Usability improvements for Intelligent Routing
Feature description.
Fill Routes was an automated route generation workflow inside FieldRoutes used by dispatchers and routing managers to schedule technicians across high-volume pest control and home service operations.
The platform generated optimized technician routes automatically, but adoption remained low because users lacked confidence in the system’s recommendations. Dispatchers frequently rebuilt routes manually due to poor visibility into stop constraints, technician fit, scheduling logic, and operational tradeoffs.
I led the redesign of the route review experience, focusing on operational clarity, automation trust, and reducing the cognitive overhead required to validate and adjust generated schedules.
Current State
Automation Adoption - 50%
Manual Route Adjustments - 60% of routes required edits
Route Validation Time - 10 min average review time
Dispatcher Confidence - 2.8 / 5
Interaction Overhead - Excessive context switching
Success Metrics
Automation Adoption - 75%
Manual Route Adjustments - Reduce to 30%
Route Validation Time - Reduce to 5 min
Dispatcher Confidence - Increase to 4 / 5
Interaction Overhead - Reduce workflow steps by 30%
Research & Discovery
Overview
I conducted 7 moderated usability testing and operational workflow analysis with Dispatchers, Routing Managers, and Field Operations teams as well as our internal Customer Relationship Managers
We observed users managing:
-technician skill matching
-geographic routing
-service windows
-recurring appointments
-day-of schedule disruptions
-territory balancing
-technician capacity constraints
A major insight was that users did not necessarily distrust the routing algorithm itself. They distrusted their ability to quickly verify whether the system had made the correct operational decisions.
The issue was not automation quality alone. It was visibility and explainability.
Key Insights
1. Route validation was too fragmented
Users constantly jumped between the map, side panels, and appointment detail screens.
2. Operational context was buried
Critical information like service type, constraints, technician fit, and scheduling conflicts required multiple clicks to uncover.
3. Dispatchers relied on mental models
Many users manually reconstructed route logic in their heads because the system did not clearly communicate scheduling rationale.
4. Automation trust depended on speed
If users could not validate a route quickly, they defaulted to rebuilding it manually.
Design Principles
We established several principles to guide the redesign:
Surface operational context immediately
Reduce navigation overhead
Improve map-to-route relationship clarity
Support rapid scanning under time pressure
Design for human oversight of automation
Preserve flexibility for manual intervention
