Skip to Main Content
Nitin OberoiDirector of Product Management, FourKites

Last week, FourKites introduced our Dynamic ETA for LTL, powered by the latest machine learning and artificial intelligence technologies. Given that this is an industry-first, we wanted to take a moment to share here the process and philosophy that went into building and releasing this feature, and how our customers can leverage it to drive higher visibility for their LTL shipments. 

Why did we build it?

Traditionally, tracking LTL loads has been one of the industry’s most intractable challenges, due to the inherent complexity of multiple terminal stops and widely variable transit times for LTL freight. The lack of predictable and reliable ETAs translates into delays and dissatisfied customers up and down the supply chain. This has meant that “day of arrival” has been the gold standard for LTL shipments, but knowing only what day your shipment will arrive is not good enough for truly effective planning. 

Now, with FourKites’ Dynamic ETA for LTL, stakeholders receive one reliable and very narrow ETA window that dynamically adjusts based on real-time status updates, empowering supply chain professionals to plan their operations effectively, and arrange dock and labor schedules to best fit their constantly evolving situation. 

How did we build it?

Within the Visibility Cloud platform, FourKites has tracked over 3 million LTL shipments for customers around the world. In so doing, we have amassed an enormous data set of shipments, terminals, carriers, routes and journeys. Leveraging this dataset and our machine learning capabilities, our team analyzed over 3 million loads, 3.2 billion LTL miles and more than 1.3 trillion data points to generate the industry’s first dynamic ETA model, applicable regardless of geography. 

FourKites’ Dynamic ETA for LTL analyzes 17 different LTL shipment parameters to predict a highly accurate ETA window, constantly reviewing and updating the prediction with any geolocation or status update received. A few of the features that we take into consideration are shipper behavior; carrier behavior; geographic information associated with the loads; seasonality; distances between intermediary points; day of the week; hour of day the loads were picked up; carrier ETAs; appointment times; and a variety of other data attributes. 

Our models are updated every month to ensure that we are capturing the latest behavior.

 

The FourKites LTL ETA contains two outputs:

  • ETA window
    • Predicts the period of time in which your shipment will arrive (as illustrated in the image above)
  • Confidence level
    • Gives additional context on the variability of the ETA by providing a number between 1-100. The higher the number, the more predictable and deterministic the lane/shipper/load in question is. At its most basic, it is the probability of the load getting delivered within the ETA window provided.

. . . . . . . . . . . . .

In discussions with our customers and partners, it was clear that they were not satisfied with the status quo for LTL. This was the sentiment that drove us to invest significant resources into improving the LTL freight shipment experience. Now, with LTL ETAs that are 6x more accurate than industry-standard delivery windows, and customers sharing results along the lines of 147% improvement in customer satisfaction, and 67% reduction in customer service calls, we’re so far very happy with the outcome.

Learn more about FourKites Dynamic ETA for LTL, with automated document retrieval capabilities.

Stay Informed

Join 30,000+ monthly readers and get exclusive ebooks, reports, and industry insights from FourKites every week.

Read our Privacy Policy