There are many reasons projects and programs fail. It is assumed that requirements are unclear, change can't be managed, sponsors aren't on board, and the plethora of other problems found in projects. But those problems are pretty much traceable to immature projects or projects - especially IT projects - in general that are low on the maturity curve.
How about projects (or programs) that are more mature on the technical and managerial side? Projects where the participants actually know what done looks like in some credible measure, know how to produce the outcomes, and generally are competent in what they do.
Why do these projects get in trouble - and they do all the time. Here's a list of sources of the Train Wreck for mature projects.
- Inattention to budget responsibilities - management and technical leaders don't keep an eye on the budget. Spending hot can not be recovered. You spend to much for an outcome, and you're over budget, which dilutes the business value of that deliverable. You can never make it up, unless you redefine the value.
- Work authorizations not always followed - the plan needs to be followed. Even in the agile world, the stories agreed to for the iteration need to appear as a result of that iteration. No rearranging the work. In the agile world this is even more important. Producing the business value in the planned order to meet the business need. In other domains, dependencies between produced outcomes is critical for distributed activities.
- Lack of an Integrated Management System - connecting all the project performance information in one place is critical to providing visibility to management. This means everything that takes place needs to have units of measure meaningful to the decision makers.
- Baseline fluctuation and frequent replanning - if you can't stick to the plan, you can't measure your progress. This does not mean blindly following a plan. But if you have a plan and you don't follow it, you're just wasting your time. If you need to change the plan - for a good reason - then do so. But record these changes so you can go back and calibrate your performance.
- Current period and retroactive changes - no cooking the books and making problems go away by changes history.
- Improper use of Management Reserve - MR is for unknowns within scope. No using MR to cover your mistakes.
- EV techniques that don't reflect actual performance - if you're not measuring physical percent complete in units meaningful to the decision makers, than you're wasting your time. The EV numbers won't be credible. You have to measure tangible evidence of progress to plan not someones opinion of progress - Show me.
- Lack of predictive variance analysis - all numbers in the baseline are random numbers. Without understanding the variances of each class of number (reference class models) and the correlations (causality) between the numbers, you'll never get a handle on why things are going wrong.
- Untimely and Unrealistic Latest Revised Estimates - stop changing the forecasts and get a handle on what is really happening. Standdown if you have to, but get it right for some period of time before proceeding.
- Progress not monitored in a regular and consistent manner - how long are you willing to wait before you find out you are late? Sample progress at least at ½ that time period, otherwise when you find out you're late it's too late to do anything about it.
- Lack of vertical and horizontal traceability cost and schedule data for corrective action - vertical traceability shows how the products are increasing in maturity. Horizontal traceability shows how you're actually doing work to increase that maturity. Both are needed for success. This is the purpose of the Integrated Master Plan and Integrated Master Schedule.
- Lack of internal surveillance controls - left alone all processes degrade. Who's watching those doing the work?
- Managerial actions not demonstrated using Earned Value - management has to walk the walk every week. Not just watch others. Management has to manage using the numbers. Only the numbers talk. No other evidence is acceptable.