How US Foods Drove 40% More Accurate Delivery Tracking With AWS

Accurate delivery timing isn’t only a convenience, it’s a business necessity. Restaurants and food-service operators depend on precise delivery windows to manage their inventory, staff schedules and customers’ expectations. When deliveries arrive unexpectedly early or late, the resulting ripple effects can disrupt an entire day’s operations.

US Foods, a leading food-service distributor, understood this challenge intimately. It partners with approximately 250,000 restaurants and food-service operators across America, and has 30,000 employees and more than 70 locations. US Foods had an ambitious goal: achieve 80% to 90% estimated time of arrival (ETA) accuracy across all markets by the end of 2024, supported by all-day machine learning operations (MLOps) monitoring. It recognized that transforming its delivery tracking system wasn’t only about improving efficiency, but also about enhancing its entire customer experience.

The Challenge: Enhancing Delivery Excellence

To manage its transportation operations, US Foods relied on a third-party fleet management system. The system provided real-time GPS tracking, a record of drivers’ hours, route optimization and vehicle performance monitoring. While this system met basic operational needs, US Foods saw an opportunity to elevate its delivery experience to better serve its customers.

The Solution: Building a Smarter Tracking System

US Foods has an established culture of innovation. It encouraged its teams to develop and test solutions using AI and ML to address challenges across the business. Building on its successful rollout of an AI-powered sales tool, US Foods decided to create its own AI-driven proprietary delivery tracking system: Where’s My Truck.

“We will continue to bring additional value and better user experiences to our customers through our digital platform, MOXē. And to highlight one functionality enhancement, we recently optimized our delivery tracking function, utilizing artificial intelligence within the Where's My Truck feature for MOXē customers to track their truck and delivery time.” – Dave Flitman, Chief Executive Officer, US Foods

Where’s My Truck was built using a comprehensive suite of AWS services, including Amazon SageMaker for model training; Amazon Elastic Container Registry, Amazon Elastic Container Service, and AWS Fargate for containerization and deployment; and Amazon Aurora PostgreSQL for database needs. The entire infrastructure was orchestrated using AWS Cloud Development Kit, with Amazon Location Service integrated to provide driver travel estimates in real time.

Using Amazon SageMaker and these supporting AWS services, the solution incorporates three sophisticated models to improve tracking and provide real-time delivery estimates.

• Likely Remaining Route Model: This adaptive system generates optimal routes when planned sequences change, ensuring real-time accuracy even when circumstances require route modifications.
• Service Time Model: By considering both stop numbers and driver behavior patterns, this model provides precise drop-off time estimates, accounting for the unique characteristics of each delivery location.
• ETA Range Estimator: Using probabilistic arrival distributions, this model creates more accurate delivery windows, helping customers plan with confidence.

Transformative Results

The team began piloting the Where’s My Truck solution in two markets in April 2024 and scaled the test to 10 markets by July. Where’s My Truck uses AI to help MOXē customers monitor and track their trucks as well as delivery times in real time. The results were so significant that the company decided to implement the solution across all markets before the year ended.

As US Foods CEO Dave Flitman noted, “Over the past few months in our pilot markets, we’ve improved our delivery window accuracy by 40%. And we’ve seen a significant reduction in delivery-related customer service calls in those markets because of the better customer experience.” The system now processes thousands of transactions every 15 minutes, providing real-time updates and accurate delivery estimates across the entire network.

Looking Forward: The Next Evolution

US Foods plans to iterate and enhance Where’s My Truck to optimize both model accuracy and the overall customer experience. Version 3.0 of the system will further refine drive time estimates, improve layover routes and address market-level variations, pushing the boundaries of what’s possible in delivery tracking accuracy.

The Future of Delivery Tracking

The US Foods journey with Where’s My Truck demonstrates how AI-powered solutions can transform delivery operations, significantly improving the customer experience and operational efficiency. By applying advanced technology and maintaining a customer-first approach, US Foods has created a system that not only meets current needs but also can be trained and updated to improve over time. Its success showcases how AI can strengthen supply chain resilience by providing real-time visibility and predictive insights that help companies proactively manage disruptions while maintaining service levels across distribution networks.

To learn more about how AWS can help you use the cloud and artificial intelligence to improve supply chain efficiency, visit aws.com.