Monday, November 21, 2005

Why IT projects continue to fail

Many IT projects fail. There are many reasons that are accepted as to why so many IT projects fail. Several articles have been written on the subject and most people seem to agree with the basic concepts that there are 5 or 6 basic reasons why IT project fail. I will not go into detail on those but there are some referenced articles below.

The biggest problem that I have today though is why so many IT projects continue to fail. I propose that there are two viable reasons for the continued failure of IT projects.

Lessoned Learned: A total lack of the IT community to accept the fact that creating and reviewing lessons learned from similar projects can save time, money, expense, training and preserve jobs. Most other industries view lessons learned as the most valuable part of Project Management while in the IT community, even if it is done it is an afterthought, and rarely if ever used as research before beginning a similar project. This attitude towards lessons learned has caused us to time and again repeat the mistakes that were made before. Project Managers should understand that every overcome issue or risk provides a valuable lesson learned, documenting that and ensuring that other Project Managers can acquire that knowledge is something IT Project Managers need to concentrate on.

Estimating: Two things about estimating that I believe are important parts of why IT projects fail so often. The first is who does the estimate the second is understanding why who does the estimating is so important.

Developers and most IT focused people in general should not IMHO (In my humble opinion) ever give final estimates. Why you ask? Ah, I'm so glad you did. The main reason is that IT focused people and developers see things very differently than many business people. For lack of a better example IT people see things in 0's and 1's and business people see them in dollars and cents. I do believe that there are business minded IT people out there who see things both ways but they are the exceptions not the rule. In my dealings with projects over the past 15 years I have seen more projects fail due to bad estimations than any other single item. Most projects that fail are already late and over budget before they are abandoned. In many cases there were other contributing factors but always there were tasks that did not have proper time assigned to get the work done.

In my historical research of my own projects over 60% of the estimates done by IT people were 40% or more under the actual time needed to complete the task. 20% of the remaining tasks were over 30% below.
(80% of all tasks estimated by IT people in my experience were 30% or more under the actual time needed to complete the task.)
What of the other 20 % of tasks? 40% were below estimate and 40% above estimate. Only 4% of all tasks estimated by IT people were completed on time and within budget.

While I do believe that this data may be flawed in that the estimates may have been adjusted down by business people or project managers, again, in my experience they have been adjusted up to allow for the possibility of under estimating. Often a project is adjusted to meet a deadline date but the hours needed are then adjusted across resources.

There are two possible solutions for the poor estimates:

First would be to train IT people on proper methods of estimating, taking all IT and other business considerations into effect. The second would be to use developer and IT focused peoples input but use business people to create the actual estimate. The business people used to create those estimates should have training in estimating.

Why IT projects Fail Links:
* http://www.computerworld.cmanagementics/management/project/
* http://www.computerworld.cdevelopmentics/development/story/0,10801,71209,00.html

Estimating Links:
Ideas on Estimating Web Development Costs (from Crosstalk magazine)
SCEW: Software Cost Estimation Web Site
Estimating Software Costs from SD Magazine (requires paid registration), Part 2, Pt 3, Pt 4
Parametric Estimating Handbook
"Reducing Bias in Software Project Estimates" from Crosstalk magazine
To Err is Human, To Estimate, Divine, an Information Week review of 2 estimation products: Cost Xpert and KnowledgePlan
Estimating Object-Oriented Software Projects with Use Cases well-written, but long thesis paper

1 comment:

Mark Schneider said...

Thanks for an excellent blog on estimation and project failure.

The links to estimation resources on the web are very useful too.