In the past several weeks, I have posted on project management best practices. And we talked about the track record of the IT industry as being, at best, a mixed bag. It turns out for the UK government, that would be putting a very positive spin on their IT projects. The Times recently ran an article* detailing the eight worst areas in the government which have cost the taxpayers dearly. IT projects had 3 blatant failures of the eight and had a major hand in 2 of the remaining. How did it come to this? Why is practice of IT reasonable for more than half of the UK government wasteful initiatives? And these are not minor blowups. The Fire and Rescue Plan, which started out as a 120 million pound effort to consolidate 46 control rooms to 9 regional centres was finally axed after it cost 469 million pounds! A straightforward project (how many of us have consolidated call centres, trading floors or command centres in the past 10 years? I would venture 50% of your firms have done this) that cost 4 times the estimate and never even delivered! And that is not the worst one: the NHS records project has cost 6.4 billion pounds to date with at least 2.7 billion of that wasted. While there are 60 million citizens in the UK, many of us in industry have customer bases in the tens of millions where we keep critical financial data safe and accessible for our customers. So while I would agree the health records breaks new ground in some areas, it is not the Manhattan project. This is a very doable project where given the monies spent, it should have already delivered significant benefit and capability. And yet very little has been delivered and there is low confidence this will change in the near future for the program. Overall, how can IT projects have 3 of the 8 slots of government failure areas when the defense industry only has 1 (and you could argue IT projects contributed heavily to that one)?
I think this dismal track record in government for IT projects is due to some common issues and a few unique ones. First, typically there is poor and ambiguous sponsorship. And it is compounded by very weak and changing requirements. Too many parties and groups in government with a stake and a reason to argue and change the project. Applying the methods of political processes of lengthy debate, consensus and influence to defining and running a project are a recipe for disaster. And I suspect the contractors doing the work likely encouraged changes and debate as this was then an opportunity to grow scope with much higher margin work (change orders are always much more profitable than the original bid). Second, the approach undoubtably used a waterfall method. And given the size and scope and vast array of stakeholders, each step (e.g. requirements definition, etc) took an extremely elongated time. An elongated schedule with a cumbersome and bloated program structure to match the stakeholder complexity would certainly have multiplied the costs. Still, it takes even more to cause such spectacular blowups.
There is an excellent book, Software Runaways, available that documents such ‘death march’ projects. ‘Death march’ projects are the kind of massive program that everyone knows is doomed to failure and yet everyone is still lashed to the ship on this voyage to failure. It is a fascinating read and some ways like watching a traffic accident unfold — it’s pretty awful but you can’t tear yourself away. Of course, the UK government and its contractors do not have a monopoly on such spectacular program failures (though they certainly seem to be doing their best to enable additional chapters to be written). What is relevant here though is that the book does an excellent job of reviewing about a dozen of the more interesting past IT program failures and identifying the root causes. These rot cause include the poor sponsorship and ill-defined requirements we discussed above. But it also describes the mentality that sets into a large program team on their ‘death march’. In essence, even though the members of the team know there are massive flaws in the program, because of the complexities and different agendas and influences of a large complex program team, they are often unable to repair them from within. Even worse, when an external party identifies the flaws, the program team then bands together to defend from such external attacks at all costs. Their identity has become so caught up in the program that they would create a Potemkin village to demonstrate that there are no flaws.
So, it is the instinct of these large program organizations that assume a life of their own with all the members now vested in its survival (not its delivery but its survival) that when combined with major flaws (such as ill-defined requirements or the wrong methodology) creates the spectacular failures. Or, put another way, it is how humans work together within large programs that if based on poor practices, can be multiplied to raise the negative results to such an irrational level.
With that in mind, what are some approaches to prevent this from occurring? I would suggest the basic ones of ensuring there is clear sponsorship, proper steering committees should help, but more radical changes to the approach would be better. Let’s take the command centre consolidation. Why do all 46 into 9 in one waterfall or ‘big bang’ approach? Instead take two regions of the 9 and do two pilots, each constructed with their separate sponsors, steering committees and contractors. Set an overall schedule for them to deliver to a well-defined, but high level set of requirements or outcomes. The team that completes their work on time and meets the requirements will have an opportunity to bid on the next two regions to be consolidated. And the one that does not meet the bar will result in the contractor being barred from the next round of work and a negative performance mark will go on the sponsors and government leads. Now, the payoff for contractor encouraging endless changes by government is gone. Further, you are breaking up the work into more doable components that can then be improved in the next regional implementation. Smaller problems sets are eminently more doable then massive ones. By changing the approach to more incremental work with short cycles and aligning program structure and incentives to getting real results, I think you would find a dramatic difference in the delivery of the project.
While these changes would certainly improve project delivery, I am sure there are several other elements that have caused impact and problems. What would you change? How do we get IT projects to not be a huge portion of wasted taxpayer funds?
I look forward to your comments.