The system is hardened and deployed. A defined group of real users now works with it daily, and the engagement enters the phase with the least code and the most consequence: the pilot.

A pilot is not a soft launch. It is an experiment with a decision at the end: expand, fix, or stop. Pilots that forget this become the permanent kind, and you already know how those die.

#Design the pilot to produce a decision

Three design choices upfront decide whether week six ends in a decision or a shrug.

A defined cohort: one team, named users, chosen with the champion. Broad enough to be believed, small enough that you can talk to every user personally. The cohort should include at least one sceptic, because a pilot that only converts believers proves nothing to the people who decide budgets.

Success criteria inherited, not invented: the scope document from lesson four already fixed the metric, the dated baseline, and the target. The pilot's job is to move that number and document the movement. Resist the temptation to add new success criteria mid-pilot; that is how the goalposts learn to walk.

An end date with a meeting on the calendar: the expand-fix-or-stop decision, scheduled before the pilot starts, with the sponsor in the room. The date creates the deadline energy that keeps everyone honest, including you.