The situation
Fleet compliance was tracked in a workbook maintained by one operations coordinator. It covered nine hundred vehicles across owned and partner fleets, with expiry dates entered by hand from documents emailed in by depot managers.
It worked, in the sense that it mostly caught things. But it depended entirely on one person's attention, it was always slightly out of date, and partner fleet data arrived in whatever form the partner felt like sending.
The baseline run
The first step was a single batch verification of the entire roster. The output showed a meaningful number of vehicles whose recorded status and actual status had diverged — mostly lapsed paperwork on partner vehicles, plus a handful of records that had been wrong since they were first entered.
That run did more to justify the project than any projection. It turned an abstract risk into a specific list.
The ongoing process
A weekly job now submits the roster and writes the results back. Alerts fire on vehicles whose status changed since the previous run, which keeps the notification volume low enough that people still read it. Data-quality failures route to a separate queue so they don't drown the compliance signal.
Where it landed
The coordinator's job changed from maintaining a list to acting on exceptions. Partner fleets are verified against records rather than against documents the partner chose to send. And the historical runs are retained, which turned out to matter the first time someone asked what the fleet's status had been in a specific month.
A composite account drawn from typical engagements. Details have been generalised to illustrate the integration pattern.