Good article over on CIO Mag: ‘Contract Sadness’ by Michael Schrage.

Too many CIOs cut enterprise software deals that look fabulous to the CEO and CFO but commit the people who do the real work to a nightmare of unrealistic expectations.

I have seen this happen too often – companies choosing a winning bid because it ticks all the boxes on the ITT (Invitation to Tender), thus meeting all requirements, often at the lowest cost, only to be hit with unexpected costs once the project is underway. This approach is flawed. It is impossible to produce a perfect list of requirements for a system that has yet to be built. But companies want to be able to compare potential suppliers’ bids on a ‘fair and equal’ basis, so will not allow any changes to that initial requirements list (all hail the list!). Any attempt to do so and you are told to either respond to the exact requirements or drop out of the process. If you want to win the bid, your choice is simple: keep the proposal as basic as possible, to secure the lowest bid, and ensure you have a well-written ‘change management’ clause for costs incurred when the customer starts to realise that requirements list needs to change… This is a lose:lose situation for everyone (except for the lawyers, as usual).

Instead of trying to write the perfect system requirements that all potential ‘suppliers’ must respond to, it’s far better to keep the initial outline simple (i.e. crystal clear in defining the intended purpose of the system, not simple as in only list 2 features) and invite ‘partners’ to bid to assist with the project. Then select up to 3 partners who best fit the skill set you are looking for, and run a short pilot to test that their skills meet expectations. Then choose your partner and work with them to define a ‘realistic’ (i.e. never going to be perfect, you will still need to factor in changes) system design and project plan with associated costs. You are far more likely to get the end result you want, within a cost and time frame that you expect.

%d bloggers like this: