Real-Time Operations Dashboard That Gave a Transit Authority a Single View of 200+ Daily Operations
The regional transit authority achieved what its board had mandated and its operations team had needed for years: a single, real-time view of the entire organization. By consolidating seven disparate data sources into an intelligent dashboard powered by AI-driven anomaly detection, the authority replaced stale weekly reports with live insights that enabled faster decisions, earlier problem identification, and more efficient deployment of its 180-vehicle fleet. Automated board reporting and proactive ridership forecasting further freed leadership from manual data assembly, redirecting that capacity toward strategic operations.
A regional transit authority operating 45 bus routes across 12 maintenance facilities with a fleet of 180 vehicles and 300 employees had no unified view of its operations. Maintenance data, fleet status, scheduling, ridership figures, and financial performance lived in separate systems that were never designed to communicate with each other. Leadership relied on weekly summary reports assembled manually by department heads, meaning operational decisions were consistently made on stale data.
The authority needed a real-time operations dashboard that consolidated data from every operational domain into a single, actionable view, enabling faster decisions, earlier problem detection, and more efficient use of its assets and staff.
Client: A regional transit authority
Employee Size: 300 employees
Industry: Government & Public Sector
Services: - Custom Software Development - AI Automation & Data Analytics - Dashboard & Reporting Development - Cloud Infrastructure
The transit authority's operational data was scattered across at least seven distinct systems. Fleet maintenance was tracked in a legacy CMMS (computerized maintenance management system). Scheduling and dispatch used a dedicated transit operations platform. Ridership data came from fareboxes and automatic passenger counters that exported daily CSV files. Fuel consumption was logged by the fleet fueling system. Financial data lived in the authority's ERP. HR and staffing records were in a separate personnel system. Safety and incident reports were filed in a document management system.
None of these systems shared data automatically. Department heads manually compiled weekly summaries for the leadership team, a process that consumed an estimated 40 staff hours per week across the organization. By the time leadership reviewed the data in their weekly operations meeting, it was already three to five days old. Emerging problems, such as a spike in vehicle breakdowns on a particular route or a ridership trend that warranted schedule adjustment, were identified too late for timely intervention.
The maintenance backlog was a particular pain point. The authority's fleet of 180 vehicles required rigorous preventive maintenance to meet federal safety standards and maintain service reliability. However, the maintenance team had no real-time visibility into which vehicles were approaching service thresholds, which were already overdue, and which had recurring issues that suggested deeper mechanical problems. The maintenance director relied on printed spreadsheets updated twice per week, and vehicles frequently ran past their maintenance windows because the data did not surface the need in time.
On-time performance, the transit authority's most visible public metric, had been declining for three consecutive quarters. Leadership suspected that the decline was driven by a combination of factors including vehicle reliability, scheduling inefficiency, and route-level demand imbalances, but they had no way to correlate data across systems to confirm the root causes or measure the impact of corrective actions.
The authority's board of directors had grown increasingly frustrated with the inability to get clear, current answers to straightforward operational questions. Board meetings frequently devolved into debates about data accuracy rather than strategic discussion, because different departments presented numbers that did not reconcile. The board issued a directive requiring the authority to implement a unified operational reporting capability within the fiscal year.
The authority needed a solution that would: - Consolidate data from all operational systems into a single view - Provide real-time visibility into fleet, maintenance, scheduling, ridership, and finance - Detect anomalies and emerging problems before they escalate - Reduce the manual effort required for leadership reporting - Support data-driven decision-making at all levels of the organization
JSG designed and implemented a real-time operations dashboard platform that unified the authority's disparate data sources into a consolidated, interactive view with AI-powered anomaly detection and automated alerting.
Key Components
Unified Data Pipeline N8N orchestrated data ingestion from all seven source systems on schedules ranging from real-time (dispatch and GPS) to hourly (maintenance and ridership) to daily (finance and HR). Data was normalized, validated, and loaded into a centralized data warehouse. Pipeline health monitoring ensured that data freshness was maintained and that any source system outage or feed failure was detected and flagged immediately.
Executive Operations Dashboard Power BI embedded dashboards provided the authority's leadership with a single-screen view of all 200+ daily operations. The primary dashboard displayed fleet availability, active routes, on-time performance, maintenance status, ridership by route, and daily financial metrics. Drill-down capabilities allowed users to move from portfolio-level views to individual route, vehicle, or facility detail with a single click.
AI-Powered Anomaly Detection & Ridership Prediction Azure OpenAI models were trained on 24 months of historical data to detect operational anomalies in real time. The system flagged unusual patterns such as sudden drops in ridership on specific routes, vehicles with accelerating maintenance frequency, fuel consumption outliers, and scheduling gaps. A separate predictive model forecasted ridership demand by route and time period, enabling the scheduling team to adjust service levels proactively rather than reactively.
Maintenance Intelligence Layer A dedicated maintenance view surfaced vehicles approaching preventive maintenance thresholds, flagged overdue units, and identified vehicles with recurring failure patterns that warranted deeper inspection or retirement. The system calculated a health score for each vehicle based on age, mileage, maintenance history, and recent repair frequency, giving the maintenance director an objective basis for prioritizing work orders and capital replacement requests.
Automated Alerting & Escalation N8N powered an alerting system that delivered targeted notifications based on role and responsibility. Maintenance supervisors received alerts when vehicles crossed service thresholds. Operations managers were notified when on-time performance dropped below targets on active routes. Leadership received daily digest emails summarizing key metrics and any anomalies detected in the prior 24 hours. Critical alerts, such as safety-related anomalies or system-wide performance degradation, triggered immediate SMS notifications.
Board Reporting Package An automated monthly board reporting package was generated directly from the dashboard data, formatted to the authority's standard presentation template. This eliminated the manual assembly process and ensured that board members received consistent, accurate, and current data for every meeting.
## Quantifiable Impact
- 200+ daily operations visible in a single consolidated view
- Maintenance backlog reduced by 42% within 90 days of launch
- On-time performance improved from 78.3% to 91.6% over two quarters
- 40 staff hours per week reclaimed from manual report compilation
- Leadership operations meetings reduced from 3 hours weekly to 90 minutes
- Anomaly detection identified 23 emerging issues in the first quarter that would have previously gone unnoticed for days or weeks
- Board reporting assembly time reduced from 5 days to 4 hours
- Fuel cost savings of $127,000 annually through consumption anomaly identification
- Frontend: Next.js (dashboard web application)
- Backend: .NET API
- Cloud: Microsoft Azure
- Database: Azure SQL (centralized data warehouse)
- Reporting & Visualization: Power BI (embedded real-time dashboards, board reporting package)
- AI & Automation: Azure OpenAI (anomaly detection, ridership demand prediction, vehicle health scoring)
- Workflow Orchestration: N8N (data pipeline orchestration, health check monitoring, alert triggers, report generation)
- Communication: Twilio (SMS critical alerts), SendGrid (email digests, notification delivery)
- Data Sources: Integration with CMMS, transit dispatch platform, farebox/APC systems, fleet fueling system, ERP, HR system, document management
Ready to Modernize Your Business?
Let's talk about where technology can move the needle first.