How can we estimate project proposals without all the information to hand? This is not an unreasonable request. The other week I spoke to a BA who was facing this problem—another department had written a project proposal and wanted to put it in front of the board of directors for approval, but they needed an estimate from a development team. Because of the way the proposal had been written (with lack of BA and development input) there was critical information missing.

And yet despite that suboptimal genesis, the question is still a valid one—they have a project idea, they’d like to know how big a deal it is, and the technologists can surely help them.

Together we devised a couple of approaches. The BA had already found one such. Wary that the department’s wishlist would be lengthy as well as woolly she was planning to ask them to divide their requirements into phases, to make the estimation task easier. The phases should be something like high, medium and low priority. I would also prefer the priority to be based on value, with the assumption being that the project might get canned at the end of any particular phase, so prioritise accordingly.

My own suggestion was to have the development team estimate ranges: “We are 90% confident that it will be more than X months, and 90% confident it will be less than Y months.” We may find that the range we give is enough for the board to make a decision. But if that range is too wide for comfort then the next question to the developers is: “What one or two pieces of information will help you narrow that range?” And so we repeat until we get a sufficiently narrow estimation band.

And of course the two approaches can be combined: You can estimate ranges for phases of the original project.