Loadsmart uses machine learning to provide instant freight hauling quotes to shippers

Trucks_parked_along_I-5.jpg

Though the freight industry is worth over $700 billion, it is still stupendously fragmented. Over 90% of the fleet companies operating in the ecosystem own 6 or fewer trucks and the majority are also conventional in their approach to freight hauling, lacking technology influx which inadvertently reduces their scope of business. On the other end of the spectrum, large shippers who ship hundreds of loads every week find it hard to get capacity as they do not have a dependable system to rely upon.

Loadsmart, an on-demand truckload shipping business, is shrinking the freight ecosystem by bringing shippers and carriers together. “When we started the company, only about 25% of the drivers had smartphones. Even today, most of the brokers depend on email and phone calls to move freight,” said Felipe Capella, co-founder of Loadsmart. “We saw an opportunity to bring in technology and automation into this cycle.”

The company’s first feature was about instant pricing and booking for full truckloads, which when launched in 2015 was the first of its kind in the U.S., allowing shippers to book a full truckload in under 5 seconds. Interested shippers could open the Loadsmart web portal and could book a truck even without signing up, which was an instant success.

With the initial features working out well, the company moved on to integrate its solution into enterprise accounts using its API. “It meant that bigger clients like Daimler, who move hundreds of loads per day, could see our instant pricing on full truckloads in their own system, without having to do anything,” said Capella. “They don’t have to email or call us to know how much it costs to move the load as they can already see that inside their system. All they have to do is press a button if they like our pricing.”

This saves clients a lot of effort with regard to calling their carriers, and sending emails notifying them of loads. Automation of its sales processes has also saved Loadsmart a lot of time and resources, as the company does not have a dedicated sales team to gather prices load-by-load.

Loadsmart differs from other competitors in this niche as it has invested heavily in its engineering division instead of the sales division - thereby allowing it to assemble a team of machine learning engineers and data scientists. The company gathers data from every single one of the 272,000 carriers running on the U.S. roads and uses the data in its sourcing algorithm, to find the best possible match for its shipping clients.

“When we run our sourcing algorithm, we don’t only run it based on the data of the carriers that are currently signed up with us. We run it against every single carrier in the U.S., and based on our technology, we can onboard any carrier in the U.S. in around 3 minutes,” said Capella. “We have APIs of insurance companies to check for liabilities, technology processes to check for safety scores, and we check with the IRS to see if the IRS number in the W9 matches, to check for fraudulent behavior.”

Analysing data from various data sources, Loadsmart can zero in on the estimate for the cost of hire based on the load, time of loading, commodity involved, and the lane it gets hauled in. This allows the company to guarantee capacity to the shippers instantaneously with the estimated cost for delivery, even before it confirms a truck to haul the capacity. Loadsmart is changing the traditional freight brokerage system in place by siphoning up the principal risk associated with the transaction.

The risk involved would require the company to tread carefully, and with the backing of concrete data analytics. For example, Loadsmart could calculate the cost of hire to be $1,000 for a specific haul, which might turn out to be $1,100 in the end, leading to $100 loss to the company. When this feature was launched in 2015, the startup did make a few mistakes, but with more data flowing in and a better-trained machine learning model, the company is now reaping rich for its persistence.

“We rely on data and machine learning processes to source a load. Right now, more than 60% of our loads are sourced through our algorithms, which makes it very efficient,” said Capella.

Data analytics is powering the company in understanding different aspects of trucking, that is not just about digital freight brokerage. For example, Loadsmart has analyzed all the truck roadside inspections done in the U.S. in the past few years, and built a heat map of the trucks stopped, which helps it create a profile of how a specific trucking company works. With hundreds of data sources, Loadsmart has established a system that can pinpoint with huge accuracy, the carriers that would be the best possible match for moving a specific load.

The company has shown impressive growth over the years and has recently signed a partnership with Daimler Trucks North America (DTNA) for hauling their loads across the U.S. through a digital platform focussed at serving the auto major’s contracted freight carriers.

Stay up-to-date with the latest commentary and insights on FreightTech and the impact to the markets by subscribing.