Foto 110882762 © Leowolfert

As shippers grapple with supply chain planning in the face of volatility and rapid swings in demand, data science and artificial intelligence (AI) are rapidly emerging as vital tools.

Abe Eshekenazi, CEO of the Association for Supply Chain Management (ASCM), noted that supply chains had been undergoing an enormous amount of disruption, which had increased the pressure on firms to be more precise in their forecasting.

Traditional supply chain planning tools had lost traction in the pace of change, which had shrunk time horizons in terms of how far out companies could predict what they need to produce and when, he told the Reuters Supply Chain Planning event on Wednesday.

Human experience and historical data that used to guide planning decisions were no longer adequate, said Brian Hecht, supply chain specialist at Kroeger, which is forcing a shift from the art of forecasting to the science of forecasting.

“Now changes are happening so fast that we have to move more towards science,” he added.

Moreover, the transformation of supply chains to a digital environment is beginning to stretch the capabilities of individuals, said Mr Eshkenazi.

As a result, data analytics has become more embedded in supply chain planning. Anything that is a transactional activity requiring a lot of individual intervention should move to this sphere, leaving humans free to pursue other tasks, he said, pointing to purchasing.

“We don’t need to renegotiate every contract. We can use analytics, and we can use AI to better optimise our purchasing,” he added.

Another element driving firms to shift to data analytics is the lack of storage space at stores, said Mr Hecht. Storage space at the back of stores has largely disappeared over the years, leaving retailers with no redundancy to hold stock there.

Bill Mines, SVP finance, logistics & supply chain at Walmart, said the retailer did not want to build up redundancy. Instead, it aims to leverage AI and data science.

“You couldn’t build in any level of redundancy to cope with what we’ve had to cope with in the past 18 months,” he said. “That’s where the data science really comes in.”

Walmart has aggressively embraced these tools as it shifted from spreadsheets to web-based and data science modules at the back end, and is developing web-based scenario-planning functionality.

“We’ve definitely seen a much broader, as well as quicker, adoption of advanced analytics in the supply chain space,” reported Jackie Guan, senior director data & analytics at the retail giant.

Six years ago, data science models were basically a dirty word among the firm’s operators; now everybody is asking for them, she said.

One initiative has been the development of an in-house forecasting platform at distribution centre area level, which is used daily and weekly.

“It allows us to plan our labour more accurately,” Ms Guan said. “We were able to dramatically improve our forecasting ability.”

In response to the labour challenges the company has experienced, her team developed an optimisation model and corresponding tool for labour recruitment and planning. In any of its distribution centres there are up to two million configurations of different applicants, job calls and schedules, so the tool helps make faster and better decisions, she said.

Now Walmart is working on a tool, and corresponding app, to give estimated arrival times for trucks, a project that involves investment in sensors, data technology and advanced analytics.

While data analytics and AI offer massive improvements, it is vital to start with the actual business problems and targeted outcomes a company wants to address, before getting to the data and the technology. Understanding the processes is of the utmost importance to get on the right track, said Mr Hecht. This slows down progress a bit at the beginning, “but slow is smooth, and smooth is fast.”

Comment on this article

You must be logged in to post a comment.