Usage logs from a fleet management platform showed something we did not expect after six months in production: the single most-used feature, by a wide margin, was a filter dropdown that took us about a day to build.
What it did
"Show only vehicles that need attention today." One filter, applied to the main vehicle list, using rules already computed elsewhere in the system — service due, document expiring, unresolved fault code. Nothing about it was novel engineering. It combined three things the platform already knew.
Why it mattered more than the dashboard
We had built a full analytics dashboard as one of the headline deliverables — fleet-wide trends, utilisation charts, cost breakdowns by depot. It gets opened, but rarely, and mostly by managers preparing a monthly report. The filter gets opened by every dispatcher, every single morning, because it answers the question they actually have at 8am: what do I need to deal with right now, out of ninety vehicles.
The pattern behind it
Dashboards answer "how are we doing", which is a question people ask occasionally and with time to think. Operational software mostly needs to answer "what do I do next", which is a question people ask constantly and under time pressure. We had scoped generously for the first kind of question and adequately for the second, and usage data corrected us.
The feature that gets opened every day is usually smaller and less impressive than the one that gets opened once a month. Scope for the daily one first.
What we did with the finding
- Promoted the filter to the home screen, rather than one option among several on a list view — it is now what a dispatcher sees first.
- Added two more filters in the same spirit, built from combinations of data the system already had, none of them requiring new architecture.
- Reviewed the analytics dashboard's scope down slightly for the next phase, on the basis that it was solving a real but occasional problem, and we had more valuable occasional-problem work waiting.
The general lesson we took
Usage data six months in is a better scoping tool than any amount of upfront discovery, because it tells you what people actually reach for rather than what they said they would need. We now schedule a usage review at that six-month mark on every ongoing engagement, specifically to catch this kind of mismatch before a second phase compounds it.