Here’s a nice brief article by Frank Hayes from Computer World discussing the latest Chaos report:
We're losing ground. Only 28% of IT projects succeed these days, down from 34% a year or two ago. Outright failures -- IT projects canceled before completion -- are up to 18% from 15%. The remaining 51% of IT projects are "challenged" -- seriously late, over budget and lacking expected features.
Those numbers are from the just-completed Chaos report from The Standish Group. Standish has been doing this study since 1994, and ever since, we've been steadily improving our ability to deliver projects. A decade ago, only 16% of IT projects were successes. By last year, it was twice that. Now we're backsliding.
… [via Greg Belanger on the ScrumDevelopment yahoo group]
Hayes recommends keeping projects small, which is nice but I’m writing a book where this is one of my recommendations … so I wish he’d keep his big gob shut :)
Two things worth noting:
-
According to Chaos a “Successful” projects is
on time, on budget and with all the features and functions originally promised.
I can live with on time and on budget but given the
finding (from Standish's chairman Jim Johnson) that
on average, 45% of built requirements are never used, and 19% are only used rarely then a project probably isn’t successful if it delivers what is in the original spec, because
the spec was a guess and is wrong at the start. So, according to this definition a successful project is one that doesn’t benefit from the learning done during it’s execution and delivers a product that the customer mostly doesn’t want. Hmmmm.
-
Statistically speaking (and forgive me because I’m not a statistician) when the numbers go up and down like described in the first chapter (you can see the 1994 – 2000 numbers
here) then you can’t predict trends – i.e. you can’t conclude that we are in fact backsliding. Just because it was sunny all last week and then it rained suddenly today doesn’t mean that monsoon season is upon us. So, we aren’t necessarily getting better or worse …