgets and Estimates should not be directly related. If your estimates are of units of work which are small, comparable to other recent work, and in the near-future, then they are going to be fairly accurate. That's great! Now, what's the ROI on that bit of functionality? If it isn't significantly higher than the estimated cost, drop that work immediately. Find something else to do. Agile is about meeting the constantly changing needs of the business, and if the costs aren't far outweighed by the benefits, then you need to find something more valuable to work on.
Let's say the estimate is 6 months at around $150,000 for 3 people, but the result will save the company $500,000/year. Yes, definitely do that (unless you find something even more valuable). As far as the budget is concerned, it shouldn't be rocket science: set aside $150,000 for the next 6 months. The part that's missing is that during those 6 months the project should be evaluated after every release to see if the next increment will deliver any more of the value.
Suppose that 3 months into the project the team has met the most critical goals of the system. Call me a dreamer! Maybe the ROI of the work completed so far is $400,000/year with $75,000 spent. Is that additional $100,000 of ROI with the additional $75,000 of investment? Probably not, and so maybe the project gets cut short. The leftover budget gets used on the next project, and nobody should be disappointed or upset.
Suppose that 3 months go by and everything seems to be on-track. So we're 50/50 on investment and returns. Let it ride, we're probably fine, but we should still compare this to the other (new) tasks which have come up since the last quarter.
Suppose that 3 months into the project things are not looking so good. The estimates were far too low, and progress is slow, only 25% completion. Things are going half as fast as expected. You could try to extrapolate that what you thought would cost $150,000 will end up costing $300,000 for the same ROI. Is that still worth the time? Probably not, but it isn't all bad. Sure, you spent $75,000 so far, but the ROI should be about 100,000/year, so it isn't a total waste. We still have the $75,000 we didn't use, and we didn't devote even more time/money into a faltering project! Still not a bad scenario, though not ideal.
The budget for a project is the amount of money we've devoted for RIGHT NOW to work on something. It is related to the estimates, but is not simply a copy-paste job. Certainly, a project may exceed the estimates, but is shouldn't ever exceed the budget. A decision to spend a specific amount of money on a project is the budget.