Proving before scaling.
WHAT A PILOT SHOULD ESTABLISH
That the data is accurate That devices survive the environment That connectivity works at the sites That the value is real and measurable What it actually costs to operate
WHAT SIZE TO RUN
Enough devices to see variation, few enough to manage.
WHAT DURATION
Long enough to cross the conditions that vary: seasons, weather, usage cycles.
WHY
Problems specific to a season appear only in it.
WHAT TO MEASURE
Device failure rate Battery performance against prediction Data completeness Support contacts Time spent per device
WHAT DATA COMPLETENESS MEANS
The proportion of expected readings actually received.
WHY IT MATTERS
Decisions based on incomplete data may be wrong, and the gaps are invisible.
WHAT TO VALIDATE
Readings against independent measurement.
WHAT TO ASK USERS
Whether the information changed anything they did.
WHY THAT QUESTION DECIDES THE PROJECT
If nothing changed, scaling multiplies cost without benefit.
WHAT TO RESOLVE BEFORE SCALING
Everything the pilot revealed.
WHAT TO RESIST
Scaling because the pilot technically worked.
