Logistics Twitter Hashtag Crisis Alerts

Summary

A colombian logistics company was losing revenue due to delivery delays caused by spontaneous protests and blockades across Colombia. The company lacked real-time situational awareness and only discovered disruptions after customers complained. This project created a real-time alerting pipeline using Twitter data and AWS services, enabling drivers to proactively reroute and avoid protest zones.

Business Challenge

During the 2021 national strikes, Colombia experienced severe transportation blockades. The company faced significant losses:

• Fresh coffee and flower deliveries were spoiled in transit.
• Urgent medical shipments were delayed.
• Shopify merchants switched to competitors due to poor reliability.

The company had no real-time way of knowing where protests were occurring. Information arrived too late, typically through:

• Customer complaints after hours of delay.
• Driver WhatsApp messages sent from blocked highways.

Solution and Architecture

To solve this, I built a real-time Twitter ingestion and alert system. It detects protest-related spikes in specific regions and alerts drivers via SMS. A dashboard highlights risk zones for the operations team.

• Tweet Ingestion: AWS Kinesis streams filter geolocated tweets containing keywords like #ParoNacional, #BloqueoEnCali, #MedellinProtest.
• Real-Time Processing: AWS Lambda functions analyze tweet frequency by region and match against active delivery routes stored in DynamoDB.
• Alerts: Twilio sends SMS alerts to drivers when a risk zone is detected on their route.
• Visualization: QuickSight shows affected areas on a live map with delivery impact reports.

Steps Taken

• Identified relevant hashtags and implemented keyword filtering logic in Spanish and English.
• Set up a geofenced Twitter ingestion stream using AWS Kinesis.
• Developed a Lambda processor that detects abnormal tweet activity and cross-references affected delivery routes.
• Connected Twilio to dispatch rerouting alerts via SMS to field drivers.
• Designed a live operations dashboard in QuickSight for real-time monitoring and historical trend analysis.

Impact and Results

The implementation significantly reduced disruption-related losses. Key outcomes include:

• 35% reduction in average delay time across protest-affected routes.
• $8,000/month saved in spoiled goods and rerouted deliveries.
• Live dashboard enabled proactive rerouting and daily risk planning.
• Historical protest data now supports route optimization and driver scheduling.

This solution turned reactive delivery operations into a proactive, data-driven system—preserving both profit and client trust.