Also, a phrasing tip: "I don't know" is more diplomatic than the more satisfying, but also more career-limiting, "How on earth should I know?"
In any event, the rule of corporate numbers means you should only utter a number after you're confident you can meet or beat it. To do that when estimating projects, you have to be in a position to plan the work -- you must be able to break down the work into tasks take no longer than a week, described as an outline about three or four levels deep (the work breakdown structure). Once you have that, you can probably estimate each task with a reasonable degree of accuracy.
Dealing with contigencies
I should say your estimate would be reasonably accurate except for the contingency piece, which is critical, because projects with no surprises are exceedingly rare. Something always happens to throw off the original schedule, and here's how to handle that:
If the estimate should not include provision for contingencies, add 15 percent for contingencies anyway, and hide it. If the reason for this isn't clear, re-read the rule of corporate numbers. If you're told to include provision for contingencies, add 30 percent: 15 percent where it can be seen, and 15 percent that's hidden.
Dealing with complexity
One more thing: Some projects are so complex that understanding them well enough to estimate their cost would require more time and effort than you have to spend.
If you're faced with estimating one of those projects, don't scramble to do the research in what you laughingly call your spare time, and don't violate the rule of corporate numbers, either. Instead, recommend that the requester charter a small project before the project, whose purpose is to provide a reliable estimate. Here's your script:
We have two choices. One is that I can make up a number. When the shooting is over, we'll have spent something -- possibly a lot more than my number -- after which you'll be angry at me and everyone else will be angry at both of us. Or, the business can decide this idea has enough potential that it's worth investing a modest amount of time and effort to do a professional job of estimation. If you like that idea, give me a week to put a plan together and I can tell you what the estimate will cost us.
Then loop back to the top and create the work breakdown structure for the estimation effort. That shouldn't take more than a week no matter what's going to happen at the end of it.
This story, "The black art of estimating IT project costs," was originally published at InfoWorld.com. Read more of Bob Lewis' Advice Line blog on InfoWorld.com. For the latest business technology news, follow InfoWorld.com on Twitter.