Paper is a resilient technology. It works without a battery, in a truck cab, with cold hands, and everyone already knows how to use it. Replacing it is not primarily a software problem.
Watch the process before designing it
We spent three days in a dispatch office before drawing a screen. The finding that shaped the product was not in any brief: the cost was not data entry, it was the correction loop afterwards.
Do not make everyone change at once
The finance team did not ask for a new system. Forcing them to change their export format alongside everything else would have created a second front. We matched the formats they already used.
Run the old process in parallel
Three weeks of paper alongside the platform, at one depot. Nobody had to trust us in advance; the numbers matched and then the paper stopped.
Adoption is a sequence, not a launch date.
Give the drivers something better, not just something digital
The driver did not care about management reporting. He cared that the form now told him immediately when an odometer reading was implausible, rather than someone phoning him about it on Thursday.