CEO Blog - How to bring an IT project (or anything else) in on-time, on budget and on-specification

At exactly this time last year, I was on my way to the UK Defence Academy, Shrivenham do a talk at the kick-off meeting for a huge new IT project that will combine logistics for the UK army, navy, and air force in one centralised operation. 

They wanted to hear how uFlexReward combined the holy trinity of being on time, on budget and on quality with a complex new design and build. They were interested because, as in every nation, defence tech projects have a troubled history. They make war look simple. 

Fortunately, I could tell them about some very easy-to-use and understand ideas, that could make a success of any IT project, including theirs (which wouldn’t be normal). They weren’t my ideas.  They were based on the work of four Nobel prize winners, and a Harvard Law School and Oxford University professor, so they could feel safe. 

Planning – As Mike Tyson said, “Everyone has a plan until they get punched in the mouth”.  As Michael suggests, very few plans go ... to plan. One reason for this is “Optimism Bias” for which Daniel Kahneman won the Nobel prize for suggesting human beings are psychologically programmed to think things will be ok, even if everyone else who tried it crashed and burned (IT projects again).  To guard against this we asked outsiders, who had no emotional connection to our plan, to review our assumptions. They suggested important changes. This is called taking “the Outside View”. It works. 

Forecasting – When setting our targets for delivery, budgets and performance we took account of uncertainty as well as bias. When you do something hard, that has never been done before, you are insane if you think you can predict an exact date, cost and outcome. But many IT projects are launched with bold promises that circle them like albatrosses. To avoid this trap, we used spread targets to set a range and then narrowed the range, based on experience, as the project unfolded. This way your endpoints get closer as you gain confidence. The longer the project goes on the more accurately you project the end date and cost. As long as you are within your spread boundaries it is a win. But what about the disaster of going outside the spreads? We used Reference Class Forecasting, to manage that. We referred to the work of Professor Flyvbjerg and Dr Budzier’s McKinsey / Oxford University database of 2,000 IT projects to check our assumptions were within the norms of high and low boundaries and then used Option Theory to take stock at breakpoints where the project could be paused, continued or even scrapped (with no loss of face). Bragging about fixed-cost delivery by a specific date is the definition of insane. Don’tdo it.

Supplier Selection – This was our biggest innovation. Obviously, as the market does not offer buyers a total reward platform, Unilever faced a make not buy decision, so we had to select a software development partner. Just as the MOD has to for its new logistics system. The traditional route is RFP, ITT and then a shoot-out that is often won by the lowest cost bidder, who is then chained to a contract that manages risk through fixed price delivery, milestone payments and a fierce change management process. We felt the adversarial nature of this was not just unhelpful. In fact, it is guaranteed to fail for everyone, except the lawyers. So, we invited fifteen suppliers to make a pitch (over a two-day speed dating format) and selected three to build a prototype, towards which we made a financial contribution. In effect, we were paying three suppliers to sell to us, but in return, we worked alongside their development teams on a daily basis. All three suppliers would have produced a successful product, but this approach allowed us to choose the supplier who had the best cultural fit. We are super Agile and Collaborative (ref MacNeill “Relational Contracting”) and wanted to work with a supplier who was willing to work fast and agree on changes on the fly in an informal process using the principle of local decision making and devolved, presumed pre-approval. Without daily contact in the pilot builds identifying that supplier would have been hard. Finally, our Outside View team had a vote on supplier selection to guard against the risk of bias. In another innovation, we saved time by carrying on with development on a time and materials basis while the full contract crawled through the legal process.

The Outside View, Reference Class Forecasting, Spreads, Options Theory and Relational Contracting delivered for us. We took only 18 months to go from speed dating with suppliers to a global roll-out to 100,000 employees in 120 countries with a product that works super-fast, with zero downtime, and rock-solid security (reward is, after all, highly regulated data).

I don’t know how the MOD went on to handle its new project, maybe our ideas are too whacky for a huge defence project, but the audience liked what they heard, and I made some good new friends and had a sobering conversation over dinner about how we much we pay soldiers to risk their lives and how important non-economic rewards are to the military. More on that in another CEO blog.

Ken Charman - CEO of uFlexReward