Pilots are popular because they are safe. Nobody was ever fired for running one. They are also, in our experience, the most reliable way to spend a year proving something everybody already believed and then not doing it.
What a pilot proves
That the idea works under conditions chosen to make it work: the friendliest team, the cleanest data, the most patient sponsor. This is not nothing, but it is rarely the thing anyone doubted.
What a pilot avoids
Who owns it afterwards. What the evaluation threshold is. What happens when the friendly team is replaced by the one that has three other priorities. Whether IT will approve it. Whether the data is like this everywhere. Every one of these questions is deferred to ‘after the pilot’, and after the pilot the budget is spent.
4 of 5
abandoned systems we have reviewed that were technically fine
How it was measured
They died in the handover, nobody owned the thresholds after the consultants left. None of them died in the pilot.
What to run instead
A thin slice: one job, end to end, in production, in the real environment, with the real access controls and an owner named before it starts. It is slower to launch than a pilot and faster to be real. If it survives one queue it earns the next; if it does not, you have learned something a pilot could not have told you.
A demo takes a fortnight. A system your staff rely on at nine on a Monday takes twelve weeks.
Marek Halloran, founding partner