- XPO Logistics will pilot its XPO Smart predictive analytics tool in its less-than-truckload network in North America. The tool was previously used to optimize labor productivity in the company's warehouses and distribution networks. CIO Mario Harik said it successfully reduced costs and increased fulfillment speeds.
- The tool is a suite of applications that "uses proprietary algorithms and site-specific machine learning to determine how individual output contributes to collective goals" according to an XPO press release. It compares "real-time productivity rates with the number of active dock workers, using machine learning to predict how adjustments in labor levels affect productivity [and track them] against production targets."
- LTL terminal managers in Massachusetts, Michigan and North Carolina are using XPO Smart, and the company plans to expand the technology to its 290 U.S. terminals over the upcoming quarter.
As supply chains globalize and seek to become more responsive to market shifts, logistics providers see value in semi- and fully-automated tools to identify areas where operational efficiency could be improved and track performance against those targets in real time.
A recent report found even among the companies not yet using predictive analytics, 57% of them had plans to adopt it within the next five years, and 90% believe it will have a material impact on supply chain operations over the next 10 years.
This is particularly relevant for LTL, last-mile shipments and warehousing operations as volumes of smaller packages increase due to the continued expansion of e-commerce.
"Our pilot LTL operators are acting on insights gained from site-specific machine learning and our predictive analytics," Harik said in a statement.
Having personnel that are able to work with these new technologies and accurately interpret the tools' results is a step towards developing a successful digital supply chain operation. Otherwise, even the best algorithm will suffer from a "garbage-in, garbage-out problem" where its recommendations can only be as correct as the information it was given to start with.