Case Study
TSIP
Giving transit signal engineers the visibility they needed to actually prioritize interventions.
Role
Product Designer
Timeline
5 months
Domain
Urban Transit / Gov Tech
Background
TSIP is a program within a mid-size U.S. city's Department of Transportation managing traffic signal priority for the city's bus rapid transit network. TSP hardware was installed at 140+ intersections, allowing buses to request early green lights or extended greens to reduce delay.
The operations team — four signal engineers responsible for the entire network — had almost no visibility into how well the system was actually performing. Their primary tool was a legacy reporting system that generated weekly CSV exports: rows of intersection IDs, request counts, and grant/deny rates. No visualization. No filtering. No real-time view.
Challenge
The team was entirely reactive. Engineers learned about problem intersections from bus operator complaints, then spent hours pulling logs to confirm what was happening. When a council member asked why Route 4 was consistently late, the team couldn't answer without days of manual data analysis.
The data existed — 143 TSP-enabled intersections generating continuous performance logs. The problem was that the data couldn't tell the engineers where to look.
Key Findings
40% of engineering time was spent on data retrieval, not decisions
A workflow audit shadowing two engineers for a full work week revealed that roughly 40% of their time was spent pulling CSVs, building pivot tables, and creating one-off charts for internal reports — before any signal diagnosis had even started.
22 underperforming intersections were invisible in the existing tool
Log analysis showed a clear disparity: of 143 TSP-enabled intersections, 22 had grant rates below 60%. These weren't flagged anywhere — they lived silently in the CSV alongside the 121 that were working fine. Without sorting or filtering, the problems were effectively invisible.
Key Results
Data work per engineer
~16 hrs/week → ~3 hrs/week
81% reduction
Time to identify problem
5–10 days → same day
Threshold-driven
Misconfigured intersections
7 found in 90 days
Avg. 4 months undetected
Network TSP grant rate
74% → 81%
+7 pts