Operational and talent benefits behind data science
Using analytics to drive automated decision-making can make companies less dependent on single human resources, according to MercuryGate CEO Monica Wooden.
There are many reasons why advancements in data science herald promising things for transportation procurement and management systems.
But Monica Wooden, chief executive officer for TMS software provider MercuryGate, noted an interesting one this week in a presentation at SMC³’s annual JumpStart conference in Atlanta.
Aside from all the operational benefits that smarter systems provide, Wooden said that being able to convert knowledge into automated decision-making makes it easier to bring new talent into a company.
The thinking goes like this: the more artificial intelligence is brought into a company’s freight processes, the more the system automates not just data entry but actual decisions, the faster the learning curve for new hires, and the more easily someone can cover for a colleague who’s on vacation or pulled away on another project.
It’s an overlooked aspect of automation, the impact not only on luring talent, but in maximizing talent once it’s in the company. This is an especially important issue for two reasons:
• Freight transportation has long been seen as an industry where experience and expertise are irreplaceable, and while that’s correct, decision automation can go some way to bridging the experience gap;
• And as transportation procurement, planning and execution become more complex (due to the amount of data companies can now collect), it becomes important for shippers and 3PLs to reduce their vulnerabilities, and one of those vulnerabilities is becoming reliant on single employees to handle critical functions.
Wooden talked in depth about other, more operational advantages of embedded analytics into freight management.
For instance, say a shipper has repetitive moves with multiple legs being created in the TMS, only it’s not apparent to the transportation planner how often this exact move happens.
“You’ll have the system intelligently tell you this type of move happens multiple times each month,” she said. “Maybe you tell the system if this move happens four times in the next 30 days, automatically create a template for that move.”
That creates a situation where errors are eliminated the next time the load is built since it's essentially pre-built.
That may seem like a relatively simple example, but as Wooden pointed out, data science allows a system to build upon its learning. In other words, as one decision is automated, the system builds that automation into its future decision-making. Each automated function is a building block for more automation down the road.
This leads into another point of discussion Wooden addressed. TMS's long ago evolved into systems designed to allow shippers and 3PLs to manage by exception. People only get involved when an issue comes up, allowing the shipments experiencing no problems to be essentially automated.
But Wooden was asked, could the system automate the exception resolutions as well? She described a tiered approach to exception management.
“Say you’re expecting a status message from your carrier every four hours,” she said. “The system would query the carrier about the missing status image. You can tell the system that if you don’t get a status message after 30 minutes, send an alert to a human’s dashboard.”
From a procurement perspective, Wooden described using a mix of different decision-making criteria in an automated tender environment. The criteria would be weighted differently depending on the needs of the shipper.
Again, these are relatively simple examples, but it’s also important to understand that not everyone even has a basic TMS. Wooden’s straw poll at the SMC³ event showed around half of attendees didn’t have a TMS, and virtually no one was using embedded analytics in their TMS.
“We’re trying to take you from algebra to geometry to trigonometry to calculus to differential equations,” she said. “You have to keep progressing. We’ve seen shippers do some of the most incredible automation.
“But the hardest thing for people to do is change. Instead of having customers write their ‘as is’ and ‘to be’ processes, I tell them, let us show you the best practices of day in the job of a TMS. Forget what you’re doing today, let us show you what you’re doing tomorrow.”
A big problem with integrating data science into decision-making is giving up control. In that sense, Wooden compared TMS automation to a driverless car. Would people jump in a driverless car right now?
“A lot of people just don’t trust the optimizer,” she said. “Humans believe they can do it so much better. So it’s trusting the results of the optimizer. Trusting the technology. Stopping the touching everything.”