The business case for our workflow platform promised savings of around four hundred thousand a year, made up of reduced manual handling, fewer errors and faster turnaround. It was approved on those numbers. It was delivered roughly on time. It went live, the project closed, the team was thanked and reassigned.
A year later I went looking for the follow up review and there was not one. Not a poor one, not a disputed one. Nothing had been scheduled, because nothing in our process schedules it.
That is not an accident of our organisation. Funding governance is built entirely around the decision to spend. There are templates for the case, a committee to approve it, a stage gate to check progress and a closure report to confirm delivery. Every one of those artefacts is about whether the project happened. None of them is about whether the thing the project was funded to achieve actually occurred, and the people who would have to answer that question have all moved on to the next case by the time an answer would be possible.
The effect is corrosive in a quiet way. Estimates in business cases become bids rather than forecasts, because there is no consequence to optimism. Nobody learns which categories of benefit we habitually overstate, and in my experience it is always the same two: time saved by staff who then do other work rather than leaving, and error reduction that was never measured before the project either. And when a genuinely valuable initiative comes along, it competes for money against numbers that nobody has ever had to defend afterwards.
We asked for one change and got it, which surprised me. Any case above a threshold now carries a benefits review twelve months after go live, with a named owner who is not the project manager, and with the baseline measurement taken before delivery starts rather than reconstructed later. The first two reviews were uncomfortable. One initiative had delivered about a third of what was claimed; another had delivered more, in a category nobody had predicted.
Both results were useful. Neither would have existed under a process that stops caring the day the system works.
– Serguey Shinder
Top comments (0)