Pilot fleets work best when the scope is narrow but real: one region, one supervisor group, or one asset class. Everyone should know which decisions the pilot is meant to inform.
Pick machines that will disagree with each other. A pilot of five identical, well-treated assets teaches you very little; a pilot spanning a hard-worked breaker, a quiet one and a carrier will surface the differences that make the case, or fail to, which is also worth knowing early.
Assign clear ownership for data quality from the start: who confirms installs, who fixes mis-tagged assets, who responds when a unit stops reporting. Without that, dashboards age quickly.
Budget an hour for the install conversation. Battery-powered units need no wiring, which makes fitting fast, but where the unit goes still decides whether the data is any good: flat surface, label facing the sky, away from the point of impact and never fully enclosed in metal.
Expect a gap between the machine's own counter and SolidTrack activity time, and decide in advance how you will use each figure. The counter may record engine-on time, while SolidTrack derives activity from motion telemetry. Neither is automatically wrong, but a pilot that has not agreed which figure supports each decision will stall on that argument.
Define two or three success measures up front, such as reduced idling, faster dispatch to stuck machines, or fewer emergency service runs. Write down the current number before the units go on, because reconstructing a baseline afterwards never convinces anyone.
Give it a full maintenance cycle. Utilization patterns may appear early, but the operational value becomes clearer when the first configured interval comes due and the workshop can review activity, calendar requirements and service history together.
When those measures move in the pilot, you have a case for scaling hardware and training calmly instead of rushing.