The chargers we were issued took five minutes per meter. With the crew sizes and the daily targets we had, that number set the ceiling on the entire programme — not the driving, not the paperwork, not the reconnection itself.
The usual response is to buy more chargers. We costed it. It was expensive and it didn’t solve the real problem, which was that five minutes of dead time per meter multiplied by everything.
So I opened one up.
I reverse-engineered the board, designed a replacement PCB, printed an enclosure, and tested it against the meters we actually had rather than the ones the datasheet assumed.
Twenty seconds. We built 56.
The part worth taking from this isn’t the hardware. It’s that the constraint everyone was planning around arrived in a box with a specification on it, and nobody had checked whether it was load-bearing. Most programme bottlenecks I’ve been handed since have had the same shape.