Except I don't think it's complete, or complete enough
It's a good discussion as far as it goes, but here's what he says about archiving data:
Following delivery of the Post Implementation Review Report, the project database is archived. Building a repository of past projects serves as both a reference source and as a training tool for project managers. Project archives can be used when estimating projects and in developing metrics on probable productivity of future teams.
Now, that's necessary, but not sufficient. What you really have to do is update the enterprise estimating models, not just archive the project. Maintaining models is really the only way to improve estimating. If I say that a "hard" specification requires "X" hours with "Y" skills, the credibility of X and Y are on the line in proposals and in execution.
Too many organizations "build a repository of past projects" without putting the effort into mining the information in that repository to refine a model of how the organization really works.
And of course, the outliers have to be dealt with, either in the footnotes, or in the distribution of possible outcomes. After all, X and Y should be expected values if they are single numbers, and if not, then they should be ranges, better yet: percentile rankings. To wit: a hard specification requires X hours at the 95th percentile. Now we have something we can work with.
My advice: don't just close; be a Closer!
Are you on LinkedIn? Share this article with your network by clicking on the link.
