Steven Mezak, who runs an application development consultancy that uses offshore services and publishes “Runtime,” a software outsourcing newsletter, is running an interesting column this week, titled, “How to Estimate the Cost of Your Outsourcing.”
He says that several teams responded to a fixed price RFP from a client, which “contained a reasonably good specification although there were some missing details about the interfaces with other systems.” The average estimate was 2,900 person hours to develop the software. The client had made an estimate of 600 hours, and they selected a service provider who had come in with a similar estimate.
And that gets to the theme of Mr. Mezak’s article. Why such a wide variance in those two estimates?
He walks through the various schemes that have been concocted through the year to estimate lines of code — the counting of functional points, the constructive cost model, the time needed to concoct each user screen.
But the bottom line is that a fixed price project may not be the best approach for choosing a vendor — unless, that is, you have ample experience in estimating what the job consists of, do a smackdown job of speccing it and then figure out a fair price for the work.
Too bad we’ll never know if the team selected for this work came in on time, in budget and with satisfactory results.