Clarke Power · 2020
Digital Inspection App
Replacing clipboards and guesswork with structured fleet telemetry.
The Problem
Clarke Power's field technicians completed vehicle inspections on paper forms. Forms were lost in trucks, left illegible by handwriting, or never returned to the office for entry. The business had no reliable data on fleet health, no inspection completion rate visibility, and no ability to correlate service history with equipment failures downstream. Every insight about maintenance patterns had to be assembled manually — when it was assembled at all. Leadership was making fleet investment decisions with almost no data.
Reflection
“The 60% velocity improvement came from the structured capture design, not the mobile technology itself. When technicians don't need to write sentences, they move faster and more consistently. The data quality improvement was equally significant — and it was a direct result of making the right input the easiest input. That principle applies to almost every data collection problem I've worked on since.”
Approach
Design with technicians, not for them
Before writing requirements, I ran observation sessions with four field technicians across two service sites. The non-obvious insight: the app had to be operable with one hand while holding a flashlight in a dark equipment bay. Every interaction was redesigned around taps, not typing. Structured dropdowns replaced written descriptions. Photos with auto-tagged metadata (timestamp, GPS, technician ID) replaced written defect notes. The design constraint — one hand, low-light, time pressure — drove every UX decision.
Offline-first architecture
Field locations have unreliable or absent cellular connectivity. I designed the data flow around the assumption of no network: inspections are completed and stored locally on-device, then synced in batch when connectivity is restored. This eliminated the primary failure mode of a connected app — a dropped connection that corrupts or loses an in-progress form. Technicians never wait for a network response. Sync happens invisibly in the background.
Structure the capture to structure the data
Open text fields produce unstructured data that requires manual review and interpretation before analysis. Every inspection field was designed as a structured input: pass/fail toggles, severity ratings, dropdown categories, and photo attachments. The constraint served both the technician (faster to complete) and the analyst (immediately queryable, no ETL cleaning step). The right answer was always the easiest answer to select.
Architecture
Capture at the edge, sync when connected, query immediately. No data entry step, no cleanup required.
Impact